Risk Management for Prop Traders: Sizing, R-Multiples, Kelly

Ask ten funded traders why they blew their last evaluation. Nine will describe a strategy problem – the setup didn’t work, the news move went the wrong way, the spread widened. Almost none will mention position sizing. Almost all of them had a sizing problem.

Here is the math that explains why. Suppose you have a real edge – say, a strategy that wins 55% of the time with a 1:1 reward-to-risk ratio. That’s a perfectly good edge. Plug it into the expectancy formula:

E = (Win% × Avg Win) − (Loss% × Avg Loss) = (0.55 × 1R) − (0.45 × 1R) = 0.10R

You expect to earn 10% of the amount you risk on each trade, on average. If you risk 2% of your account per trade, that’s 0.2% expected per trade. Over 100 trades, the expected return is 20%. Sounds fine.

But expected return isn’t what kills you – variance is. With 100 trades at 2% risk, a Monte Carlo simulation of that edge shows a roughly 20–30% probability of hitting a –10% drawdown somewhere along the path. On a funded account with 8–10% max drawdown, you have a real risk of failing the evaluation even with a profitable strategy. Lower your risk per trade to 0.5%, run the same simulation, and the drawdown distribution compresses dramatically.

Risk management is not a feel-good topic. It’s the lever that decides whether a real edge ever gets to compound.

The classic 1% rule says: risk no more than 1% of your account equity on any single trade, where “risk” is the distance from entry to stop-loss × position size. It predates electronic trading and comes from the pit-era observation that sizing was the single largest explanatory variable in trader survival.

On a personal live account, 1% is a sane default for discretionary trading. On a prop-firm evaluation with an 8–12% overall drawdown and a 4–5% daily loss limit, 1% per trade is already aggressive:

• A 5-trade losing streak at 1% risk = –5% account. That’s the entire daily loss limit on one bad morning.
• Add typical spread, commission, and slippage, and a theoretical 1% loss often realises at 1.1–1.3%. The slippage compounds faster than most traders model.
• Overlap of correlated trades (EUR/USD long + GBP/USD long) doubles or triples your effective exposure without feeling like it.

Practical recommendation for evaluation-phase trading: 0.25–0.5% risk per trade, with a maximum concurrent exposure cap of 1.0–1.5% across all open positions. If your strategy doesn’t produce enough R per week at 0.5% sizing to hit the target in the evaluation window, the answer is almost always to improve the strategy or pick a different evaluation type – not to raise the risk.

For any trade, the correct position size is:

Position size = (Account equity × Risk%) / (Entry − Stop distance × Pip value)

That’s it. Every trading calculator on the internet is implementing some variation of this. A worked example on a $50,000 funded account, risking 0.5%:

1. Risk budget = $50,000 × 0.005 = $250.
2. Trade = EUR/USD long at 1.0800, stop at 1.0780 → 20-pip stop.
3. Standard lot pip value for EUR/USD = $10 per pip.
4. Position size = $250 / (20 × $10) = 1.25 standard lots.

Three things that reliably break the formula in practice:

• Spread and commission. A 20-pip stop with a 1-pip spread is effectively 21 pips to the stop, reducing the effective risk ceiling.
• Slippage on stop. Real stops fill, on average, worse than the stop price – particularly during news. Budget for 10–30% extra on the stop.
• Correlated positions. Two EUR-denominated longs at 1% each is not 2% exposure; it’s closer to 1.7–1.9% depending on correlation. Conservative traders treat correlated trades as one position.

The free position-size calculator on the site handles these mechanics automatically.

The idea, popularised by Van K. Tharp, is simple: measure trade results not in dollars, but in units of the amount you risked (R). A trade where you risked $250 and made $500 is a +2R trade. The same strategy executed on a $5K account or a $500K account produces the same R distribution; only the dollar amount changes.

Why this matters for prop traders:

• Expectancy becomes account-size-independent. Your strategy has an expectancy of, say, +0.6R per trade whether you’re on a $25K or $200K account.
• Daily loss limits become clearly defined. If your daily loss limit is 4% and you risk 0.5% per trade, you can sustain 8R of losses in a day before disqualification. That’s a concrete, trackable number.
• Journaling improves. Logging every trade in R lets you diagnose which setups are +EV and which are leaking R, without account-size noise.

A reasonable target for a funded-account trader: an expectancy of +0.3R to +0.7R per trade, with R-variance low enough that a 10-trade losing streak doesn’t cross the daily loss limit. If your expectancy is higher than +1R, either you’re a genuinely excellent trader or (more likely) your sample is too small.

The Kelly criterion, introduced by J. L. Kelly Jr. in a 1956 Bell System paper, gives the bet-fraction that maximises the long-run geometric growth rate of a bankroll given known win probability and odds:

f* = (bp − q) / b

where p is win probability, q = 1 − p, and b is the net odds received on a winning bet. For our 55/45 edge with 1:1 R/R, b = 1:

f* = (1 × 0.55 − 0.45) / 1 = 0.10 → 10% of bankroll per trade.

Ten percent per trade is insane for trading, and every serious practitioner knows why: the Kelly formula assumes known win rate and odds. Real trading edges are estimated, often over-estimated, and drift with market regime. Full-Kelly sizing against an over-estimated edge loses money faster than a random strategy.

Standard practice is fractional Kelly – usually 1/4 to 1/2 of full Kelly. On our 55/45 example, quarter-Kelly is 2.5% per trade, and half-Kelly is 5% per trade. Those are still aggressive numbers for a funded account. A reasonable rule of thumb: size at fractional Kelly based on a conservative estimate of your edge, capped by the 1% / 0.5% rule above.

For most evaluation-phase traders, the Kelly output is a theoretical ceiling; the funded-account drawdown rule is the practical one. Use whichever number is smaller.

Risk of ruin (RoR) is the probability that your account hits some catastrophic drawdown given your edge, risk per trade, and number of trades. The classic closed-form approximation (Vince, 1992) is:

RoR ≈ ((1 − edge) / (1 + edge)) ^ (capital / risk_per_trade)

where edge is expectancy as a fraction (e.g., 0.10 for the 55/45 example) and capital is the number of risk-units until ruin. On a funded account with 8% max drawdown and 0.5% risk per trade, capital = 16 units:

RoR ≈ (0.90/1.10) ^ 16 ≈ 3.3%

A 3.3% probability of hitting the 8% drawdown sounds low, but it’s over a long horizon. On shorter horizons – an evaluation window – the probability is higher because variance dominates expected value. Run the same formula with capital = 8 (4% daily loss / 0.5% risk):

RoR_day ≈ (0.90/1.10) ^ 8 ≈ 18%

You have a roughly 1-in-5 chance of hitting the daily loss limit in a long losing run given that exact edge and risk-per-trade. Cutting risk-per-trade to 0.25% doubles capital to 16 daily risk-units, pushing RoR_day back down to about 3%. This is the concrete payoff to conservative sizing: not less money on average, just dramatically fewer catastrophic outcomes.

Putting it all together – the framework a disciplined funded trader actually uses:

1. Fix risk per trade in R and in %. 0.5% is the default; 0.25% if your variance is high.
2. Compute position size before placing the trade. Use a calculator, not mental math. Mental math is where 0.5% turns into 0.8%.
3. Cap daily loss at 2R (not the firm’s limit). If the daily limit is 4%, stop trading at –2%. The firm’s limit is a hard stop; yours should be softer, to leave recovery room.
4. Cap correlated exposure. Treat two correlated positions as one. 0.5% risk on EUR/USD and 0.5% risk on GBP/USD is 0.8–0.9% effective.
5. Plan for news windows. Either close positions before major releases or halve position size.
6. Journal every trade in R. At the end of each week, compute win rate, average win R, average loss R, and expectancy. Adjust size only if the 30-trade rolling expectancy changes materially.
7. Scale size down after a losing week. Drop to half-size for 10 trades after any week that’s –3R or worse. Scale back up only after you’re back to break-even.

None of this is exciting. It’s supposed to be boring – boring is the point. The traders who stay funded for years across multiple firms are the ones who implement five of these rules rigorously and ignore the rest.