Risk & Psychology

Why most traders lose

In 2018 the European Securities and Markets Authority (ESMA) forced CFD brokers to publish what share of their retail clients lose money. The disclosed figures sat between 74% and 89% of retail accounts. That is not a guess from a forum. It is brokers reporting on their own clients, under regulatory pressure.

The easy explanation is "they had bad strategies". Some did. But when you look at how retail accounts actually blow up, the pattern is usually about risk, not entries:

• Too much size for the account. One bad day takes 20% or more, and the trader never recovers emotionally or mathematically.
• No defined loss per trade. Stops get moved, widened or removed, so one loss can equal ten wins.
• Overtrading. Barber & Odean (2000), "Trading is hazardous to your wealth", studied tens of thousands of US brokerage households. The most active fifth earned about 11.4% a year net, against roughly 17.9% for the market over the same period. Activity itself, through costs and poor decisions, was the drag.
• No idea of their own numbers. Most losing traders cannot tell you their win rate, average win or average loss. They are running a business with no accounts.

Module 02 gave you a setup model. This module is about the part that decides whether that model ever pays you. A decent setup with strict risk can make money. A great setup with sloppy risk almost always loses.

Expectancy: the only number that matters

Expectancy is how much you make, on average, every time you take a trade. If it is positive, the more trades you take (following the same rules), the more money you should make over time. If it is negative, no amount of discipline or hard work will save you.

In plain words: take the share of trades you win and multiply it by your average win. Then take the share you lose and multiply it by your average loss. Subtract the second from the first.

Expectancy = (Win rate × Average win) − (Loss rate × Average loss)

We measure the wins and losses in R, where 1R is the amount you planned to risk on the trade. You will see why in the "Thinking in R" section below.

Now look at a trader who feels much more successful day to day.

A high win rate feels good. It is not the goal. Expectancy is the goal, and it needs both halves of the formula.

Expectancy also tells you how to improve. There are only three levers: win more often, win bigger, or lose smaller. Losing smaller is the one most traders ignore, and it is the easiest to pull. If the scalper above simply kept every loss at exactly −1R and lifted the average win from 0.4R to 0.5R, expectancy would go from −0.02R to (0.70 × 0.5) − (0.30 × 1.0) = +0.05R. Small, but positive, and positive is the only number that compounds in your favour.

To calculate yours, you need at least a few dozen journaled trades with the result of each one written in R. If you do not have that yet, start now: from today, every trade in your journal gets an R result.

Win rate vs reward: the breakeven table

Every reward-to-risk ratio has a breakeven win rate: the win rate at which you make exactly zero (before costs). Win more often than that and you make money; win less often and you lose.

In plain words: you need enough wins to pay for your losses. If each win is worth 2 losses, you only need to win 1 trade in every 3 to break even.

Breakeven win rate = 1 ÷ (1 + Reward-to-risk)

For a 2R target: 1 ÷ (1 + 2) = 1 ÷ 3 = 33.3%.

Illustrative figures, before spread, commission and slippage.

Your real edge is the gap between your actual win rate and the breakeven win rate for the R:R you actually achieve. If your A+ setups hit 2R and win 45% of the time, your edge is 45% − 33.3% = about 12 percentage points. That gap is what pays you.

Two important cautions. First, higher targets get hit less often, so you cannot just pick 5R and assume your win rate stays the same. Chapter r3 deals with this. Second, the table ignores costs, which raise every breakeven number a little. Chapter r3 shows you how much.

The breakeven lab lets you move both sliders and watch the expectancy change. Spend five minutes finding the combinations where a small drop in win rate turns a profit into a loss.

Thinking in R

Van Tharp popularised R-multiples in Trade Your Way to Financial Freedom. The idea is simple: 1R is the amount you planned to lose if the trade hit your stop. Every result is then expressed as a multiple of that amount.

In plain words: divide what you made or lost by what you planned to risk.

R-multiple = Profit or loss ÷ Planned risk

• You risk $200 and make $500: $500 ÷ $200 = +2.5R.
• You risk $200 and get stopped cleanly: −$200 ÷ $200 = −1R.
• You risk $200, slip through the stop on news and lose $300: −$300 ÷ $200 = −1.5R.
• You risk $200 and close at breakeven: 0R.

Why bother? Because dollars lie. A $1,000 win on a day you were oversized looks great but might be a 4R gamble. A $150 loss might be a perfect −1R trade on a small account. R strips out account size and position size and shows you the quality of the decision.

R also makes your journal comparable across markets. A −1R on MNQ and a −1R on MGC are the same event, even though the point values are completely different. When you review in Chapter r9, everything will be in R.

The asymmetry of losses

Losses and gains are not symmetrical. If you lose 50% of your account, you do not need a 50% gain to get back. You need 100%, because you are now growing a smaller base.

In plain words: the percentage you need to recover is the loss divided by what is left.

Gain needed to recover = Loss % ÷ (100% − Loss %)

Illustrative figures.

Small drawdowns are cheap to fix. Large ones are brutally expensive, and above about 40% they become almost impossible for most traders, because the size and pressure needed to recover lead to more mistakes.

This is the single best argument for small risk per trade. You are not protecting yourself from one bad trade. You are protecting yourself from ever entering the part of the table where the maths turns against you.

Variance: why good traders have bad months

Even with a real edge, results arrive in a random order. You do not get your 45% win rate evenly spread as nine wins in every twenty trades. You get clumps of wins and clumps of losses. This randomness is called variance.

Take a solid strategy: 40% win rate, 2R average win, 1R average loss. Expectancy = (0.40 × 2) − (0.60 × 1) = +0.20R per trade. Over many trades it makes money.

Now look at a single month of 20 trades. Each win is +2R and each loss is −1R, so the month is negative if you win 6 trades or fewer (6 wins = +12R, 14 losses = −14R, net −2R). With a true 40% win rate, the probability of 6 or fewer wins in 20 trades is about 25%.

So a genuinely profitable trader following this strategy should expect roughly one losing month in four, purely from luck. Nothing is broken. Nothing needs fixing. That is what an edge of +0.2R looks like up close.

This matters because most traders abandon a working system during a normal losing month and jump to something new, which resets their sample to zero. Mark Douglas made this point in Trading in the Zone (2000): any single trade's outcome is random, even when the edge across a series of trades is real. You judge a system on a large sample, not on this week.

The EdgeFlow risk hierarchy

Every risk rule in this module serves one of three goals, in strict order of priority:

1. Survive. Never lose so much that you cannot keep trading, financially or emotionally. This means small risk per trade, hard daily limits and no single trade that can hurt you badly.
2. Be consistent. Take the same quality of setup, at the same risk, with the same rules, so your results are a fair sample of your edge. You cannot measure an edge you apply randomly.
3. Grow. Only once you have survived and been consistent do you increase size, and only by rules (Chapters r2 and r9).

The order is not negotiable. A trader who chases growth before consistency is just adding size to noise. A trader who chases consistency at the expense of survival (for example by forcing trades to "keep the streak going") is gambling with the account.

Illustrative rules; you will build your own in Chapter r9.

Fixed-fractional risk

Fixed-fractional risk means you risk the same percentage of your account on every trade. If you risk 0.5% and your account is $50,000, 1R is $250. If the account grows to $60,000, 1R becomes $300. If it falls to $40,000, 1R becomes $200.

This does two useful things automatically. When you are winning, your size grows with the account. When you are losing, your size shrinks, so each new loss hurts a little less in dollars. You never need to decide "should I trade smaller now?" in the middle of a drawdown, because the rule has already done it.

The alternative, picking a fixed number of contracts ("I always trade 2 NQ"), ignores the stop distance and the account size. Two NQ with a 10-point stop risks $400. Two NQ with a 40-point stop risks $1,600. Same "size", four times the risk.

Choosing your risk per trade

Classic trading books often quote 1–2% per trade as the rule. Treat that as the old ceiling, not a target. It came from an era of longer-term swing traders taking a handful of trades a month. Intraday traders take far more trades, hit losing streaks more often in calendar time, and frequently trade prop accounts with tight drawdown limits.

Look at what a run of losses does at each risk level (each loss is taken from the shrinking balance):

Illustrative figures, compounding each loss on the remaining balance.

Ten losses in a row sounds extreme. It is not. With a 40% win rate, Chapter r4 shows it is quite likely somewhere in a year of active trading. At 2% risk that streak costs you nearly a fifth of the account. At 0.5% it costs under 5%, and you barely notice it.

EdgeFlow's guidance:

• 0.25%: your first 50–100 live trades, a new strategy, or the first weeks back after a drawdown.
• 0.5%: the default for most members on personal and prop accounts.
• 0.75%–1%: only with a proven sample (Chapter r9 gives the rules) and an account where a 10-loss streak does not threaten any hard limit.
• Above 1%: not part of the EdgeFlow plan.

The position-sizing formula

In plain words: work out how many dollars you are willing to lose, work out how many dollars one unit (one contract, one lot, one coin) loses if the stop is hit, and divide the first by the second. Then round down.

Position size = Dollar risk ÷ (Stop distance × Value per point per unit)

where Dollar risk = Account size × Risk %.

The three steps, every time:

1. Dollar risk: account × risk %. Example: $50,000 × 0.5% = $250.
2. Risk per unit: stop distance in points (or pips, or dollars of price) × the value of one point for one contract or lot.
3. Divide and round down: dollar risk ÷ risk per unit. Always round down to a whole contract (or the smallest lot step your broker allows).

The sizer lab does these three steps for any market. Use it before every trade until you can do the maths in your head, and keep using it after that, because tired traders make arithmetic mistakes.

Sizing futures: NQ vs MNQ and rounding down

Index futures are where most new traders get their sizing wrong, because a full-size contract is large. One NQ is $20 per point. One MNQ is $2 per point, exactly one tenth.

Notice what happened in the NQ line. The honest answer was "zero contracts". Many traders round 0.5 up to 1 NQ, which doubles their risk to $500, or 1% of the account. Do that on every trade and your carefully chosen 0.5% risk is fiction.

The same logic applies to the S&P: ES is $50 per point and MES is $5 per point. A 10-point ES stop is $500 per ES contract but only $50 per MES.

Illustrative stops.

Rules for futures:

• Always round down. If the formula says 0.6 contracts, the answer is zero full-size contracts; move to micros.
• Micros first. Until your dollar risk per trade comfortably covers at least one full contract at your typical stop, trade micros. Micros also let you size by grade, which full contracts usually cannot.
• Never shrink the stop to fit a contract. If 1 NQ only fits with a 12-point stop and the setup needs 25, the stop is wrong and you will get stopped out by noise (Chapter r3).

Sizing forex and crypto perps

The formula is identical. Only the "value per point" changes.

Forex (EURUSD): one standard lot (100,000 units) is about $10 per pip, a mini lot (10,000 units) is about $1 per pip.

Crypto perpetuals (BTC): here you size in coin or in notional value. The stop distance is easiest to use as a percentage.

That last line is the most important idea in this section. Leverage does not change your risk. The stop and the position size do. Leverage only changes how much margin the exchange holds while the trade is open. A trader who says "I used 50× so I was risking more" is confused. They were risking more because they held a bigger position, and the leverage merely allowed it.

Two practical warnings for perps:

• Liquidation must sit well beyond your stop. If high leverage puts your liquidation price closer than your stop, you can be liquidated before your stop triggers, and liquidation fees make it worse.
• Fees matter in R. At an illustrative 0.05% taker fee per side, a $10,000 position costs $5 to enter and $5 to exit: $10 total, which is 0.1R on a $100 risk. Chapter r3 covers costs in detail.

Grade-based sizing

Module 02 taught you to grade every setup A+, A or B. Those grades are not just labels. In EdgeFlow they set your size, so that your best setups carry the most risk and your marginal setups carry the least.

Illustrative stops.

Here "1R" means your full base risk from your plan (0.5% in this example). An A setup risks three quarters of that; a B setup risks half.

Why this works: your grading is an honest estimate of probability. If A+ setups really do perform better, putting more on them raises your overall expectancy. If your journal later shows that B setups have negative expectancy, they are already at half size, so the damage while you collect the data is small. And "below B = no trade" removes the most dangerous trade of all: the one you take because you are bored.

The Kelly criterion and why traders use a fraction

In 1956 John Kelly, a researcher at Bell Labs, published a formula for the bet size that grows wealth fastest over the long run when you have a known edge. Traders still talk about it, so you should understand it, and understand why nobody sensible uses the full amount.

In plain words: the Kelly percentage is your win rate, minus your loss rate divided by your reward-to-risk ratio.

Kelly % = Win rate − (Loss rate ÷ Reward-to-risk)

17.5% per trade is absurd for a real trader, and here is why:

• You do not know your true edge. Kelly assumes your win rate and payoff are exact. Yours are estimates from a limited sample, and if you overestimate your edge you will bet too big. In the standard mathematical model, betting more than twice the true Kelly amount makes long-run growth negative.
• Full Kelly is a stomach-churning ride. In the textbook model, a full-Kelly bettor has about a 50% chance of seeing the account halve at some point. Half Kelly keeps about 75% of the growth rate while cutting that chance to roughly 12.5%.
• Trades are not independent coin flips. NQ and ES trades on the same morning are effectively one bet (Chapter r5). Kelly does not know that.
• You have hard limits. A prop firm's max drawdown or your own personal stop-loss is a wall. Kelly assumes you can keep betting smaller forever.

Ralph Vince's optimal f is a related idea that works from your actual trade history instead of a single win rate. It has the same weakness: it finds the size that would have been best on past data, which is reliably too aggressive for the future.

So professionals use Kelly as a ceiling and a sanity check, not a target. If your Kelly figure is 10% and you risk 0.5%, you are at one twentieth of Kelly. That is deliberately conservative, and it is the right place to be while your sample is small.

When to change size

Size changes should come from rules you wrote in advance, never from how you feel in the moment.

Reasons to reduce size (do it immediately):

• You hit a drawdown circuit breaker (Chapter r4: −6R in a week halves size; −10% from peak quarters it).
• You are trading a new market, session or setup type.
• Your pre-session state check (Chapter r8) scores low.
• You are trading around major news with wider spreads and slippage.

Reasons to increase size (do it slowly, and only on a schedule):

• Your account has grown, so your fixed-fractional 1R has grown automatically. That is already scaling up.
• You have 100+ journaled trades with positive expectancy at your current risk, and you followed your rules on at least 90% of them. Then step up by a small amount, for example from 0.5% to 0.6%, not straight to 1%.

Never a reason to change size:

• "I'm due a win" after a losing streak.
• "I'm on fire" after a winning streak.
• Wanting to make back today's loss.

A stop is an invalidation, not a pain limit

Most beginners set stops by asking "how much can I stand to lose?" That is backwards. A stop answers a different question: at what price is my reason for this trade proven wrong?

If you are long because price swept sell-side liquidity and then shifted structure upward, the trade idea is wrong once price trades back through the sweep low. Anywhere before that, the idea is still alive. Anywhere after that, holding on is hope.

So the order is always:

1. Find the invalidation price on the chart.
2. Put the stop just beyond it.
3. Then size the position so that stop equals your planned 1R (Chapter r2).

The pain limit is handled by size, not by stop placement. If the correct stop is far away, you trade fewer contracts. If you cannot afford even one micro at the correct stop, you skip the trade.

Where EdgeFlow stops go

Module 02 gave you the setup model. Here are the stop locations that go with it, from tightest to widest:

Illustrative guide; your journal will show which works best for each setup.

The EdgeFlow default for an A or A+ reversal is beyond the sweep. The sweep low (for a long) is the price that, if broken, says the liquidity grab failed and the move down is genuine. A stop inside the FVG or order block is a valid, more aggressive choice, but it should be a decision you make on purpose and track in your journal, not a habit to make the R:R look better.

"Beyond" means past the level plus a buffer, never exactly on it. Stops placed exactly on an obvious low are themselves liquidity, and Module 02 taught you what price likes to do to obvious liquidity.

Stops that are too tight

A stop that sits inside the market's normal noise gets hit by randomness, not by the setup failing. The trade then goes on to do exactly what you expected, without you.

A simple way to measure noise is the Average True Range (ATR): the average size of a candle's full range on your entry timeframe (most platforms show it with a 14-period setting). If the 5-minute ATR on NQ is 12 points, a single ordinary 5-minute candle covers about 12 points. A 6-point stop can be taken out by half of one normal candle.

EdgeFlow rule: put a buffer of about 0.25 × ATR beyond your invalidation level, measured on your entry timeframe.

A second check: if your stop distance is smaller than about 1 × ATR on your entry timeframe, ask whether you are really trading the setup or just trying to make the R:R look attractive. Tight stops are not wrong, but they need a precise entry and a market that is moving cleanly, and they should show up in your journal as a separate, tracked variation.

The hidden costs: spread, commission and slippage in R

Every trade pays costs: the spread or the tick you give up getting filled, the commission and exchange fees, and slippage when a stop order fills worse than its price in a fast market. These are small in dollars, but you should measure them in R, because that is where they do the damage.

In plain words: convert your round-trip cost into points, then divide by your stop distance in points.

Cost in R = Round-trip cost in points ÷ Stop distance in points

Take an illustrative all-in cost on MNQ of about 1 point per contract per round trip ($2 per MNQ): roughly half a point of commission and fees plus half a point of spread and slippage.

Same market, same target in R, same contract. The tight-stop trader needs over four percentage points more win rate just to stand still. On a scalping strategy with 4-point stops, the same 1-point cost would be 0.25R per trade, and that alone can wipe out a real edge.

The same logic applies everywhere: wider spreads on EURUSD outside the London and New York sessions, wider spreads on CL around the weekly inventory report, and taker fees on BTC perps (an illustrative 0.05% per side on a position with a 1% stop is about 0.1R round trip).

Planning the trade in R

Before you enter, write down the whole trade in R. If you cannot fill in every line, you do not have a trade plan yet.

The EdgeFlow trade-plan lines:

1. Entry price and order type.
2. Stop price, and the invalidation it sits beyond.
3. Stop distance in points and the cost in R.
4. Size from the sizer (Chapter r2), by grade.
5. Target 1 and Target 2, with the liquidity each one aims at, and the R of each.
6. Management rule: how much you take off at each target, and whether and when the stop moves.
7. Blended R if both targets are hit, and the result if the stop is hit first.

Writing it this way does three things. It forces you to find the target before you are emotionally involved. It tells you whether the trade is worth taking at all (a 0.8R plan on a B setup usually is not). And it gives you a clean record to compare against what actually happened.

Partials and moving to breakeven: what it really does to your numbers

Taking partial profits and moving the stop to breakeven feel safe. They raise your win rate and reduce the number of full losses. But they also cut some of your best trades short. Whether they help is a numbers question, not a feelings question.

Here is an illustrative 100-trade sample of the same setup, with a 3R final target:

• 25 trades go straight to 3R.
• 10 trades reach 1R, pull back to the entry price, then go on to 3R.
• 20 trades reach 1R, then reverse all the way to the original stop.
• 45 trades never reach 1R and hit the stop.

Illustrative sample, before costs.

In this sample, the version that feels best (C, 55% winners) has half the expectancy of the version that feels worst (A, 35% winners). The 10 trades that dipped back to entry before running are the reason: breakeven stops turn them from +3R into 0R or +0.5R.

This does not mean partials are always wrong. In a different sample, where most trades that reach 1R then fail, partials and breakeven stops can raise expectancy. They also reduce variance, which matters on a prop account with a tight drawdown. The point is that you must test it on your own journal, not adopt it because it makes losing feel less bad.

EdgeFlow defaults (change them only with journal evidence):

• Move the stop to breakeven only after a new structure break in your favour, not at a fixed R. That way the new stop has its own invalidation logic.
• Partials at the first liquidity target, sized so the remaining position still matters (for example 40% off at T1, 60% held to T2).

Targets: liquidity, not hope

Module 02 taught you that price moves from liquidity to liquidity. Your targets come from the same idea. Before entry, ask: where is the opposing liquidity?

For a long, that means buy-side liquidity above you:

• The nearest internal high (often Target 1).
• Equal highs, the previous day's high, the session high (Asia or London) or the previous week's high.
• The draw on liquidity you identified for the session.

When price is already beyond the obvious pools, or for a runner, use the standard-deviation extensions from Module 02: project the dealing range beyond its end at −0.27 and −0.62.

Three target rules:

1. Target the level, then subtract a little. Exit a few ticks before the liquidity, because price often reaches the pool without trading through it by much.
2. If the nearest liquidity gives less than about 1.5R, the trade usually is not worth it. Either wait for a better entry or skip.
3. Never move a target further away during the trade because "it looks strong". That is a new trade decision made while emotional. If you want a runner, plan it before entry.

Reward-to-risk myths

Myth 1: "Only take trades with 3R or more." A 3R target needs at least a 25% win rate, but further targets are hit less often. What matters is expectancy at each target, from your own data.

Myth 2: "A tighter stop gives a better R:R, so it's always better." It changes the R:R on paper, but it lowers the win rate and raises costs in R. Check both effects before switching.

Myth 3: "R:R is fixed when you enter." Your realised R:R depends on management. Partials, early exits and breakeven stops all change the number you actually get. Your journal should record the planned R and the actual R separately.

Myth 4: "Low-R:R strategies are for amateurs." Some excellent strategies win often with small targets. They are only fragile when costs are large compared with the stop, which is why you measure cost in R.

What drawdown actually measures

A drawdown is the fall from your highest account value (the peak) to the lowest point that follows it (the trough), before a new peak is made. It's measured from the peak, not from your starting balance. If you started at $50,000, grew to $54,000 and then fell to $51,000, you're not "up $1,000". You're in a $3,000 drawdown.

Why measure from the peak? Because that's the money you actually had, and that's what your brain anchors to. A trader who gives back $3,000 of gains feels the loss as sharply as a trader who loses $3,000 of starting capital. The peak-to-trough number is also what prop firms use for their trailing limits (Chapter r6).

You should track drawdown two ways:

• In percent of the peak: drawdown ÷ peak equity × 100. This tells you how hurt the account is.
• In R: drawdown ÷ your current 1R in dollars. This tells you how many full losing trades the drawdown is worth, which makes it comparable across account sizes and across months.

The maximum drawdown is the deepest drawdown in your history. It's the single most important risk number in any track record, because it shows what you actually had to sit through to earn the return.

Notice the last line. You lost 6.0% of the peak, but you need 6.38% to get it back. That small gap is the start of a much bigger problem.

The recovery table

When you lose money, you have a smaller account to earn it back with. So the percentage gain needed to recover is always bigger than the percentage you lost. The formula in plain words: divide the loss by what's left.

Recovery % = loss % ÷ (100 − loss %) × 100

At small drawdowns the gap is minor. At large ones it becomes brutal.

Illustrative figures; the maths holds for any account size.

You met the −50% / +100% example in Chapter r1. The table shows the full curve. Up to about −15%, recovery is roughly proportional: lose 10%, make back 11%. Past −25% the curve bends hard, and past −50% recovery becomes unrealistic for most traders, because the edge that produced a modest yearly return now has to produce a double or more.

There's a second, hidden cost. In a drawdown you're also trading smaller if you size as a percentage of current equity (Chapter r2). That protects you on the way down, which is exactly the point, but it also means each winning trade earns less on the way back up.

How long losing streaks really get

Most traders badly underestimate how long their losing streaks will be. If your win rate is 40%, five losses in a row sounds unusual: the chance of any specific five trades all losing is 0.6 × 0.6 × 0.6 × 0.6 × 0.6 = 7.8%. That feels rare.

But you don't take five trades. You take hundreds. And the question that matters is not "what's the chance these five lose?" but "what's the chance that somewhere in my next 100 trades, five in a row lose?" That chance is far higher, because there are dozens of overlapping places a streak can start.

The table below shows the probability of seeing at least one losing streak of a given length in 100 trades, assuming each trade is independent.

Probabilities for 100 independent trades, calculated exactly.

Read the 40% row slowly, because that's where many good reward-to-risk traders live. Over 100 trades, a five-loss streak is close to certain (97.6%). An eight-loss streak is a coin flip (49.0%). One in five traders with this exact edge will see ten or more losses in a row.

None of those traders did anything wrong. Their edge is identical. The streak is just where the randomness happened to cluster.

Use the streak lab now. Enter your own win rate and number of trades, run it a few thousand times, and look at the distribution of longest streaks. Then write the 90th-percentile number on a sticky note. That is the streak you should plan to survive.

Expected worst streak over a year of trades

A year of trading for an active intraday trader is often around 250 trades: roughly one trade per trading day. Over that many trades, streaks get longer still. The simulation below ran 100,000 sequences for each win rate and recorded the longest losing streak in each.

Monte Carlo results from 100,000 simulated sequences per row, independent trades.

So a 40% win-rate trader taking 250 trades a year should expect a worst streak of about nine losses, and should be prepared for 13. For the 250-trade year at 40%, the exact probability of at least one 10-loss streak is 45.2%: close to even odds.

Here's how to turn that into a sizing check. Take your 1-in-10 worst streak, add a margin of a few trades for the bad patches that aren't perfectly consecutive (losses mixed with small winners), and ask: at my current risk per trade, what would that cost?

• 13 losses at 0.5% risk: 1 − 0.995¹³ ≈ 6.3% drawdown.
• 13 losses at 1% risk: 1 − 0.99¹³ ≈ 12.2% drawdown.
• 13 losses at 2% risk: 1 − 0.98¹³ ≈ 23.1% drawdown.

If the number at your current risk would breach your personal max loss or a prop-firm limit, your size is too big for your edge. Not "a bit aggressive". Too big.

Drawdown circuit breakers

A circuit breaker is a rule that automatically reduces or stops your trading when losses reach a set level. You decide the levels in advance, when you're calm, so that you don't have to make the decision in the middle of a losing streak, when you're least able to make it well.

EdgeFlow uses three tiers:

R is measured at your normal full size, even after size has been reduced.

These breakers stack. If you hit −6R for the week, you're on half size. If a further losing run takes you to −10% from peak, you drop to quarter size.

How often will these trigger? At 0.5% risk, it takes about 21 straight full losses to go from peak to −10%, so the top tier should be rare if your edge is real. At 1% risk it takes only about 11, which the streak table says is not rare at all for a 40–45% win-rate trader over a year. That's one more reason EdgeFlow defaults to 0.5% or less.

The recovery protocol

Cutting size is the easy half. The hard half is when to size back up. Most traders do one of two things wrong: they go straight back to full size the moment they have a winner, or they stay small forever out of fear.

The trap of waiting to fully recover first is mathematical. Suppose you're 10% below your peak and trading at one-quarter of 0.5% (0.125% per trade). To earn back the 11.1% you need, at 0.125% per R, you'd need about 85R of profit. At +0.35R per trade, that's around 240 trades: a year of trading at tiny size. That's not a protocol. That's a punishment.

So EdgeFlow steps up based on process and short-term results, not on getting all the money back:

1. Quarter size (after a −10% trigger). Trade only A+ and A setups. Step up when you've completed 10 trades at this size with every trade rule-compliant and net R for those 10 trades of zero or better.
2. Half size. Same conditions: 10 rule-compliant trades, net R zero or better.
3. Full size. Normal rules resume, including B setups at their normal grade size.

If you hit −3R in a day while on reduced size, that day stops as normal. If you lose 6R (at full-size R) from the start of any step, you drop back one step. "Rule-compliant" means you'd mark it Y in the journal field covered in Chapter r9: correct setup, correct grade size, stop placed before entry, no moved stops.

Variance or broken edge? A checklist

Every drawdown raises the same question: is this bad luck, or has my edge stopped working? Both happen. The cost of getting it wrong is high either way: abandon a good system in a normal streak and you throw away your edge; keep trading a broken one and you bleed out.

First, know what normal looks like. Take a system with a 40% win rate and 2R winners. Its expectancy is +0.2R per trade: a genuine edge. Yet calculated exactly, the chance it's still negative after a run of trades is:

Exact binomial probabilities for 40% win rate, +2R winners, −1R losers.

One in six traders with this real edge will be negative after 50 trades. So a losing 50-trade stretch, on its own, proves very little. You need to look at why you're losing, not just that you're losing.

Run this checklist at every −10% trigger, and at the end of any losing month:

1. Did I follow my rules? Pull every trade in the drawdown. What share were rule-compliant? If it's below 90%, the problem is execution, not the edge. Fix that first.
2. Is the streak inside the expected range? Compare your current losing run with the 90th-percentile streak for your win rate. If it's inside, variance is the likely explanation.
3. Are losses normal size? Average loss should be about −1R. If it's −1.3R or worse, you have a slippage, stop-placement or sizing problem.
4. Are winners normal size? If your average winner has shrunk from 2R to 1.2R, check whether you're cutting winners early (the disposition effect, Chapter r7) or whether targets have stopped being reached.
5. Has the market changed? Compare average daily range now with your backtest period. A setup built in a high-volatility market can struggle when ranges halve.
6. Is one setup, session or market responsible? Split the drawdown by setup, grade and session. If all the damage is in one place, cut that one place, not the whole system.
7. What does the backtest say? Did your backtest or earlier live record contain drawdowns this deep? If so, this is familiar territory. If this drawdown is 1.5× or more the worst you've ever recorded, treat the edge as suspect.

If items 1, 3 and 4 are clean and 2 and 7 say "within range", it's almost certainly variance. Keep going under the recovery protocol. If items 5, 6 or 7 raise flags, pause live trading on the affected setup, return to the replay or sim, and collect 30–50 fresh trades before you risk real money on it again.

What ruin really means

In textbooks, risk of ruin means the chance of losing your entire account. That's the wrong definition for a real trader, because nobody actually trades down to zero. Long before that, one of three things happens:

• You hit your personal maximum loss. The level at which you'd stop trading, or be forced to, because the money matters too much or the recovery is too steep. For most people this is somewhere between −15% and −30%.
• You hit a prop-firm maximum drawdown. On a funded or evaluation account, this might be $2,000 or $2,500 on a $50,000 account (illustrative). Hit it and the account is closed. For prop traders, ruin is a 4–5% move, not 100%.
• You quit. A drawdown deep enough to break your confidence ends your trading as surely as a margin call, even if there's money left.

So in EdgeFlow, ruin = reaching the loss level you've decided you can't go past. For this chapter we'll use a 20% peak-to-trough drawdown as the personal ruin line. You should replace it with your own number.

The three levers of ruin

Three things decide how likely you are to reach your ruin line:

1. Edge: your win rate. More winners means fewer and shorter losing streaks.
2. Payoff: your average winner compared with your average loser (in R). Bigger winners mean each win repairs more damage.
3. Size: your risk per trade. This decides how far each loss moves you toward the line.

Edge and payoff combine into expectancy (Chapter r1). But they're not perfectly interchangeable: two systems with the same expectancy can have very different streak behaviour. A 60% win-rate system with 1R winners suffers fewer long streaks than a 35% win-rate system with 3R winners, even though the second may earn more per trade.

The table shows the effect of edge, with size held at 1% per trade. Each row is 20,000 simulated runs of 200 trades with 2R winners. "Ruin" means a 20% peak-to-trough drawdown at any point.

Monte Carlo results, 20,000 simulated runs per row, independent trades. Illustrative strategy figures.

A five-point drop in win rate, from 40% to 35%, multiplies the risk of ruin by about five. And here's the uncomfortable part: you can't fully control edge or payoff. Markets change, and your live win rate may come in below your backtest. Size is the only lever you control completely, every single trade.

Ralph Vince's work on optimal f showed that for any edge there's a bet size that maximises long-run growth, and that sizes above it actually reduce growth while increasing drawdowns. His point that matters most for traders: the growth-maximising size produces drawdowns almost no human can tolerate. That's the same reason traders use fractional Kelly (Chapter r2). Professional sizing sits well below the mathematical maximum on purpose.

Why tiny changes in risk % matter so much

Moving from 1% to 2% risk per trade sounds like a small adjustment. It's not. It changes the shape of every possible future.

Here's the same strategy (40% win rate, 2R winners, +0.2R expectancy) at five risk levels, 200 trades each, 20,000 simulated runs per level, profits compounded:

Monte Carlo results, 20,000 runs per row. Illustrative strategy figures; real results would vary with costs, slippage and changing conditions.

Look at the jump from 1% to 2%. The median return roughly doubles, which is why the higher number is tempting. But the chance of a 20% drawdown goes from about 1 in 15 to more than 2 in 3. At 3%, almost every path hits the ruin line at some point. Many of those paths still end in profit after 200 trades, but you'd have had to sit through a 33% drawdown, on average, to get there. Most traders don't.

Monte Carlo: seeing a thousand futures

A Monte Carlo simulation is a simple idea with a grand name. You tell a computer your win rate, average winner and risk per trade. It then "plays" your strategy, say, 200 trades in a row, by randomly deciding each trade's result with the right odds: a 40% chance of +2R, a 60% chance of −1R. That gives one possible future equity curve. Then it does it again. And again. A thousand times.

Each curve is a version of your next 200 trades that could happen with exactly your edge. Some are lucky, some are cruel, most are in between. Laying them on top of each other shows the full range of what your edge can produce, which a single backtest can never show you. A backtest is one path. A Monte Carlo is the whole road map.

What you can learn from it:

• The realistic range of returns, not just the average.
• How deep drawdowns get on a typical path, and on a bad one.
• How often you'd hit your ruin line at a given risk %.
• How long flat or losing stretches last even when the edge is real.

There are two common ways to run one:

1. Parametric: you enter a win rate and an average win and loss (what the lab does). Clean and fast, but it assumes every winner is exactly the average size.
2. Resampling your own trades: you feed in your actual journal of R results, and the simulation shuffles and redraws them. This keeps your real mix of small and large winners, but needs at least 50–100 trades to be meaningful.

Open the montecarlo lab now. Start with 40% win rate, 2R average winner, 1% risk and 200 trades, and compare your results with the table in the previous section. Then change one input at a time.

Reading the cone: median, best 5%, worst 5%

When you plot a thousand curves together, they fan out into a shape traders call the cone. Narrow at the start, because after a few trades nobody has moved far, and wider over time as luck accumulates. Three lines through the cone tell you almost everything:

• The median (50th percentile): half the paths finished above it, half below. This is your "typical" outcome, and it's a better guide than the average, which a few lucky runaway paths can inflate.
• The best 5% (95th percentile): only 1 in 20 paths did better. This is your realistic good year. Don't plan on it.
• The worst 5% (5th percentile): only 1 in 20 paths did worse. This is the line you plan around. If you can't live with the worst-5% outcome, your size is too big.

For the 40% / 2R strategy at 1% risk over 200 trades, the simulation gives:

Monte Carlo results, 20,000 runs. Illustrative strategy figures.

Read the worst-5% line carefully. One in twenty traders with this exact edge, doing everything right, makes only about 5% over 200 trades. Some will be negative for most of the period. If you are that trader, nothing is wrong with you. You drew a bad sample of the same edge.

You should also look at the maximum drawdown distribution, not just final returns. At 1% risk, the median path's worst drawdown was 11.8%, and the worst 5% saw 21.0% or more. So even at 1%, a bad-luck path crosses the 20% ruin line.

Choosing your risk % from the simulation

Here is the EdgeFlow method for choosing your normal risk per trade. It replaces "1% because everyone says 1%" with a number that fits your edge and your limits.

1. Set your ruin line. Personal account: the peak-to-trough drawdown you'd stop at (say 20%). Prop account: the max drawdown in dollars.
2. Enter honest inputs. Live or forward-tested win rate and average winner. If you have fewer than 100 trades, knock 5 points off the win rate.
3. Run the simulation at several risk levels. For example 0.25%, 0.5%, 0.75%, 1%.
4. Apply two tests. The chance of reaching your ruin line must be 5% or less, and the worst-5% max drawdown should be no more than about three-quarters of your ruin line, so even a bad path leaves you room.
5. Pick the largest risk % that passes both. Then round down.

Rerun this every quarter with your updated numbers. As your live sample grows and your stats firm up, the right risk % may rise. If your win rate slips, it should fall.

Correlation: when two trades are really one

Every calculation so far assumes each trade is independent. That fails the moment you hold two positions that move together. Correlation measures how closely two markets move together: +1 means they move in lockstep, 0 means no relationship, −1 means they move in opposite directions.

The most common trap at EdgeFlow is NQ and ES. Both are US stock index futures, and their intraday moves are usually very closely linked. If you go long NQ at 0.5% risk and long ES at 0.5% risk on the same morning, you haven't taken two half-sized trades. You've taken roughly one 1% trade on "US stocks go up". If the market drops, both stops are likely to be hit together.

Other pairs to watch:

Relationships are general tendencies, not constants; they change over time.

How many trades before you trust your stats

Your win rate after 20 trades is a very rough guess. The question is how rough. Statisticians answer with a confidence range: the band within which your true win rate probably sits, given what you've seen.

The plain-words formula for the 95% range around a win rate is: win rate ± 2 × the square root of (win rate × loss rate ÷ number of trades). (The exact multiplier is 1.96; 2 is close enough to remember.)

95% confidence ranges, normal approximation, rounded to whole percentages.

After 20 trades, a "45% win rate" could be a losing system or a superb one. After 100, the range is still ±10 points. You need a few hundred trades before you can be fairly confident to within 5 points.

Practical rules:

• Under 30 trades: you have an anecdote, not a statistic. Trade minimum size.
• 30–100 trades: a rough guide. Stress-test everything with a lower win rate.
• 100–200 trades: enough to make sizing decisions, with caution.
• 200+ trades per setup: enough to compare setups, sessions and grades against each other with some confidence.

Split samples shrink fast. If you have 200 trades but want to know your Silver Bullet win rate in the London session on A+ grades, you might have 15 trades in that cell. Treat it as a hint, not a fact.

Why the best traders stop early

Look at any trading journal with a few hundred trades in it and you'll usually find the same pattern: the worst days were not caused by the first loss. They were caused by what happened after it. The first loss is the market. The fourth and fifth are the trader.

Brett Steenbarger, a psychologist who has worked with professional traders for decades, describes how performance drops when a trader's emotional state shifts from focused to frustrated (Enhancing Trader Performance, 2006). Jared Tendler's framework (Chapter r8) calls this sliding from your A-game into your C-game. Either way, the result is the same. Decisions made after a string of losses are worse than decisions made fresh.

A daily limit is not an admission of weakness. It's a business rule that caps the cost of your worst state. Professional desks have run this way for decades: a trader who hits their loss limit stops for the day, regardless of how good they believe the next setup is.

There's a second benefit. A good day also has a limit of sorts. The ICT approach is built around a small number of high-quality windows (the London and New York killzones, the Silver Bullet hours). Once those windows are done, the best setups of the day are usually gone. Trading on into the afternoon chop, whether you're up or down, tends to give back money.

Building your daily loss limit

Build your limits in R, then convert them to dollars. That way they scale automatically with account size and risk %.

EdgeFlow's default daily loss limit is −3R on closed trades. Here's why 3R:

• It allows for two normal losses plus one more, which covers most ordinary bad days without stopping you on the first losing trade.
• At 0.5% risk, −3R is −1.5% of the account. Even five bad days in a row only costs about 7.3% (0.985 to the power of 5 ≈ 0.927), which stays well inside the flat part of the recovery table.
• Combined with the two-loss rule in the next section, you'll rarely reach it. It's the backstop, not the plan.

The formula: daily loss limit ($) = account equity × risk % × 3.

Three extra rules make the limit work in practice:

1. The limit is on closed P&L plus open risk. If you're down $500 and you open a trade risking $250, you're already at the limit if that trade loses. Count the stop before you enter.
2. Hitting the limit ends the session, not the hour. Close the platform. No sim "just to watch". No returning in the evening session.
3. The limit doesn't reset with a win. Down −2R, then +1R, then −2R is a −3R day. Stop.

Open the limits lab and enter your own account size and risk %. It will show your daily, weekly and monthly limits in dollars, and the number of full losses you can absorb at each level.

Max trades per day and the two-loss rule

A loss limit caps how much you lose. A trade limit caps how many decisions you make. Both matter, because overtrading is one of the best-documented ways retail traders destroy their returns. Barber and Odean's study of tens of thousands of US brokerage accounts ("Trading is hazardous to your wealth", 2000) found that the households that traded most earned the lowest net returns, largely because of costs and poor timing.

EdgeFlow defaults:

• Maximum 3 trades per day across all markets. Most good days in the EdgeFlow model contain one or two A-grade setups. A fourth trade is usually a forced one.
• The two-loss rule: after two consecutive losses in a session, stop trading for that session (London or New York). You may trade the next session only if you're inside the daily limit and pass the pre-session state check in Chapter r8.
• One re-entry per idea: if you're stopped out of a setup, you may re-enter once if the setup is still valid (for example, a new MSS after a second sweep). After that, the idea is dead for the day.

Weekly and monthly limits

Daily limits stop single bad sessions. Weekly and monthly limits stop slow bleeds: weeks where no single day hits −3R, but every day loses a little.

EdgeFlow defaults (consistent with the circuit breakers in Chapter r4):

At 0.5% risk, −10R in a month is about −5% of the account. That's a bad month, but it sits comfortably in the part of the recovery table where a normal month or two of trading closes the gap.

News and event risk

Scheduled US economic releases can move NQ 100+ points or gold $20+ in seconds (illustrative), with spreads widening and stops filling far from where you placed them. In R terms, a 1R stop can easily become a 2R or 3R loss in a fast release. That breaks every calculation in this module.

Know the big ones and their times. Dubai doesn't use daylight saving, so the Dubai time shifts by an hour when the US changes its clocks. Times are shown as US summer (EDT) / US winter (EST).

Check an economic calendar each week; release dates move and US holidays shift some releases.

EdgeFlow's news rules:

1. Flatten before tier-one releases. Be flat at least 5 minutes before CPI, NFP and the FOMC decision on any market they affect. No exceptions for "it's nearly at target".
2. Skip the first 15 minutes after. No new entries until 15 minutes after a tier-one release (08:45 ET / 16:45 or 17:45 GST for CPI and NFP). On FOMC days, no new entries until the press conference has finished, which often means no New York PM trading at all.
3. FOMC day is a half day. Trade the morning killzone if you want to, at normal size, and be done before the afternoon.
4. CL traders respect EIA Wednesday. Flat on CL/MCL 5 minutes before 10:30 ET.
5. Know the calendar on Sunday. Write the week's tier-one events into your plan before Monday's session.

Prop-firm drawdown types: static, intraday trailing, end-of-day trailing

Prop-firm evaluations and funded accounts all have a maximum drawdown: a floor below which the account is closed. The rules vary between firms and change over time, so always read the current rules for any account you trade. But nearly every firm uses one of three types. We'll use the same illustrative account for each: $50,000 starting balance, $2,000 maximum drawdown.

1. Static drawdown. The floor is set once at the start and never moves. Here it's $50,000 − $2,000 = $48,000, forever. Grow the account to $55,000 and you have $7,000 of room. This is the most forgiving type.

2. Intraday trailing drawdown. The floor follows your highest account value at any moment, including open (unrealised) profit on a live trade. It only moves up, never down. This is the harshest type, because a trade that runs into profit and then comes back can raise your floor without any closed gain.

3. End-of-day (EOD) trailing drawdown. The floor follows your highest end-of-day closing balance. What happens during the day doesn't move it; only where you finish the day does.

Illustrative $50,000 account, $2,000 max drawdown.

On many trailing accounts, the floor stops trailing once it reaches a set level, often around the starting balance. That point is important: once reached, the drawdown effectively becomes static. Check exactly where it is for your firm.

Daily loss limit vs max drawdown

Many prop accounts also have a daily loss limit (DLL): a maximum you can lose in one day. Depending on the firm, hitting it either fails the account or simply locks you out until the next session. These two limits do different jobs:

• Max drawdown: the total distance from the floor. Hitting it ends the account.
• Daily loss limit: a cap on one day. It protects the max drawdown from a single bad session.

Your personal daily limit should sit well inside the firm's DLL. If the firm's limit is your stop, you'll often blow through it on slippage. If it's your target, you're risking the account every bad day.

Consistency rules, payouts and the buffer strategy

Many prop firms have a consistency rule: no single day may make up more than a set share of your total profit, measured either during the evaluation or before a payout. The share differs between firms; we'll use 30% as an illustration. The aim is to filter out traders who pass on one lucky, oversized day.

The formula: best day ÷ total profit must be at or below the limit. If it's above, you need more total profit until it isn't.

This is one more reason to keep size consistent and use daily targets loosely. A 3R day at normal size is great. A 10R day at triple size might feel great but can delay your payout for weeks.

The buffer strategy. On a funded account with a trailing drawdown, your first goal is not a payout. It's building a buffer: enough profit that the floor is far below you, or has stopped trailing.

1. Phase 1: build the buffer. Trade normal size until your balance is at least 10R above the point where the floor locks or, for a static account, 10R above the floor. No withdrawals.
2. Phase 2: withdraw from the excess only. When you take a payout, leave at least the 10R buffer intact. If the payout would take your cushion below 10R, take a smaller one.
3. Phase 3: never trade bigger to make the payout faster. Payout timing comes from consistency, not size.

A prop-firm risk plan that survives 10+ losses

Put everything together and you can design a prop plan around one simple goal: the account must survive the longest losing streak you're likely to see. From Chapter r4, a 45% win-rate trader should plan for streaks of 10–11 losses in a year of trades. The plan below survives 13.

For a $50,000 evaluation with a $2,000 end-of-day trailing drawdown and $3,000 profit target (illustrative):

Illustrative account and rules; adjust to the firm's current terms.

How safe is 13 losses? At a 45% win rate, the chance of 13 or more losses in a row somewhere in 200 trades is about 3.5%. At 40% it's about 9.5%. And because of the daily stop and the two-loss rule, a 13-loss streak would take at least five trading days to happen, giving you several chances to notice and review.

Size also changes your chance of passing far more than most traders expect. The simulation below used a 40% win rate, 2R winners and the $2,000 trailing drawdown (on closed balance) with a $3,000 target, 50,000 runs per row.

Monte Carlo results, 50,000 runs per row, independent trades, no costs or daily limits. Illustrative.

Bigger size passes faster when it works, but fails about half the time. Smaller size takes longer, often a few months, but passes most of the time. Evaluations cost money, and every reset is a cost. The slower path is usually the cheaper one.

Your brain is not built for trading

For most of human history, a loss could mean you didn't eat, and a threat ignored could kill you. So your brain learned to feel losses sharply, act fast on recent events, find patterns everywhere and stick with the group's story. Those instincts kept your ancestors alive. In a market, almost every one of them costs money.

Trading is unusual because the feedback is noisy. In most of life, a good decision usually leads to a good result fairly quickly. In trading, a perfect A+ setup can lose and a reckless impulse trade can win. Your brain learns from results, not from decisions, so it happily learns the wrong lessons.

This chapter walks through the biases that do the most damage to discretionary traders, names the research behind each one, and gives you a rule for each. You won't remove the biases. Nobody does. The aim is to decide in advance, when you're calm, so the bias has nothing to change in the moment.

Prospect theory and loss aversion

In 1979, Daniel Kahneman and Amos Tversky published prospect theory, one of the most cited papers in economics. Its central idea is simple: people don't judge outcomes by final wealth. They judge them as gains or losses from a reference point, and losses hurt far more than equal gains feel good.

In a follow-up in 1992, Tversky and Kahneman estimated that a loss weighs roughly 2 to 2.5 times as much as a gain of the same size. Losing $500 feels about as bad as winning $1,000 to $1,250 feels good. That's loss aversion.

Prospect theory also found two behaviours that matter directly for trading:

• In gains, people play safe. Offered a sure $500 or a 50% chance at $1,000, most take the sure thing. In a trade, that's taking profit early because "a bird in the hand".
• In losses, people gamble. Offered a sure −$500 or a 50% chance of −$1,000 (and 50% of nothing), most take the gamble. In a trade, that's moving your stop or refusing to close, because a sure loss feels unbearable and the gamble offers a chance to escape it.

Put those together and you get the most expensive pattern in retail trading: small wins, big losses. It's not a knowledge problem. It's prospect theory running your mouse.

The rule: every trade has a hard stop placed in the order ticket at entry (a bracket or OCO order). Stops can be moved only toward profit, and only by a rule written in your plan (for example, "to breakeven after price closes beyond the next swing", from Chapter r3). A stop moved away from profit is logged as a rule break, whatever the result.

The disposition effect (cutting winners, holding losers)

In 1985, Hersh Shefrin and Meir Statman named the disposition effect: the tendency to sell winners too early and ride losers too long. It follows directly from prospect theory. A winner sits in the "gains" zone, where you play safe and bank it. A loser sits in the "losses" zone, where you gamble on it coming back.

Terrance Odean tested this on real brokerage data in his 1998 paper, Are Investors Reluctant to Realize Their Losses? Using thousands of discount-brokerage accounts, he found investors realised gains roughly one and a half times as readily as losses. Worse, the winners they sold tended to keep outperforming the losers they held. They were cutting the wrong positions.

For an intraday trader, the disposition effect shows up as:

• Closing at +1R when the plan said +2.5R, "because it pulled back a bit".
• Letting a −1R loser run to −1.5R or −2R "to give it room".
• Taking partials on every winner but never on a loser.

Here's what it costs in numbers, using expectancy from Chapter r1 (expectancy = win rate × average win − loss rate × average loss, all in R).

The rule: targets are set at entry as a limit order at the liquidity level you planned (Chapter r3). You may close early only for a pre-written reason, such as "a 5-minute body close back through the entry FVG". "It felt like it was turning" is not on the list.

Overconfidence and overtrading (Barber & Odean)

Brad Barber and Terrance Odean's 2000 paper has a title that says it all: Trading Is Hazardous to Your Wealth. They studied over 60,000 US households with discount-broker accounts in the early 1990s. The households that traded the most, the top fifth by turnover, earned about 11.4% a year net, while the market returned about 17.9%. The trading itself, through costs and poor timing, was the damage.

Their 2001 follow-up, Boys Will Be Boys, looked at gender as a proxy for overconfidence, since psychology research shows men tend to be more overconfident in areas like finance. Men traded about 45% more than women, and their extra trading reduced their net returns by more than it reduced women's. The lesson isn't about gender. It's that confidence drives activity, and activity is expensive.

For you, overconfidence shows up after a good run. Three green days and suddenly B setups look like A+, you add a second market, and the two-trade cap starts to feel like it's for beginners.

Recency, the hot hand and the gambler's fallacy

Recency bias means the last few trades feel like the truth. Win three in a row and the next one feels certain. Lose three and your edge feels broken. But three trades tell you almost nothing. As you saw in Chapter r5, you need dozens to hundreds of trades before your stats mean much.

Recency splits into two opposite errors:

• The hot hand: "I'm on a streak, so the next one will win." You size up exactly when you feel most sure, which is when you're most exposed to a normal reversion.
• The gambler's fallacy: "I've lost four in a row, so I'm due a winner." You size up to catch the win you're "owed". The market owes you nothing.

If your trades are roughly independent (which most well-defined setups are, close enough), each trade's chance of winning is the same regardless of what came before.

The rule: size is set by grade and by the drawdown circuit breakers from Chapter r4, and nothing else. Size may go down after losses (−6R week = half size). It never goes up because of a streak in either direction.

Confirmation bias and the story you want

Confirmation bias is the habit of seeking, noticing and remembering evidence that supports what you already believe, and ignoring the rest. In ICT terms: once you've decided NQ is bullish, every FVG is a buy zone, every sweep of a low is "smart money accumulating", and the bearish displacement on the 15-minute is "just internal structure".

It gets worse with social media. You can always find a chart, a thread or a video that agrees with you. Searching for agreement feels like research. It's actually just collecting comfort.

Three habits that fight it:

1. Write the opposite case first. Before every session, write one sentence for the bullish case and one for the bearish case, then say which you favour and why.
2. Name your invalidation in advance. "My bullish bias is wrong if the 1H closes below 19,950." If that happens, you're flat or neutral. No re-reading the chart to find a new reason.
3. Fix your timeframe before you look. As Module 02 said, flicking timeframes until one agrees with you is confirmation bias in its purest form.

Sunk cost and anchoring to your entry

Sunk cost is letting money already spent (or lost) drive your next decision. "I've lost $600 today, I can't stop now, I have to make it back." The $600 is gone whatever you do next. The only question that matters is whether the next trade is a valid setup at valid size.

Anchoring is fixing on a reference number and judging everything relative to it. The most common anchor in trading is your entry price. The market doesn't know or care where you got in. Yet traders constantly manage positions around it: "I'll get out at breakeven", "I'll hold until it gets back to my price".

Here's how it plays out. You go long NQ (illustrative) at 20,000 with your stop at 19,980 below the order block, trading 5 MNQ. That's 20 points × $2 × 5 = $200 at risk. Price drops to 19,990 and the 5-minute closes below the FVG that justified the trade. Your plan says the idea is invalid. But you're anchored to 20,000, so you think "I'll get out when it gets back to breakeven". Price keeps falling, and now 19,980 is "too close to bother", so you move the stop.

That one trade has anchoring, sunk cost and loss aversion stacked on top of each other.

The rules:

• Manage every open position by asking: "If I were flat right now, would I enter this trade here, with this stop?" If the answer is no, close it or tighten to your rule.
• After your daily loss limit (−3R in the EdgeFlow plan), you're done. Today's losses don't create a reason to keep trading. They're the reason the limit exists.
• Never use "back to my entry" as a target. Targets are liquidity levels, not your break-even price.

Spot the bias: real scenarios

Read each scenario, name the bias, then check the rule. The bias lab below lets you do this interactively with more variations.

1. The early exit. You're long MNQ from an A setup with a target at the previous day's high, +2.5R away. Price reaches +1.2R, then pulls back 8 points. You close "to lock it in". Price hits your target 20 minutes later. Bias: disposition effect (fear of a gain turning into a loss). Rule: target is a resting limit order. Early exit only on a pre-written structural reason, such as a body close back through the entry FVG.

2. The extra room. Short GC at 2,400 (illustrative), stop 2,404. Price trades to 2,403.50 and you move your stop to 2,408 "because it's about to sweep the high and reverse". Bias: loss aversion and the risk-seeking in losses described by prospect theory. Rule: stops move only toward profit. A stop moved away is a rule break and is logged.

3. The streak. You've won four trades in a row this week. Today you see a B setup on ES and take it at full A+ size because "I'm reading it perfectly right now". Bias: overconfidence and the hot hand. Rule: size comes from grade only. B = 0.5R, always.

4. The owed winner. Three losses in a row on NQ. You tell yourself the fourth has to work and double your contracts. Bias: gambler's fallacy. Rule: size never increases after losses. If this would break the daily limit, you're done.

5. The one-sided research. You're bullish on BTC. You spend 30 minutes reading posts from bullish accounts, skip the 4H bearish CHoCH, and go long. Bias: confirmation bias. Rule: write the opposite case and your invalidation level before the session. A 4H CHoCH against you means no trade until it's resolved.

6. The breakeven hope. Long CL at 80.00 (illustrative). The trade idea is invalidated at 79.85 but you hold "until it gets back to 80". Bias: anchoring to entry. Rule: "would I enter here now?" If no, close.

7. The make-it-back session. You're down −2.5R by 10:15 ET (18:15 GST in US summer). The plan says stop at −3R, so you take a marginal setup because "I've already lost so much today that one more doesn't matter". Bias: sunk cost. Rule: after two losses, take a 15-minute break and only A+ setups for the rest of the day (the two-loss rule from Chapter r6). At −3R, platform closed.

8. The quiet market. NQ is ranging and nothing on your list is setting up. You switch to EURUSD, a market you don't normally trade, and take a "setup" there. Bias: overconfidence plus action bias (the urge to do something). Rule: only markets in your written plan. No setup = no trade is a valid, profitable day.

Rules beat willpower

You may have heard of ego depletion, Roy Baumeister's idea from the late 1990s that self-control is a limited resource that runs down with use. It's a popular explanation for why traders make worse decisions late in the session. Be careful with it: large, pre-registered replication attempts in the 2010s found much weaker or no effects, so the theory is now contested. Don't build your plan on it.

What's much less contested is simpler: in-the-moment willpower is unreliable, especially when you're tired, stressed, or staring at a red P&L. So don't rely on it. Rely on decisions you made earlier, when you were calm.

That's why every bias in this chapter has a rule, not a mindset. Here's the summary you can print:

Automate wherever you can. Use bracket orders so the stop and target exist before you can argue with them. Use your platform's daily loss lockout if it has one. Keep the rules on a card next to your screen. Every rule you automate is one less decision your stressed brain gets to make.

Emotions are information, not instructions

Every trader feels fear, greed, frustration and excitement. Trying not to feel them doesn't work, and the traders who claim they feel nothing usually just haven't noticed. Brett Steenbarger, a psychologist who has spent decades working with professional traders, makes a useful point across his books (The Psychology of Trading, 2003, and Enhancing Trader Performance, 2006): emotional state affects how you process information, and the goal is to notice it and work with it, not suppress it.

So treat each emotion as a piece of data with a question attached:

The key skill is naming the emotion out loud or in writing as it happens. "I'm frustrated." "I'm scared of missing this." Labelling creates a small gap between feeling and action, and that gap is where your rules can work.

FOMO and chasing

FOMO (fear of missing out) is the urge to enter because price is moving without you. It usually hits right after a displacement candle leaves your planned entry behind. The trouble is that chasing doesn't just change your entry. It wrecks your reward-to-risk, because the stop still needs to go at the structural invalidation.

The FOMO rules:

1. No fill, no trade. If price leaves your entry zone without a fill, the trade is cancelled. Write "missed" in your journal with a screenshot.
2. Re-entry only on a new setup. If price pulls back and forms a fresh, gradeable setup, you can take that. You can't take the old one at a worse price.
3. Minimum R before entry. If the planned target is less than 2R from your actual entry, you don't enter. (Chapter r3 explained why.)
4. Track the misses. After a month, look at your "missed" log. You'll find plenty of trades that ran without you and plenty that reversed. Seeing both kills the belief that every missed move was free money.

Revenge trading

Revenge trading is trading to get back at the market, or to get back to zero, after a loss. It's driven by anger and by the reference point from prospect theory: you measure the day from your starting balance, and being below it feels intolerable.

You'll recognise it by the signs: faster entries, bigger size, lower-grade setups, and the thought "I just need one good trade". It's the single most common reason traders break daily loss limits.

The anti-revenge rules:

• The two-loss rule (Chapter r6): after two consecutive losses, stand up, step away for at least 15 minutes, and only A+ setups for the rest of the session.
• Size never goes up intraday after a loss. It can only stay the same or go down.
• Make the limit physical. Use your platform's daily loss lockout if it has one. If not, set an alarm at −2R that tells you to stop and run the reset routine below.

The five types of tilt

Poker players call it tilt: emotional play that drags your decisions below your normal standard. Jared Tendler, a mental game coach who worked with poker players before traders, wrote The Mental Game of Trading (2021). He argues that tilt isn't one thing. It comes in distinct types, each with its own trigger, and you fix it by finding your pattern.

Tendler's practical point is that you need to write down your own tilt profile: the triggers, the thoughts, the feelings in your body, the actions, ordered from mild to severe. Once you know your early signs, you can step in at level 2 instead of discovering at level 9 that you've blown the day.

Here's a simple version to fill in:

1. Level 1–3 (early signs): e.g. "I start checking P&L every 30 seconds", "my jaw tightens", "I say 'come on' out loud".
2. Level 4–6 (building): e.g. "I start looking at a second market", "I think about adding contracts".
3. Level 7–10 (full tilt): e.g. "I'm entering without a screenshot", "I've moved my stop".

Your rule: any level 4 sign = reset routine. Any level 7 sign = done for the day.

Your A-, B- and C-game

Tendler also describes your performance as a range: your A-game (your best decisions), your B-game (solid, ordinary execution) and your C-game (your worst habits under stress). Everyone has all three. What separates good traders is that their C-game isn't a disaster.

Write a short list for each:

• A-game: "I wait for the full checklist. I take the trade at planned size. I leave the target alone. I log it calmly."
• B-game: "I'm a bit hesitant, I miss one setup, I check P&L too often, but I follow every rule."
• C-game: "I enter early before the MSS confirms, I take B setups at A size, I move stops, I trade outside my session."

Tendler calls the improvement process the inchworm: you improve by first fixing the worst of your C-game (pulling the back end of the range up), which then makes room to push the front of your A-game forward. In practice, that means your first job isn't to have more great days. It's to make your bad days less bad.

The pre-session state check

Pilots run a checklist before take-off, even after thousands of flights, because skill doesn't protect you from fatigue or distraction. Your state check works the same way. It takes two minutes and tells you whether to trade full size, reduced size, or not at all.

Score each item 0, 1 or 2. Be honest. Nobody else sees it.

Scoring:

• 0–3: Full size. Trade your plan as normal.
• 4–7: Reduced. All grade sizes halved (A+ = 0.5R, A = 0.375R, B = 0.25R), and your max trades for the day drops by one.
• 8 or more: No trading. Do review work, backtesting or chart study instead.

Hard overrides (regardless of total):

• Item 8 scores 2 (no plan written) → no trading until it's written.
• Item 5 scores 2 (you need money today) → no trading. Urgency is the single best predictor of forcing trades.
• Item 4 scores 2 (you broke a rule last session) → maximum reduced size, and you must have reviewed that session in your journal first.

Log the score in your journal every day. After a month you can check whether your low-score days really do outperform your high-score days. For most traders they clearly do, and seeing that in your own numbers is what makes you respect the check.

If–then plans that work in the moment (Gollwitzer)

Psychologist Peter Gollwitzer's 1999 paper on implementation intentions showed that a goal ("I'll stick to my stops") is much weaker than a specific plan in the form "If situation X happens, then I will do Y." Later reviews of many studies found these if–then plans have consistent, meaningful effects on whether people actually follow through.

The reason is that the decision is pre-made and linked to a cue. When the cue appears, the response is close to automatic. You're not deciding under pressure. You're just executing what you decided earlier.

Good if–then plans are specific, observable and short. Here's a starter set. Copy the ones that fit, then write your own for your tilt profile.

• If I've taken two losses in a row, then I stand up, leave the desk for 15 minutes, and only A+ setups are allowed when I return.
• If I reach −3R for the day, then I close the platform and write the journal entry.
• If price leaves my entry without filling me, then I cancel the order and write "missed" with a screenshot.
• If I notice I want to move my stop away from profit, then I take my hand off the mouse, say "plan, not P&L", and look at the invalidation level.
• If I catch myself checking P&L more than once a minute, then I hide the P&L column and set a price alert at my stop and target.
• If a high-impact release (CPI or NFP at 08:30 ET, 16:30/17:30 GST; FOMC at 14:00 ET, 22:00/23:00 GST) is less than 5 minutes away, then I'm flat or my stop is already in the market and I don't touch it.
• If I'm up +3R for the day, then I either stop or only take A+ setups at half size (protecting a good day from overconfidence).

The reset routine (step-away protocol)

When you notice a level 4 tilt sign, or after two losses, run this exact routine. It takes 15 minutes and you do it every time, not just when it feels necessary, because the times it feels least necessary are often the times you need it most.

1. Flatten or confirm. If you're in a trade, either close it or confirm the stop and target are in the market. Then cancel any pending entries.
2. Hands off, stand up. Physically leave the desk. Distance is the point.
3. Breathe slowly for two minutes. Slow breathing with a longer out-breath (for example, in for 4 counts, out for 6) is a simple way to calm your body down. It's not magic; it just gives the spike of emotion time to pass.
4. Name it. Say or write one line: "I'm feeling ___ because ___." For example: "I'm angry because my stop got wicked out before the move."
5. Check the facts. Where am I against my daily limit, in R? How many trades left? Was each trade today on the checklist?
6. Re-read the plan. Your bias, levels and invalidation for today. Has anything changed structurally, or only emotionally?
7. Decide with a rule. Return at the same or lower grade standard, or end the session. If you're unsure, end it. A missed opportunity costs nothing; a tilted session can cost a week.

Thinking in probabilities

Mark Douglas's Trading in the Zone (2000) is still one of the most recommended trading psychology books, because it tackles the root cause of most fear and greed: expecting any single trade to be right. Douglas lists what he calls five fundamental truths about the market. In our words:

1. Anything can happen. No setup is guaranteed, however perfect it looks.
2. You don't need to know what happens next to make money. You only need an edge and the discipline to repeat it.
3. Wins and losses are randomly distributed for any edge. You can know your win rate over 100 trades without knowing the order the wins and losses arrive in.
4. An edge is just a higher probability of one outcome over another. It's a tilt in the odds, not a prediction.
5. Every moment in the market is unique. This setup isn't the last one, even if it looks identical.

If you truly accept these, a loss stops being evidence that you're wrong or bad. It's just one draw from a distribution you already understand. That's the link to everything in Chapters r1 to r5: expectancy, streak lengths and Monte Carlo cones are thinking in probabilities, written as numbers.

A practical way to make this real is to think in batches of 20 trades. Before each block of 20, write: "I'll take every valid setup at the right size. I expect roughly 8–9 winners and 11–12 losers if my win rate is about 42%, in no particular order." Then judge yourself at the end of the block, not after each trade. At 42%, 20 trades gives an expected 0.42 × 20 = 8.4 winners, but a single block can easily come in well above or below that.

Why a written plan beats a good memory

You'll make your worst trading decisions at the exact moments your memory of your rules is weakest: after losses, during fast markets, when you're tired. A rule you have to remember is a rule you can negotiate with. A rule written on a page next to your screen is much harder to argue with.

A written plan does three jobs:

1. It makes decisions in advance. Every hard question (how big, how many, when to stop) is answered while you're calm.
2. It makes breaking a rule visible. If it's written, you can mark it "followed" or "broken" in your journal. If it's in your head, it quietly shifts.
3. It makes improvement possible. You can only test and change a plan you can see. Vague habits can't be reviewed.

Van Tharp made a similar point in Trade Your Way to Financial Freedom: your edge is a system, and position sizing is part of it, not an afterthought. The plan below is that system, on one page.

Write your risk plan

Here's every field your plan needs, with the EdgeFlow defaults from this module. Fill it in using the plan lab below, print it, and keep it where you can see it while you trade.

Here's how those numbers turn into contracts on a real (illustrative) trade.

The pre-trade checklist

The risk plan sets the rules. The pre-trade checklist checks them before every single entry. It should take under a minute. If any line is a "no", there's no trade.

1. State check done today and size level known (full / reduced / no trading)?
2. Market and session are on my plan?
3. No high-impact news within the next 5 minutes (CPI, NFP, FOMC, or others listed in the plan)?
4. Daily and weekly limits: this trade's full loss would still keep me inside both?
5. Trade count: am I under my max trades for today?
6. Setup and grade: does it meet every item for its grade from the Module 02 setup model? Screenshot taken?
7. Stop is at the structural invalidation (beyond the sweep, OB or FVG), with an ATR buffer if needed?
8. Target is at a liquidity level and is at least 2R from entry after costs?
9. Size comes from the sizer and is rounded down?
10. Bracket order ready: entry, stop and target all in the ticket?
11. Correlation: no open position in a linked market the same way, or combined risk ≤ 1R?
12. Emotion check: can I name what I'm feeling, and is it below level 4 on my tilt profile?

Journal fields that actually matter

Many traders' journals are either empty or a diary of feelings. Neither helps. You want a small set of fields you fill in every trade, so you can sort and filter later. Brett Steenbarger's work stresses that structured self-observation, done consistently, is how traders find both their real strengths and their repeated mistakes.

Record R, not dollars, as your main result. Dollars change with size; R lets you compare a trade from your first month with a trade from your tenth.

Process score vs P&L

Your P&L tells you what happened. Your process score tells you whether you did your job. In the short run they often disagree, because variance can reward bad trades and punish good ones. Over the long run, a high process score is what produces a good P&L.

Process score = trades where every rule was followed ÷ total trades × 100.

Use this simple grid to read each week:

The most dangerous cell is the top right: making money while breaking rules. It teaches your brain that the rule breaks work.

The weekly review (30 minutes, exact questions)

Do this once a week at the same time, for example Saturday morning. Set a 30-minute timer. Answer these questions in writing:

Numbers (10 minutes)

1. How many trades did I take, and what was my total in R?
2. What was my process score?
3. What was my result in R on rule-following trades only, and on rule-breaking trades only?
4. Did I hit any limit (daily −3R, weekly −6R)? Did I respect it?
5. What was my average state check score, and how did low-score and high-score days compare?

Behaviour (10 minutes)

6. Which rule did I break most often, and what was happening just before each break?
7. Which tilt type showed up, and at what level did I notice it?
8. Did I miss any valid A or A+ setups? Why (hesitation, distraction, outside session)?
9. Which trade am I proudest of for its execution, regardless of the result?

Action (10 minutes)

10. What's one if–then plan I'll add or sharpen for next week?
11. Is anything in my risk plan unclear or causing confusion? (Change it only at the monthly review unless it's a safety issue.)
12. What's my one focus for next week? (One. Not five.)

The monthly stats review

Once a month, look at your journal as a dataset. The goal is to find where your edge really lives and cut what doesn't work. Calculate expectancy (Chapter r1) for each slice: by setup, by session, by day of the week, by grade and by market.

But 15 trades is a small sample. As Chapter r5 showed, you can't trust stats from a handful of trades. So use a staged rule instead of cutting on one bad month:

• Under 30 trades in a slice and negative: reduce that slice to B size (0.5R) and keep collecting data.
• 30+ trades and still negative expectancy: cut it from the plan.
• 30+ trades and clearly positive: keep it, and consider whether its grade sizes are right.

Also check these every month:

• Does grade predict results? A+ should beat A, and A should beat B. If B setups perform as well as A+, your grading isn't discriminating, and you need to tighten the checklist.
• Day of the week: some traders find one day consistently negative (often around major news days).
• Costs: total commission and fees in R. If costs are eating more than about 0.1R per trade, you're trading too often or with stops too tight.

When to raise your risk

Raising risk is the most dangerous decision in your plan, because it usually happens after a good run, which is exactly when overconfidence (Chapter r7) is highest. So make it a rule, not a feeling.

Raise risk by one step only if all of these are true:

1. At least 100 trades logged at your current risk level.
2. Expectancy over those trades is at least +0.20R after costs.
3. Process score is at least 90% over the last 4 weeks.
4. You haven't hit the −10% drawdown trigger during that period.
5. You're at or within 3R of your equity peak (don't size up in a drawdown).
6. State check scores are mostly in the full-size range.

Steps: 0.25% → 0.5% → 0.75% → 1.0%. One step at a time, and 1.0% is the ceiling for personal accounts in the EdgeFlow plan (Chapter r2 explained why). This is also why traders who use the Kelly criterion use only a fraction of it: full Kelly assumes you know your edge exactly, and you never do.

Lower risk automatically if any of these happen:

• −6R week → half size for the rest of the week.
• −10% from peak → quarter size, full review, and only return to normal by stepping back up (Chapter r4's recovery protocol).
• Process score under 75% for two weeks in a row → drop one step until it's back above 90% for two weeks.

Your first 90 days

Here's a structured way to put the whole module into practice. The goal of the first 90 days isn't to make money. It's to build the habits and the data that will make money later.

At day 90, run the "when to raise your risk" checklist. If you pass, you've earned the next step. If you don't, you haven't failed. You've found exactly what to fix, which is the whole point of a review process.

A few practical habits for the 90 days:

• Same routine every day: state check, plan written, session, journal, done. Routines remove decisions.
• Track the streak that matters: consecutive days with a 100% process score, not consecutive green days.
• Share your plan with someone. Ask in any EdgeFlow chat for a peer to review your one-page plan. Saying your rules out loud to someone else makes them harder to break.

Review the moduleGlossary flashcards →