Part 4: Checking and deciding · Chapter 13 of 14 · 12 min
A risk plan on trial
A 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%.
Why a written plan
In the previous chapters we measured risk and drawdown. What remains is the hardest part: applying them when the account falls. At that moment judgment worsens: you raise the risk to recover, you change the rules, you skip the system. A risk plan serves to decide beforehand, with a cool head, what to do in each situation, so that afterwards you only have to execute.
A plan does not create an edge. It limits how much you can lose and gives you a fixed moment to stop and look at the data.
The model plan
Here is a plan of seven rules, which we use as an example and which you can adapt:
- Risk per trade: 1% of current capital (fixed fraction, chapter 5), computed from the stop (chapter 8). Never increased after a loss.
- Positions: one at a time, or a maximum total open risk of 2% (chapter 11).
- Daily limit: after −2 R in the day, no new trades until the next day.
- Drawdown from the high: at −10% the risk is halved (0.5%); you return to 1% when the drawdown falls below 5%. At −20% you stop: no trades until a review has been done.
- Review every 100 trades: mean R with the error, winning share, costs, drawdown against the expected 95th percentile (chapter 4), longest losing streak (chapter 3).
- Trial at a fixed number: the number of trades (or the period) with which the system is judged is decided beforehand (chapter 12); until then the edge is considered zero and the risk stays small.
- No changes halfway: the rules change only at the review, in writing, with the date and the reason.
Before every trade, a checklist: is the stop where the idea is wrong? Are the ounces computed from the stop, rounded down? Does the total open risk stay under the cap? Is the margin used under 30%? Has the daily limit not been reached? Is the cost in R under 5%? After the trade: result in R in the log, cost, a note on whether the rules were respected.
What each rule does on the data
Let us apply the plan to the real series, rule by rule. Final capital, maximum drawdown and trades made, with base risk 1% and 2%.
LA (552 trades):
| Rules | Risk 1% | Risk 2% |
|---|---|---|
| None | 81% / dd 30.7% / 552 trades | 62% / dd 53.3% / 552 |
| Daily limit −2 R | 80% / dd 31.0% / 515 | 61% / dd 53.6% / 515 |
| Drawdown steps (−10% half risk, −20% stop) | 93% / dd 20.1% / 520, stopped on 16 September | 99% / dd 20.8% / 443, stopped on 14 August |
| All | 93% / dd 20.2% / 484, stopped on 15 September | 98% / dd 20.3% / 339, stopped on 26 June |
LD (367 trades):
| Rules | Risk 1% | Risk 2% |
|---|---|---|
| None | 109% / dd 17.4% / 367 | 115% / dd 32.2% / 367 |
| Daily limit −2 R | 104% / dd 18.0% / 340 | 106% / dd 34.0% / 340 |
| Drawdown steps | 110% / dd 14.1% / 367 | 101% / dd 20.1% / 319, stopped on 8 September |
| All | 107% / dd 14.4% / 340 | 109% / dd 20.6% / 318, stopped on 17 September |
LA + LD one position at a time (606 trades):
| Rules | Risk 1% | Risk 2% |
|---|---|---|
| None | 86% / dd 27.1% / 606 | 70% / dd 48.9% / 606 |
| Daily limit −2 R | 78% / dd 34.6% / 561 | 57% / dd 59.0% / 561 |
| Drawdown steps | 96% / dd 19.1% / 606 | 99% / dd 20.3% / 375, stopped on 21 June |
| All | 92% / dd 20.3% / 505, stopped on 8 September | 99% / dd 20.6% / 242, stopped on 28 April |
Three things.
- The drawdown step does the work. Where the system loses (LA, LA+LD), the rules stop the fall at about −20%, and final capital goes from 62% to 99% (LA at 2%) and from 70% to 99% (LA+LD at 2%). The plan stops, and the capital stays put.
- The daily limit does not change the numbers: 80% against 81% (LA 1%), 78% against 86% (LA+LD 1%). The losses of one day do not anticipate those of the next (chapter 3: chance has no memory), so skipping the following trades removes wins and losses in equal measure. Its usefulness is behavioural: it stops you chasing the losses of a bad day.
- Stopping costs when the system then recovers. LD at 2% is stopped on 8 September with 101% of capital, whereas without rules it would end at 115%. It is the price of every stop rule: sometimes it stops you just before the recovery.
Read the chart in words
Daily gold chart with, below, two curves of the capital of LA at 2% with and without drawdown rules and two horizontal lines at 100% and 80%.
Read the chart in words
Daily gold chart with, below, two curves of the capital of LA and LD together at 2% with and without drawdown rules and two horizontal lines at 100% and 80%.
And when the edge is there?
A stop rule protects from a system that loses; but if the edge is true, how much does it make you lose? On 3,000 random paths of 500 trades drawn from the results, with the drawdown rules (−10% halves, −20% stops):
| Series, risk | Without rules: median capital (5th percentile) | Touches −30% | With rules: median capital (5th) | Touches −30% | Paths stopped |
|---|---|---|---|---|---|
| LA 1% | 82% (57%) | 33.8% | 88% (80%) | 0% | 52.1% |
| LA 2% | 64% (31%) | 74.1% | 86% (80%) | 0% | 94.9% |
| LD 1% | 114% (78%) | 4.2% | 107% (84%) | 0% | 11.7% |
| LD 2% | 123% (58%) | 26.8% | 103% (81%) | 0% | 70.8% |
| SX 1% | 276% (193%) | 0% | 270% (182%) | 0% | 0% |
| SX 2% | 724% (353%) | 0% | 572% (236%) | 0% | 3.4% |
(LA and LD have a null or almost null edge, SX has a built-in edge of 0.21 R.) The balance:
- Without an edge the plan protects a lot: LA at 2%, the 5th percentile goes from 31% to 80%, and no path touches −30% (down from 74%). But the plan stops 95% of the paths: it tells you, almost always, to stop.
- With a true edge it costs little at small risk and something at higher risk: SX at 1% is never stopped and loses 2% of median capital; at 2% it is stopped in 3.4% of cases and the median capital falls 21% (from 724% to 572%).
- The signal is the frequency of stops: if your plan stops you in more than one case in five, either the risk is too high for the system or the edge is not there. LD at 2% is stopped in 71%: the rule tells you that at that risk, with a mean close to zero, the system does not hold.
What happens when you stop
Your account has fallen 20% from the high and the plan says to stop. The review has three questions, with the numbers of this book:
- Was the drawdown within the expected? Compare it with the 95th percentile at 250-500 trades (chapter 4). If it exceeds it, the system is worse than you thought.
- Is the mean R, with its error, still compatible with the edge you had estimated? (chapters 2 and 12). With few trades the answer is almost always "I do not know": in that case the decision is about the risk, not the system.
- Are costs and execution those assumed? (chapters 8 and 10.)
Then you decide, in writing: resume at half risk for a fixed number of trades, change the system (and start from zero with the trial), or quit. You do not resume the next day with the same size.
What to do with it
- Write the plan before trading, on one page, with numbers and not intentions ("1%", not "a little").
- Use the drawdown step: it is the rule that makes the difference on the data.
- Keep the daily limit as a discipline rule, knowing it does not improve the result.
- Count the stops: if you stop often, lower the risk or question the edge.
- Change the rules only at the review, in writing, never with the account in the red.
Procedure in four steps
- Write the plan on one page: risk, positions, daily limit, drawdown steps, reviews, number of trades of the trial.
- Before every trade, go through the checklist; afterwards, record the result in R.
- Every 100 trades do the review with the numbers of the book: mean ± error, drawdown against the expected, costs.
- When a threshold triggers, execute the rule without arguing; decide the changes only at the next review.
Exercises
Chapter exercises
1. In the model plan, at what drawdown from the high do you stop and do the review?
2. On the real series, did the daily limit at −2 R improve final capital?
3. The plan on LD at 2% stops on 8 September with 101%, whereas without rules the series ends at 115%. What does it show?
4. A system with a true edge is stopped by the plan in 3.4% of paths, one with a nearly null edge in 71%. How do you read it?
Have a question or want to share your exercise?
Post in the community, or join the free signals room on Telegram.
Educational content, not financial advice. Trading involves risk.