🛡️ TOOL · Is my edge real?

Size & Survive

Most traders think about being right. Surviving is about the other half — how much you put on, and what a rough patch does to you. Five quick, honest calculators. Nothing here is advice; it's arithmetic you can lean on.

How it works

Position sizer

1 You set what you are willing to loseA share of your account — not a share of the position. This is the only number you actually control before the trade.

2 Your stop decides the restThe gap between entry and stop is what one unit can cost you. A wider stop means a smaller position for the same risk.

3 We work backwards to the sizeAmount at risk divided by the distance to your stop. Size is an output here, never a guess.

“Units” means whatever one of the thing you are pricing is — one share, one coin, one unit of the currency pair. Entry and stop are prices in the same currency as your account (shown here in £). If you trade futures or spread bets, one “unit” is one contract or £1 per point, and you will usually need to round to what your broker actually lets you trade.

Risk per trade is a share of your account, not of the position. Risking 1% of £10,000 means £100 lost if the stop is hit — which is why the position can be far larger than £100.

What this can’t tell you

The arithmetic is exact. What it assumes about the real world is not.

It assumes you get out at your stop. That is the whole calculation. A gap through the level — overnight, on news, in a thin market — can cost more than the amount shown, sometimes far more. The figure is your planned loss, not your worst case.
Costs are not included. Spread, commission and slippage come on top of the loss shown. Cost of trading covers what they do to a strategy over time.
It does not know what you are trading. Contract sizes, minimum increments, lot sizes and margin requirements are all your broker’s rules — you will usually have to round the answer, and rounding up means risking more than you chose.
It says nothing about whether the trade is any good. Correct sizing on a strategy with no edge just controls how quickly you lose. Whether the edge is real is what Real or Noise? is for.
One trade at a time. If you hold several positions that tend to move together, your real exposure is the whole book, not each line — five “1% risk” trades in correlated markets can behave like one 5% trade.

Risk of ruin

1 You describe your edge and your bet sizeHow often you win, how big a win is next to a loss, and what share of the account you put at risk each trade.

2 We play out 2,000 possible runsEach run is 200 trades decided by chance in line with your win rate, re-sizing off the balance as it moves. This is a simulation, not a formula.

3 We count how many hit your ruin lineThe answer is simply the share of those 2,000 runs that fell as far as the depth you set.

“R” is whatever you risk on one trade, so a loss is −1R and a 1.5R win makes back one and a half times what you put at risk. It is the same unit Real or Noise? uses.

“Ruin” here means the depth you choose — not going to zero. Set the last slider to 50% and the answer is the chance of ever being down 50%, which most people would stop trading long before. That is the point: pick the loss you would not come back from.

What this can’t tell you

A simulated answer is only as good as what it is told to simulate.

It takes your edge as fact. The win rate and payoff you set are treated as true and unchanging. If they are optimistic — or were found by testing lots of variations — the real risk of ruin is higher than shown. Real or Noise? is where you check that first.
It is a simulation, so it is an estimate. 2,000 runs of 200 trades gives a stable figure, not an exact one, and a small number like “1%” should be read as roughly that rather than precisely.
200 trades is the horizon, not your career. Trade for longer and there is more time for a bad run to arrive, so the lifetime risk is higher than the number here.
Every trade is assumed independent and the same size in percentage terms. Real trading has clustered losses, correlated positions and changes of mind, all of which make deep drawdowns more likely than this shows.
Costs, gaps and slippage are not modelled. Each of them makes ruin more likely, never less.
Surviving is not the same as succeeding. A low number here only means you are unlikely to be wiped out at this size. It says nothing about whether the strategy is worth trading.

Drawdown maths

1 You set how far down you areA drawdown is the fall from the highest point your account has reached — not from what you started with.

2 We work out the gain needed to get backThe climb is always steeper than the fall, because the gain has to be earned on the smaller balance the loss left behind.

3 We estimate how long that would takeAt the average yearly return you set, compounding steadily.

Why the two numbers never match. Lose 50% of £10,000 and you have £5,000. Making 50% of £5,000 gets you to £7,500 — not back to even. You need to double it, so a −50% fall needs a +100% gain. The deeper the hole, the faster that gap widens.

What this can’t tell you

The gain needed is exact arithmetic. The recovery time is the part to treat carefully.

“Time to recover” is the optimistic case. It assumes you earn your average return steadily, every year, starting immediately, and never fall again on the way back. Real recoveries are interrupted by further drawdowns, which is why they usually take longer than this.
An average return is not an annual return. Averaging 12% a year does not mean making 12% each year — and a run of flat or negative years at the wrong moment stretches the recovery well past the figure shown.
It cannot tell you whether this drawdown is normal. Every strategy has a depth it visits routinely and a depth that means something has broken. This tool does not know which is which — Risk of ruin is where you ask how likely a given depth is.
It assumes you are still trading. The maths does not care how a long drawdown feels, but most people change size, change approach, or stop entirely — and any of those changes the recovery completely.
Costs, tax and withdrawals are not included, and each of them lengthens the climb.

Expectancy

1 You give three numbersHow often you win, what an average winning trade makes, and what an average losing trade costs.

2 We weigh them against each otherYour wins multiplied by how often they happen, minus your losses multiplied by how often they happen.

3 That gives what one trade is worthAveraged across every trade you take — the winners and the losers together.

These two inputs are different from the answer. “Average win” is the average of your winning trades only, and “average loss” the average of your losing trades only. Enter the loss as a positive number — it is a size, and the calculation already knows it works against you. Expectancy, the answer, is the average across all trades.

You have met this number before. It is the same idea Real or Noise? measures in R, and the same figure Cost of trading asks for as “you make, before costs” — just in pounds here. All three mean: what one trade is worth on average, once winners and losers are put together.

What this can’t tell you

An average is a summary. It deliberately throws away the thing that makes trading hard.

It says nothing about the order things happen in. A positive expectancy still arrives through winning and losing streaks, and a long enough losing run can end an account that was “profitable on average” the whole time. Risk of ruin is where that question belongs.
It is before costs. Spread, commission and slippage come off every trade, winners and losers alike — see Cost of trading for what that does to a thin edge.
It cannot tell you the figure is real. It takes your three numbers at face value. Whether a record like that is skill or luck is what Real or Noise? tests.
Averages are easily distorted by outliers. One enormous win can lift your average win well above what a typical winner actually looks like, making the whole calculation flattering.
It assumes those numbers hold steady. Win rate and trade sizes drift with market conditions, position size and your own behaviour.
Profit factor with few trades means very little. Over a handful of trades it swings wildly and can look impressive on nothing at all.

Kelly

1 You describe the betHow often it wins, and the size of a win next to the size of a loss.

2 Kelly finds the fastest-growing stakeThere is one bet size that grows an account quickest over a very long run. Bet less and you grow slower; bet more and you also grow slower, while risking far more.

3 We show the fractions people actually useBecause the fastest route is a punishing ride, and it depends on knowing your edge exactly.

What the percentage means — read this before acting on it. Kelly’s answer is the share of your account you lose if the trade loses. It is the same thing the Risk of ruin tab calls “risk per trade”, not the size of the position. So a full Kelly of 25% means a losing trade costs a quarter of everything.

Kelly is a ceiling, not a target. It assumes you know your win rate and payoff exactly. Overstate your edge even slightly and the “optimal” number is above true Kelly — where growth falls and risk climbs. That is why the fractions matter more than the headline. Take the figure over to Risk of ruin and see what it does before treating it as a size.

What this can’t tell you

Kelly is the most misused number on this page. These are the reasons why.

It assumes you know your edge exactly. You do not. Kelly is built for a known bet, like a card count — not an estimated trading edge. Every real edge is a guess with error bars around it, and Kelly has no way to be told that.
Overbetting is worse than underbetting. Past full Kelly, growth falls while risk keeps rising, so an overstated edge is punished twice. Half of a correct Kelly keeps most of the growth; double it and you can lose money on a genuinely winning system.
Full Kelly is far rougher than most people expect. Drawdowns of 50% or more are a normal feature of it, not a sign anything has gone wrong. Almost nobody can hold a position through that, which makes the theoretical optimum practically unusable.
It is a two-outcome model. One typical win, one typical loss. Real trades come in a spread of sizes, and the occasional very large loss hurts a Kelly-sized position badly.
It optimises for one thing only — long-run growth. Not for smoothness, not for sleeping at night, not for a career that has to survive the next twelve months. Those are usually what actually matters.
It cannot tell you the edge is real. Feed it a mined or lucky record and it will confidently size a strategy that does not work. Real or Noise? comes first, always.

The thinking behind it

Losses and gains aren't symmetric

READ →

Win rate is only half the story

READ →

Size is survival

READ →

You know roughly what the edge is worth and how much to put on. Now see the honest range of what that could become over years — not one hopeful line.

What's Possible →