Position Sizing with Volatility Forecasts

Ask 100 traders what makes a profitable strategy and 90 will talk about entries: indicators, patterns, signals, setups. Ask the other 10 -- the ones who are actually profitable over years -- and they will talk about position sizing. The math is unambiguous. A strategy with a 55% win rate and 1.2:1 reward-to-risk ratio is profitable. But that same strategy can produce wildly different outcomes depending on how much you risk per trade. Risk 1% of your account per trade and you have a smooth, compounding equity curve that survives multi-week drawdowns. Risk 10% per trade and you have a coin-flip between fast profits and account destruction. The entry does not change. The win rate does not change. The reward-to-risk does not change. Only the sizing changes, and it determines whether the strategy makes you money or wipes you out. This is not opinion -- it is the mathematical relationship between bet size, edge, and risk of ruin.

The Kelly criterion, developed by John Kelly at Bell Labs in 1956, answers the question: given a known edge, what is the optimal fraction of your bankroll to risk on each bet? The formula is: Kelly % = (win_rate x avg_win - loss_rate x avg_loss) / avg_win For a trader with a 55% win rate, average winner of $40 (8 points on MES), and average loser of $30 (6 points on MES): Kelly % = (0.55 x 40 - 0.45 x 30) / 40 = (22 - 13.5) / 40 = 21.25% Full Kelly says risk 21.25% of your account per trade. In practice, nobody does this because the assumptions are never perfectly known and the drawdowns at full Kelly are psychologically brutal (50%+ peak-to-trough drawdowns are normal). Most professional traders use quarter-Kelly to half-Kelly: 5-10% of the full Kelly fraction. For our example trader, quarter-Kelly is approximately 5% risk per trade. On a $50,000 prop firm account with a $2,500 trailing drawdown, 5% of the drawdown buffer is $125 -- which translates to a 25-point stop on MES. That is a reasonable, tradeable number for most intraday strategies.

Fixed position sizing ignores the most important variable: how much the market is expected to move today. A 10-point stop on a day when the expected range is 12 points is a tight stop. The same 10-point stop on a day when the expected range is 45 points is an extremely tight stop that will get hit by noise. Volatility-adjusted sizing solves this. The principle: risk a fixed dollar amount per trade, but adjust your stop width (and therefore your position size) based on the expected range. Here is a practical framework using the 1-10 volatility rating: Rating 1-3 (Quiet): Expected ES range roughly 15-25 points. Stop width: 6-10 points. At $5/point on MES, dollar risk per trade: $30-50. You can afford 2-3 MES contracts within a $125 risk budget. Rating 4-6 (Normal): Expected ES range roughly 25-40 points. Stop width: 10-16 points. Dollar risk per trade: $50-80. Standard 1-2 MES contracts. Rating 7-8 (Elevated): Expected ES range roughly 40-60 points. Stop width: 16-24 points. Dollar risk per trade: $80-120. Reduce to 1 MES contract, or sit out if your strategy does not work in trending conditions. Rating 9-10 (Extreme): Expected ES range 60+ points. Stop width: 24+ points. Many experienced traders sit out entirely. If trading, 1 MES maximum with a wide stop and acceptance that the trade may produce an outsized loss. The dollar risk stays approximately constant. What changes is the stop width and contract count. This keeps your risk of ruin stable across different market conditions.

A stop loss should be placed where your trade thesis is invalidated, not at an arbitrary point distance. But that invalidation point changes with market conditions. A support level 8 points away on a quiet day is meaningful. The same level 8 points away on a CPI day might get swept in the first minute of the release. Adaptive stops use the expected range to scale the stop to conditions. A simple method: set your stop at 30-40% of the expected daily range from your entry. On a day rated 3 (expected range around 18 points), that is a 5-7 point stop. On a day rated 7 (expected range around 45 points), that is a 14-18 point stop. The wider stop on high-volatility days is not riskier if you reduce your position size proportionally. One MES contract with an 18-point stop risks $90. Two MES contracts with a 7-point stop risks $70. The dollar risk is similar, but the wider stop gives the trade room to work in volatile conditions instead of getting stopped out by normal noise. This is the core insight of volatility-adjusted trading: you are not trying to avoid losses. You are trying to keep each loss approximately the same dollar size regardless of conditions. Consistent loss size is what makes the Kelly math work over hundreds of trades.

A complete position sizing framework for futures trading combines four elements: 1. Maximum risk per trade: Define this as a percentage of your drawdown buffer (for prop firms) or account equity (for personal accounts). A range of 2-5% is standard. For a $2,500 trailing drawdown, that is $50-125 per trade. 2. Volatility-adjusted stop width: Use the expected range (from the volatility rating or your own ATR measurement) to set a stop that gives the trade room to work. Typically 30-40% of the expected daily range. 3. Position size calculation: Divide your maximum risk by the stop width in dollar terms. If your risk budget is $100 and your stop is 15 points on MES ($75), you can trade 1 contract. If the stop is 8 points ($40), you can trade 2 contracts. 4. Daily loss limit: Cap total session losses at 2-3 trades or a fixed dollar amount (e.g., 6-8% of drawdown buffer). If you hit the limit, stop trading for the day. This prevents the revenge-sizing spiral that destroys accounts. This framework is not exciting. It will not double your account in a week. What it will do is keep you in the game long enough for your edge to compound. And compounding -- not any single trade -- is how money is actually made in futures trading. The volatility rating makes step 2 automatic. Instead of calculating ATR or estimating the range yourself, you check a single number that synthesizes 42 features into an expected range. Whether you use our rating or your own measurement, the principle is the same: size for the day you are trading, not the day you wish you were trading.