Position Sizing and Core Risk Metrics

By Anh Le

From Backtest to Risk Budget

You just finished checking whether your backtest is real or curve-fit. Now you turn its signals into positions you could actually hold, then describe the result with a small set of risk numbers. This is where a backtest stops being a chart and starts becoming a strategy.

By the end of this lesson, you can:

• Explain why position size matters as much as signal quality.
• Size a position with a volatility target.
• Read maximum drawdown, Sharpe ratio, and exposure correctly.
• Apply simple caps so a sizing rule cannot quietly take over your account.

Position Size Is Your Real Risk Dial

Two traders can use the same signal and get completely different outcomes. One sizes each trade at 2% of capital, the other at 50%. The signal may be identical, but the second trader is running a very different strategy.

Position sizing is where risk control becomes concrete. A risk budget is the amount of risk you are willing to spend. For a beginner strategy, that might mean targeting about 10% annualized volatility and accepting a maximum drawdown around 20%. Those numbers are choices, not laws. The important part is choosing them before you trade and then sizing every position to fit inside them.

Think of position size as the bridge between a forecast and a dollar amount. The forecast says “go long.” The sizing rule says “go long this much, given how volatile this asset is and how much risk the portfolio can carry.”

Volatility Targeting: Size by Risk, Not by Conviction

A common beginner mistake is to put the same dollar amount into every trade. That treats a calm utility stock and a wild small-cap stock as equally risky. Volatility targeting fixes this by making the position smaller when the asset is more volatile.

The basic rule is:

$$ w_t = \text{signal}t \times \frac{\sigma{\text{target}}}{\sigma_{\text{asset},t-1}} $$

In plain English, the weight is _____, scaled by the signal. A 10% target and a 20% asset volatility give _____ before caps.

Here is the same rule in a few lines of Python-like pseudocode:

vol = returns.rolling(60).std() * np.sqrt(252)
weight = signal * target_vol / vol.shift(1)
weight = weight.clip(-max_weight, max_weight)

Notice the .shift(1). The volatility estimate must be lagged so you are not using today’s return to size today’s position. That would be look-ahead bias, the same problem you worked to remove from your backtest.

A few practical rules:

• Use annualized volatility on both sides of the ratio.
• For a long-only strategy, the signal is usually 0 or 1. For a long-short strategy, it can be -1, 0, or +1.
• Cap the final weight. If volatility is 5% and your target is 10%, the raw weight is 200%. That is leverage, and it may not be what you intended.
• Rebalance on a schedule, such as weekly or monthly. Daily rebalancing can create costs that eat the benefit.

Warning: A volatility estimate can be wrong, especially during a crisis. A cap on position weight and gross exposure is not optional.

Try it on your capstone data: Your asset has 25% annualized volatility, your target is 10%, and your signal is 1.0. What raw weight do you get? What if the signal falls to 0.5?Answer guidance. With 25% asset volatility and a 10% target, 10/25 = 0.40, so the raw weight is 40%. With a signal of 0.5, it is 0.5 × 0.40 = 20%. If your cap is 30%, the first case is trimmed to 30%. Note which control is binding, the volatility target or the cap. That tells you what is actually limiting your risk.

Measuring Risk With Drawdown, Sharpe, and Exposure

Position sizing controls how much you put at risk. The next step is to describe what happened in the backtest in a way you can compare across strategies. Three metrics do most of that work.

Drawdown: the loss you would have lived through

Drawdown measures how far equity has fallen from its prior peak. Maximum drawdown is the _____ in the equity curve.

$$ \text{Max Drawdown} = \max_t \left( \frac{\text{Peak}_t - \text{Equity}_t}{\text{Peak}_t} \right) $$

A 25% drawdown needs a 33% gain just to get back to the old peak. That asymmetry is why drawdown matters more than many beginners expect. A strategy with a good average return but a 60% drawdown is hard to stick with in real life.

Sharpe ratio: return per unit of risk

The Sharpe ratio divides _____ by annualized volatility.

$$ \text{Sharpe} = \frac{R_{\text{annual}} - R_f}{\sigma_{\text{annual}}} $$

A Sharpe of 1.0 means one unit of excess return for each unit of volatility. Higher is better, but be careful. A very high backtest Sharpe on daily data should make you suspicious, not proud. Check for look-ahead bias, survivorship bias, and unrealistic fills before you celebrate.

Exposure: how much capital is actually working

Exposure tells you how invested the strategy is. Gross exposure sums _____. Net exposure sums signed weights, so long minus short.

If your long-only strategy has average gross exposure near 100%, it is roughly fully invested. If it is near 150%, you are using leverage, whether you meant to or not.

Note: Report all three metrics together. A strategy with a Sharpe of 1.5 and 300% gross exposure is a very different object from one with the same Sharpe and 80% exposure.

Try it on your capstone equity curve: Start at 100, rise to 120, fall to 90, then recover to 110. What is the maximum drawdown?

Click to reveal the answer

The peak is 120 and the trough is 90. The drawdown is (120 - 90) / 120 = 25%. The later recovery to 110 does not change the maximum drawdown, because the worst peak-to-trough loss already happened.

Second check: Your strategy earns 12% annualized, volatility is 15%, and the risk-free rate is 3%. What is the Sharpe ratio?

Click to reveal the answer

Sharpe = (12% - 3%) / 15% = 0.6. You can report this alongside maximum drawdown and average gross exposure in your capstone summary.

Guardrails for Sizing and Risk Reporting

• Lag your volatility estimate. Using today's return to size today's position is look-ahead bias in disguise.
• Always cap the final weight. A low-volatility asset can produce a raw weight above 100% very quickly.
• Do not report return alone. Pair it with maximum drawdown, Sharpe, and exposure so the reader can judge the path, not just the endpoint.
• Rebalance on a schedule you can defend. Daily rebalancing may keep weights close to target, but costs can erase the benefit.
• Treat a surprisingly high Sharpe as a warning sign first, and a discovery second.

What to Carry Into Your Capstone

You now have the sizing rule and the reporting vocabulary:

• Convert a signal into a weight with a lagged volatility target, then apply a cap.
• Measure the path with maximum drawdown, the risk-adjusted return with Sharpe, and the capital at work with exposure.
• Keep the sizing logic simple enough that you can explain it in one paragraph.

In the next lesson, you will put these pieces together in the capstone: one complete strategy, from data and rules to backtest, sizing, risk metrics, and a short written summary.

Front: What does a volatility target do? Back: Scales position weight so expected portfolio volatility stays near a chosen target.

Front: What is the volatility-target weight formula? Back: Signal times target volatility divided by lagged asset volatility.

Front: Why must the volatility estimate be lagged? Back: To avoid using today's return to size today's position, which is look-ahead bias.

Front: Asset volatility is 5% and target is 10%. What is the raw weight? Back: 200%, which is leverage and usually needs a cap.

Front: Why cap position weight? Back: Volatility estimates can be wrong, and uncapped weights can create unintended leverage.

Front: What is drawdown? Back: The decline from a prior equity peak to a later trough.

Front: What is maximum drawdown? Back: The largest peak-to-trough decline in the equity curve.

Front: A 25% drawdown needs what gain to recover? Back: About 33%, because the starting capital is now smaller.

Front: What is the Sharpe ratio formula? Back: Annualized return minus risk-free rate, divided by annualized volatility.

Front: What does the Sharpe ratio measure? Back: Excess return per unit of volatility.

Front: What is gross exposure? Back: The sum of absolute position weights, including shorts.

Front: What is net exposure? Back: The sum of signed weights, long minus short.

Front: What is the tradeoff in rebalancing frequency? Back: More frequent rebalancing keeps weights near target but raises trading costs.