Two traders take the same signals over the same year. Identical entries, identical exits, identical win rate. One finishes the year up 30%. The other is down 60% and has stopped trading.
The difference is not the strategy. It is how much they risked on each trade, and this is the part of the process that receives the least attention relative to how much it determines.
Entry logic is interesting to discuss and easy to argue about. Sizing is arithmetic, it is not interesting, and it decides the outcome. What follows is the arithmetic, with the numbers worked through rather than asserted.
Expectancy comes first, and nothing rescues its absence
Before you can answer any sizing question, you must establish one thing: does the strategy have a positive expected value at all?
Expectancy per trade is straightforward. Multiply your win rate by your average win, subtract your loss rate multiplied by your average loss. Express both in units of risk, usually written as R, where 1R is the amount you lose on a losing trade.
Take a system that wins 40% of the time, with an average win of 2.5R and an average loss of 1R. Expectancy is (0.40 × 2.5) − (0.60 × 1.0), which comes to 0.4R per trade. Over two hundred trades, that is an expected 80R, before costs.
Now take a system that wins 70% of the time with an average win of 0.5R and an average loss of 1R. Expectancy is (0.70 × 0.5) − (0.30 × 1.0), which is −0.05R. It feels good to trade, because most trades win. It loses money.
This distinction matters more than any sizing rule, because sizing does not change the sign of expectancy. It only changes how quickly the sign expresses itself. A negative-expectancy system traded conservatively loses slowly. Traded aggressively it loses quickly. There is no third option, and no amount of risk management converts one into the other.
Losses and gains are not symmetric
This is what makes drawdown control mathematically urgent rather than merely prudent.
A 10% loss requires an 11.1% gain to recover. A 20% loss requires 25%. A 33% loss requires 49%. A 50% loss requires 100%. A 70% loss requires 233%.
The relationship is not linear and it accelerates sharply. The practical implication is that the cost of a large drawdown is not the drawdown itself but the amount of future performance consumed getting back to level, during which you are working for nothing.
It also means that a strategy producing 20% a year with a maximum drawdown of 15% is a fundamentally different proposition from one producing 25% a year with a 45% drawdown, even though the second has the better headline. The second one requires the operator to sit through a decline that most people will not sit through, and the recovery from it consumes nearly two years of the strategy’s own returns.
Risk of ruin, with actual numbers
There is a simple model that makes the stakes concrete. It assumes even-money outcomes and a fixed number of units of capital, which is a simplification, but the direction and magnitude of what it shows are sound.
Consider a trader with a 55% win rate at 1:1 payoff, so a genuine if modest edge. Risking 2% per trade, they have fifty units of capital. The probability of losing all of it, under this model, works out to roughly four in one hundred thousand. Effectively negligible.
Now change only the position size. Same edge, same win rate, but risking 10% per trade instead of 2%, so ten units of capital. The probability of ruin rises to approximately 13%.
Nothing about the strategy changed. The edge is identical. A one-in-eight chance of total loss was introduced purely by the sizing decision.
And one more case, to show where the cliff is: a trader with no edge at all, a 50% win rate at 1:1, faces a probability of ruin of exactly one, at any position size. Given enough trades, ruin is certain. Small size delays it. It does not prevent it.
The Kelly criterion, and why nobody sensible uses it fully
Kelly gives the position size that maximises the long-run growth rate of capital. For even-money bets the formula is simply your win probability minus your loss probability.
For that 55% win rate, Kelly says to risk 10% of capital per trade. Which is precisely the size that produced a 13% chance of ruin in the previous section.
Both results are correct. Kelly maximises expected logarithmic growth, and it does so while producing drawdowns that almost no human being will tolerate. Full Kelly routinely involves declines of 50% or more as a normal feature of operation, not as a malfunction.
There is a second problem, and it is the more serious one. Kelly requires you to know your win rate and payoff ratio precisely. You do not. You have estimates drawn from a limited sample, and those estimates have wide error bars. If your true win rate is 52% and you have estimated 55%, you are systematically overbetting.
The consequence of overbetting is worse than most people assume, because the growth curve is asymmetric around the optimum. Betting half of Kelly gives you roughly three quarters of the maximum growth rate with substantially less than half the volatility. Betting twice Kelly gives you an expected growth rate of approximately zero, while carrying enormous risk along the way. For that 55% example, betting 20% per trade produces expected growth indistinguishable from doing nothing at all.
Quarter to half Kelly is where practitioners generally settle, and the reason is the estimation error, not timidity.
Per-trade risk is not your real risk
A rule such as “never risk more than 2% per trade” is incomplete on its own, because trades do not exist in isolation.
Five open positions at 2% each is 10% of capital at risk. If those five positions are long AUD, long NZD, short JPY, long gold and long an equity index, they are not five positions. They are one bet on risk appetite, expressed five ways, and they will lose together on a single afternoon.
Track total open risk, and track it grouped by common driver rather than by instrument. A practical approach is to set a ceiling on aggregate open risk, something in the region of 6% for most retail accounts, and a tighter ceiling on any single correlated cluster. When a new setup would breach the cluster limit, either it does not get taken or an existing position is reduced.
Whether you trade with Xlence or with any other venue, none of these figures depend on the platform. They are properties of your own account and your own decisions, which is why they are worth settling before you open a position rather than during one.
Work backwards from the drawdown you can actually tolerate
Most people choose a per-trade risk figure first, usually 1% or 2%, because they read it somewhere. The better order is the reverse.
Start with the drawdown you can genuinely sit through without abandoning the strategy. Be honest rather than aspirational; the number is smaller than you think, and if you have not experienced a drawdown of that size before, assume your real tolerance is lower still.
Then look at your strategy’s historical worst run of consecutive losses, and add a margin, because your sample almost certainly understates the worst case. If the longest losing streak in three years was eight trades, plan for twelve.
Divide your tolerable drawdown by that number, and you have your per-trade risk. If you can tolerate a 15% decline and you plan for twelve consecutive losses, you arrive at roughly 1.25% per trade. That figure is derived from your own constraints rather than borrowed from a rule of thumb, and it will hold up when tested because it was built to.
Then commit to it in advance, including what happens if the drawdown limit is reached: reduce size, pause, or stop entirely. A limit without a predetermined response is not a limit.
The part that is not arithmetic
All of the above assumes you follow the sizing you set. In practice, the most common failure is not miscalculation but deviation, and it deviates in a predictable direction: sizes increase after wins and after losses, for opposite emotional reasons, and both increases arrive at the worst available moment.
The defence is mechanical. Calculate size from account equity before entry, using a formula rather than judgement, and record the intended size alongside the actual one. The gap between those two columns, reviewed monthly, is usually the most informative thing in a trading journal.
Trading leveraged products carries a high level of risk and can result in losses that exceed your deposits. Past performance is not a reliable indicator of future results. This article is educational and is not investment advice.