FREE BOOK · RISK

Risk and position sizing from A to Z: simulations on real results

The Academy's eleventh book in 14 chapters: what risk is, the series of results, losing streaks, the drawdown, fixed size and fixed fraction, the risk of ruin, the Kelly criterion, the stop and the ounces, leverage and margin call, costs against the stop, several positions at once, how many results are needed and a risk plan on trial. Each chapter explains how it works, what we simulated on the real series of the previous books and what to do with it. Mistakes, glossary and a final test.

14 chapters · 4 modules · 3 hours of reading, approx. · updated 2026-11-20

How to use this book

Every chapter has figures on real gold candles, four exercises with the solution explained and a recap of the key points. Do the exercises before you move on to the next chapter. Educational content, not financial advice.

Part 1: Risk and its numbers

What risk is, the series of results, losing streaks and the drawdown.

  1. 1What risk isThe risk of a trade is the loss you accept before entering, and it is measured in R: 1 R is the distance between entry and stop. Three real series from the previous books give mean R of −0.03, +0.03 and +0.09 per trade, with a deviation of about 1 R (0.56 for the hours system): over 100 trades the result swings by ±19 R, far more than any edge we found. That is why the book is about how much to risk, not what to buy. · 10 min
  2. 2The series of resultsA series of trades is a sample: run again, it would give a different sequence. With a deviation of 1 R the error on the mean is ±0.19 R after 100 trades and ±0.08 after 552. In our series, 100 consecutive trades went from −24 to +11 R (levels bounce) and from −16 to +19 R (swing break); even the example with a true edge of 0.21 R had a block of 112 trades at +8 R against another at +36. · 12 min
  3. 3Losing streaksLosses in a row are arithmetic, not bad luck: with 50% of trades winning, in 100 trades there is an 81% chance of at least 5 consecutive losses and 17% of at least 8. In our series the longest is 7 losses (levels bounce, 21-24 July 2026), 6 (swing break) and also 7 in the example with 60% of trades winning. Chance has no memory: after five losses the next trade is worth the same as any other. · 11 min
  4. 4The drawdownThe drawdown is the distance between the account and its previous high. In our series the maximum fall is 35.3 R (levels bounce), 18.9 R (swing break), 1.7 R (Asia hours) and 9.0 R even in the example with a true edge; the series with an edge spent 64% of trades below its own high, those without a proven edge from 84% to 96%. To recover a −50% you need +100%. With 2% risk, 20 R of drawdown is −33% of the account. · 12 min

Part 2: How much to risk

Fixed size and fixed fraction, risk of ruin and the Kelly criterion.

  1. 5Fixed size and fixed fractionTwo rules to decide how much to risk: the same amount on every trade (fixed size) or the same percentage of current capital (fixed fraction). On the real series the two rules give the same result when risk is small (LA at 1%: 82% and 81%) and diverge as it grows (SX at 5%: 670% against 15,049%, drawdown 17.8% against 37.4%). The martingale, which doubles after every loss, closed in profit on the real LA series and wiped out the account or took it below 5% in 72% of 4,000 random orders of the same trades (LA, base 1%). · 13 min
  2. 6The risk of ruinThe risk of ruin is the probability of touching, at least once, a loss you can no longer afford: here −30% and −50% of the account. Without an edge it grows with time: at 2% risk per trade, in a 50%-winning model a 50% loss arrives in 26% of cases in 500 trades and in 43% in 1,000. With a true edge it stays put: SX, at 10% risk, touches −50% in 9-10% of paths at any horizon, and at 1-2% almost never. · 12 min
  3. 7The Kelly criterionThe Kelly criterion gives the fraction of capital to risk to make the account grow as fast as possible: f ≈ μ/σ² (mean divided by variance, in R). On our series it is 0 for LA, 2.9% for LD (with an error from −7% to +13%), about 21% for SX. But at full Kelly risk the median drawdown of SX is 94% and the probability of touching −50% is 46%: maximum growth is paid for with an account nobody can bear. Half Kelly gives 75% of the growth, and at twice Kelly capital falls even with a true edge. · 11 min

Part 3: From theory to the account

Stop and ounces, leverage and margin call, costs against the stop and several positions at once.

  1. 8The stop and the ouncesThe ounces to buy are the dollar risk divided by the stop distance: with 10,000 $ and 1%, a stop of 12.2 $ is worth 8.2 ounces and one of 43.3 $ is worth 2.3, and the risk stays 100 $. In our series the stop goes from 5 to 175 $ (median 21.7 $), so the quantity changes by almost four times from one trade to another. The stop, though, is an order, not a guarantee: in LA, in 2 stopped trades out of 276, the candle opened beyond the stop and the loss would have been about −2.6 R instead of −1 R. · 11 min
  2. 9Leverage, margin and margin callThe leverage you use does not decide the risk: the stop with the size does. Leverage used = risk % ÷ stop as % of price: with 1% risk and the median LA stop (0.48% of the price) leverage is 2.1 times, on median; at 5% it is 10.4 times, and 9.4% of trades would need more than 20 times. The forced close (margin call) arrives only if you do not respect the stop, hold too many positions or there is a jump: at 20 times a 2.5% move against you is enough (112 $ on 4,495 $) and in 11.5% of the days in the data a purchase at the open saw this move. · 10 min
  3. 10Costs against the stopThe cost of a trade is fixed in dollars, so in R it weighs more the narrower the stop: 0.3 $ is worth 0.007 R with a 40 $ stop and 0.10 R with a 3 $ one. In the series of the books the average cost is 0.015 R per trade (8.5 R in total over 552 trades, 8.5% of a 10,000 $ account at 1%). LD, with +0.03 R on average, breaks even at a cost of 0.87 $ and loses from 1 $ up; SD, with wide stops, holds up to 8 $. A rule: keep the cost under 5% of the stop. · 11 min
  4. 11Several positions at onceTwo positions open together do not halve the risk: they add it up according to how correlated they are. Running LA and LD together, 62% of LA trades overlap with an LD one; when they go the same way the results are correlated (+0.41) and coincide (both winning or both losing) in 65% of cases, against 47% expected if they were independent. The drawdown of the combined series (42.7 R) exceeds that of each (35.3 and 18.9), and the worst day (16 September 2026) closed 7 trades at −5.1 R. · 11 min

Part 4: Checking and deciding

How many results are needed, a risk plan on trial, mistakes, glossary and final test.

  1. 12How many results are neededTo see an edge of 0.1 R per trade, with a deviation of 1 R, you need about 385 trades for the error to equal the mean and about 785 to have an 80% chance of noticing it. At the pace of LA that is 14 months, at that of LD 20, at 5 trades a month 13 years. With 100 trades a true edge of 0.1 R is recognised only 17.5% of the time; and if you check every 25 trades a system with no edge looks "significant" at least once in 16% of paths. In our real series the lower limit of the interval is never above zero. · 10 min
  2. 13A risk plan on trialA risk plan is a set of rules written beforehand: risk per trade, cap on positions, daily limit, drawdown steps and reviews. Tested on the real series, the drawdown step (half risk at −10%, stop and review at −20%) changed a lot: LA at 2% goes from 62% to 99% of capital, and on 500 random trades the 5th-percentile capital rises from 31% to 80%. The daily limit at −2 R did not help (80% against 81%): it serves discipline, not the numbers. The price of the plan: with a true edge, SX at 2%, the median capital falls from 724% to 572%. · 12 min
  3. 14A ten-month account, mistakes, reference, glossary and final testThe plan of chapter 13 applied for ten months to LA and LD with one position at a time and 1% risk: without rules the 10,000 $ account closes at 8,640 $ (drawdown 27.1%), with the plan at 9,235 $ (20.3%), because the system has no edge (−0.02 ± 0.08 R) and the plan limits the loss without avoiding it. Then the most frequent mistakes in chapter order, the quick reference with formulas and measurements, the glossary and a final test of 15 questions. · 17 min

Another course