Position sizing in trading: the math behind risk per trade

Most traders pick a contract count by feel. They take one MNQ when they feel cautious, three when they feel confident, and five after a winner. That is not position sizing. That is mood. Real position sizing answers one question with math: given my account, my edge, and my stop, how many contracts should I trade?

This post walks through the three sizing methods that matter (fixed dollar, fixed percent, Kelly), shows the arithmetic on each, and explains why prop traders almost always use a fraction of what the math suggests.

What position sizing actually means

Position sizing is the decision of how much money you put at risk on a single trade. The risk is not the notional value of the contract. It is the distance from your entry to your stop loss, multiplied by the contract's tick value, multiplied by the number of contracts.

For one MNQ contract with a 10-point stop:

Risk = 10 points × $2 per point × 1 contract = $20

For one ES contract with the same 10-point stop:

Risk = 10 points × $50 per point × 1 contract = $500

Same chart, same setup, 25 times more risk. This is why "I take one contract" is a meaningless statement without context. The contract you trade and the stop you use define your risk far more than the share count does. Position sizing is the bridge between the trade idea and the dollar amount on the line.

Fixed dollar risk

The simplest method. Pick a dollar amount you are willing to lose per trade and back into the contract count.

Say you decide that no trade should ever cost more than $200. You see a setup on MNQ with a 15-point stop:

Max contracts = $200 / (15 points × $2) = $200 / $30 = 6.67

You round down to six contracts. Total risk: $180. If the stop is wider, say 25 points, you size down:

Max contracts = $200 / (25 points × $2) = $200 / $50 = 4

Fixed dollar sizing is honest. It forces you to size the stop, not the conviction. The downside is that it does not scale with the account. A $200 risk is reasonable on a $20k account and trivial on a $200k account. As the account grows or shrinks, you have to manually re-anchor.

Fixed percent risk

The default for most professional traders. You risk a fixed percentage of account equity per trade. One percent is the textbook number. Two percent is the upper end for traders with a real edge.

On a $50,000 account at 1% risk:

Risk per trade = $50,000 × 0.01 = $500

With a 20-point MNQ stop:

Contracts = $500 / (20 × $2) = 12.5 → 12 contracts

The reason percent sizing is the default is compounding. After a winning streak the account grows, your dollar risk grows with it, and gains compound. After a losing streak the account shrinks, your dollar risk shrinks too, and the drawdown decelerates.

The math of drawdown matters here. After ten consecutive losses, account equity is:

1% risk:   0.99^10 = 0.9044  → 9.56% drawdown
2% risk:   0.98^10 = 0.8171  → 18.29% drawdown
5% risk:   0.95^10 = 0.5987  → 40.13% drawdown

Five percent per trade sounds aggressive. Ten losses at five percent per trade halves your account. That is not a hypothetical. At a 55% win rate, the probability of a ten-loss streak somewhere in a 1000-trade sample is meaningful, not zero. Position sizing has to survive variance, not just average outcomes.

The Kelly criterion

Kelly is the answer to "what bet size maximizes long-run growth, given a known edge?" The formula in trading form:

f* = W − (1 − W) / R

Where W is win rate (as a decimal) and R is the ratio of average win to average loss. The result f* is the fraction of bankroll to risk per trade.

Example: 55% win rate, average winner is 1.5R (the average winner is 1.5 times the average loser).

f* = 0.55 − (0.45 / 1.5)
   = 0.55 − 0.30
   = 0.25

Full Kelly says risk 25% of the account per trade. On $50,000 that is $12,500 per trade. Nobody trades full Kelly. Here is why.

Kelly assumes your win rate and your win/loss ratio are known with certainty. They are not. Your sample is small, the market regime can shift, and a 55% win rate measured over 100 trades has a wide confidence interval around it. Full Kelly on an overestimated edge produces ruin, not optimal growth.

Full Kelly also produces enormous drawdowns even when the edge is real. With a 25% risk per trade, three losses in a row drops the account by:

1 − 0.75^3 = 1 − 0.4219 = 57.81%

Half your account, gone after three losses. The math says recovery is still positive expectation. Your nervous system disagrees, and your prop firm closes the account before recovery happens.

Why most traders use fractional Kelly

The fix is fractional Kelly: use a small multiple of the Kelly number (one-quarter is common, sometimes one-half).

Continuing the example above with quarter Kelly:

0.25 × 0.25 = 0.0625 → 6.25% per trade

Still aggressive for most accounts. Even quarter Kelly assumes you have correctly measured your edge. Most traders have not. They have backtested 50 trades, found a 60% win rate, and have no idea whether the next 50 will hit 60% or 40%.

This is why one to two percent per trade is the practical answer for most traders. It sits well below quarter Kelly for any realistic edge, which means you can be wrong about your win rate by a lot and still not blow up. It also produces drawdowns small enough that the trader can keep executing without flinching.

The sequence is: estimate your edge from your journal, calculate Kelly, take a quarter or less, and adjust down further if your sample is small or your win rate is volatile across regimes. For a deeper look at why win rate alone does not tell you what to risk, see the win rate vs risk-reward post.

Position sizing on a prop firm account

Prop accounts add a hard constraint that retail accounts do not have: a maximum drawdown that ends the account if breached. On a $50,000 Topstep account with a $2,000 maximum loss limit, the account is dead at $48,000. Percent-of-equity sizing on the initial balance is misleading because the real bankroll for sizing purposes is the distance to the drawdown line, not the total balance.

If you want to survive ten consecutive losses on a $50k Topstep account with $2,000 of room before the limit, your max risk per trade is bounded:

Max risk = $2,000 / 10 = $200 per trade

That is 0.4% of the account but 10% of the actual cushion. As the account grows past the trailing point, the cushion grows and risk per trade can grow with it. As it shrinks toward the limit, risk per trade has to shrink faster than percent-of-equity would suggest, because the drawdown line is fixed.

This is why prop traders who size purely off account equity blow up in evaluation more often than they should. They are sizing for the wrong number. The relevant number is the cushion, not the balance.

What to actually do

Pick a percent of equity (1% if you are still building confidence in your edge, 2% if your journal shows a stable win rate and reward-to-risk ratio across at least 100 trades). Translate it to a dollar amount. Translate that dollar amount to a contract count using the trade's stop distance. Log every trade with the dollar risk and the R-multiple result so the next sizing decision is informed by data, not memory.

The traders who survive variance are not the ones with the highest win rates. They are the ones whose position sizes are small enough that variance cannot end the account before the edge plays out. Position sizing is not a pre-trade decision. It is the decision that lets every other decision matter.

You can track risk per trade, R-multiples, and drawdown progress automatically in Tradavity so the math runs in the background while you focus on the chart.