Size Positions and Limit Risk

Choose a sizing method

There are only a few sizing approaches worth considering, and the differences between them matter less than the discipline of picking one.

What you are choosing is how your stake responds to your balance and to your recent results. Those two behaviours define the method, and everything else is detail.

Fixed-fraction stakes

Stake the same percentage of the current balance every time. The stake grows as the balance grows and shrinks as it falls, which produces two useful properties without any decision from you: it decelerates through a losing run and it compounds through a good one. This is the default for a reason, and most traders who last are running some version of it.

Consistency over emotion

The competing approach in practice is not another method but the absence of one: staking whatever seems right. This feels responsive and is a reliable way to end up largest on your worst trades, because conviction peaks in exactly the situations where you are most likely to be reading the market through a recent result. Fixed sizing has no view about your conviction, which is its main advantage.

Fixed cash amounts deserve a word because they are common among beginners and look like discipline. Staking the same money figure every time is consistent in one sense and dangerous in another: as the balance falls, that unchanged figure becomes a larger and larger proportion of what remains. The method that felt conservative at the start becomes aggressive precisely as the account weakens. Recalculating from the current balance takes a moment and removes the effect entirely.

Simplicity first

Complicated sizing schemes, scaling by volatility or by setup grade, are defensible in principle and premature for almost everyone. They add a variable to your record at a stage where you are still trying to find out whether the entries are worth anything, and a record with two moving parts answers neither question. Start with one fraction, applied to everything, and revisit the idea after you have a few months of readable results.

One exception is worth allowing: a smaller fixed fraction for a trade type you know carries more ways to be wrong, such as the reversal setups described elsewhere on this site. That is still fixed sizing; it is two fixed fractions attached to two defined categories, decided in advance. It is not the same as adjusting per trade, and the difference is whether the number was chosen before you saw the chart.

One fraction of the current balance, applied to everything. Add sophistication only once the record is readable.

Understand martingale danger

Doubling after a loss is the most persistent idea in short-horizon trading, and the arithmetic that kills it is short enough to write out.

It persists because it works almost every time, which is precisely the problem. A method that succeeds repeatedly and fails catastrophically produces a long run of confirmation before the event that matters.

Doubling after losses

The idea is that after a loss you increase the next stake enough to recover the loss and make the original gain, so any eventual win resets you to profit. On a fixed-payout contract that pays less than it risks, the required increase is more than doubling, which accelerates the sequence faster than the familiar version suggests.

Rapid ruin risk

The stake sequence grows geometrically, so the number of losses you can absorb is small and fixed by your balance. The illustration below is arithmetic, not a measured result: it shows how the required stake grows across a losing run when each step must recover everything lost so far plus the original target.

Ranges are given rather than exact figures because the multiple depends on your payout percentage, which varies. The shape is the point: within a handful of losses the required stake exceeds any sensible balance, and runs of that length are entirely ordinary. A sequence that has no memory produces streaks, and a method that cannot survive a streak will meet one.

Softer versions circulate under different names and deserve the same treatment. Adding a partial increase after a loss, or recovering across two trades instead of one, slows the sequence without changing its direction. The test is simple: does the stake rise because the previous trade lost? If yes, the method contains the same failure, arriving later and therefore with more confidence built up behind it. There is no version of increasing after losses that is safe, only versions that take longer to demonstrate the point.

Why it fails

Three separate constraints each end it independently. Your balance is finite. The platform has a maximum contract size. And your own nerve, which is the constraint that usually arrives first, fails somewhere around the point where a single contract represents more than you were originally willing to lose in a week. Any one of these is sufficient; you do not need to be unlucky, only to trade long enough.

Doubling wins often and loses everything. The long run of small recoveries is the mechanism, not a defence.

Limit exposure per session

Sizing controls the cost of one trade; session limits control the cost of a bad afternoon, and the second is usually the larger number.

A fixed fraction bounds each contract and bounds nothing about how many contracts you take. Twenty small trades in a deteriorating state of mind can cost more than any single position would have.

The two limits also fail differently when they are missing, which is worth knowing. Without a count, a session drifts long and shallow: many small trades, none of them planned, ending in a loss nobody can attribute to a decision. Without a money cap, a session can end badly in four contracts. Traders usually have a natural weakness toward one of these, and the log tells you which within a fortnight.

Trade-count caps

Decide how many contracts a session may contain, based on how many qualifying setups your rules actually produce rather than on how many the chart offers. This number is easy to set and easy to check, and reaching it is the clearest available signal that you have moved from your plan to something else. Log the count every session and the pattern becomes visible within two weeks.

Loss limits

A loss cap in money terms works alongside the count, because a session can reach an uncomfortable loss well before it reaches the trade limit. Whichever arrives first ends the session. Setting both is not belt and braces; they catch different failures, one of frequency and one of depth.

One detail makes limits far more likely to hold: decide them when you are not trading. Numbers chosen on a quiet morning are chosen by someone with no position and no recent loss, and they tend to be sensible. Numbers chosen at the screen, mid-session, are chosen by someone who wants to continue. Writing the three limits down the evening before takes two minutes and removes the negotiation entirely.

Walking away

The clock is the third limit and the most underrated. Attention degrades with time at the screen, and the trades taken in the last half hour of a long session are reliably worse than the ones taken in the first. A time limit costs you nothing on a good day and removes the tail of a bad one. You can run a fixed fraction for a fortnight on virtual funds to find out how long your own attention actually lasts before the entries start loosening.

• Count, money and clock. Three limits, first one ends the session.
• Set them from your rules, not from what the chart is offering.
• Record which limit you hit. The pattern names your weakness.
• No exceptions on good days. An exception once is an exception always.

Bound the session three ways: contracts, money and time. Whichever arrives first ends it.

Keep stakes emotion-free

Every sizing failure has the same shape: a number that was fixed in advance moved because of how the previous trade felt.

Naming the two states that produce the move makes them easier to catch, because both announce themselves clearly if you are looking.

No revenge sizing

After a loss, particularly one that felt unfair, the stake wants to go up. The reasoning offered internally is that the setup was correct and the market was wrong, so a larger position on the next one restores the balance. Nothing about that reasoning is sound: the next trade is unrelated to the last, the market has no view about your balance, and your read is currently being made by someone who is annoyed. If you feel the urge, that feeling is itself the reason to stop for the day.

Both states share a physical signature that is easier to notice than the reasoning behind them: a slight urgency about placing the next contract. Calm trading feels unhurried even when the setup is time-limited. If you find yourself moving quickly toward the stake field, that is the moment to check the number against what you wrote down, because the urgency arrived before the justification did.

No euphoria sizing

After a good run the stake also wants to go up, and this version is harder to resist because it feels like competence rather than desperation. A short sequence of wins is usually variance, and sizing up on it means your largest positions arrive just as the sequence reverts. Under fixed-fraction sizing your stake already grows with the balance, which is the version of pressing an advantage that does not require a decision.

A second habit worth having is a physical separation between deciding and acting. Calculate the session stake once, write it on paper, and place every contract at that figure without opening the calculator again. It sounds fussy and it works, because the calculation is the moment where a small adjustment enters unnoticed. Traders who report the fewest sizing problems are almost always the ones who removed the opportunity rather than the ones with more willpower.

Rules over feelings

The structural fix is to make the stake something you do not decide during a session. Write the fraction down before you start, calculate the actual amount once, and keep it visible. A number that has to be recalculated per trade is a number that can be adjusted per trade, and the adjustment always finds a justification.

There is a diagnostic worth running monthly. Go through your log and compare the average stake on trades that followed a loss with the average on trades that followed a win. Under fixed-fraction sizing those two figures should be almost identical. If they are not, the sizing method you are actually running is not the one you wrote down, and that gap explains more about your results than any entry rule.

Compare your average stake after losses with your average after wins. Any gap is your real sizing method.

Sizing takeaways

Sizing is the least interesting decision in trading and the one that most reliably separates outcomes.

Consistent stakes win

Not in the sense of guaranteeing a gain, which nothing does, but in the sense that a consistent stake is what allows a method to express whatever it is worth across a run of trades. Variable stakes weight your results by mood, so a month of trading tells you about your mood rather than about your rules. Consistency is what makes the record mean something.

Avoid martingale

The arithmetic is decisive and does not depend on anything about the market. A geometric stake sequence meets a finite balance, a platform maximum and your own nerve, and it meets at least one of them within a handful of losses. Streaks of that length are ordinary, which means the failure is not a risk to be managed but an event to be waited for.

• One fraction, current balance. Applied to everything.
• No doubling, ever. Not as a system, not as a one-off recovery.
• Three session limits. Count, money, clock.
• Run the after-loss versus after-win comparison monthly. It reveals the truth.

It is worth remembering what sizing cannot do, so that it is not asked to do it. A stake rule does not improve your reads, does not tell you when to trade, and cannot turn a rule set that sits below break-even into one that clears it. What it does is control the rate at which your method reveals itself and guarantee that you are still there when it has. That is a supporting role rather than a starring one, which is exactly why it is skipped and exactly why skipping it is expensive.

Limit each session

No figures are recommended here for the fraction, the caps or anything else, because the right numbers depend on your balance and your tolerance, and this desk has measured nothing. What can be said without measurement is that a fixed fraction decelerates in a decline while a doubling sequence accelerates, and that the difference between those two behaviours is the difference between a bad month and a finished account. Choose the numbers, write them down, and treat the writing as the part that matters.

Fixed fraction, no doubling, three session limits, and a monthly check that you are actually running them.