The Sharpe ratio measures return per unit of risk. Here are the five levers that move it, tested before and after on a sample portfolio, and the shortcuts that did not work.
What is your own portfolio exposed to? Check it with Exent →
Only two things: the return above cash, and the volatility it took to earn it. The Sharpe ratio is (portfolio return − risk-free return) ÷ volatility. The figures on this page calculate it from monthly returns, use 1–3 month Treasury bills (BIL) as the risk-free rate and annualize the result. Unless another source is named, they come from Exent’s data as of Sep 30, 2026.
That gives three routes. Raise return without adding as much volatility. Cut volatility without losing as much return. Or stop paying for things that lower return and add nothing, such as fees and duplicate holdings. Anything that raises return and volatility in the same proportion leaves the ratio where it was. If you care more about losses than swings, the Sortino ratio uses only downside volatility.
Sample portfolio: 40% SPY, 25% QQQ, 20% Apple, 15% Nvidia. Hypothetical backtest over calendar years 2017–2025, rebalanced every January, computed from yearly returns in the same data, before costs and taxes. Nine yearly returns make this a rough Sharpe estimate; read the direction, not the decimals.
| Version (backtest) | Annual return | Volatility of yearly returns | Sharpe | Worst year |
|---|---|---|---|---|
| Before: sample as is | 28.2% | 31.8% | 0.96 | −28.3% |
| Merge QQQ into SPY (remove duplicate) | 26.9% | 29.4% | 0.97 | −24.7% |
| Add 10% gold (GLD) | 27.3% | 29.2% | 0.99 | −25.5% |
| Cap Nvidia at 7.5%, rest to SPY | 24.1% | 26.8% | 0.94 | −25.8% |
| Add 10% T-bills (BIL) | 26.0% | 28.7% | 0.96 | −25.3% |
| Add 10% bonds (AGG) | 25.8% | 29.2% | 0.94 | −26.7% |
| Add 10% long Treasuries (TLT) | 25.5% | 29.7% | 0.92 | −28.5% |
| S&P 500 alone (SPY) | 15.0% | 16.6% | 0.85 | −18.2% |
Two lessons. Every lever lowered the worst year, which may matter more to you than the ratio. And the Sharpe ratio barely moved, because the sample was dominated by one stock with an exceptional run. Improving risk-adjusted return is usually a matter of hundredths, not a transformation.
Duplicate funds and stock-specific bets add volatility without a reliable reward. Of the 465 largest US stocks with five years of data, only 101 had a higher 5-year Sharpe ratio than SPY to Sep 30, 2026 (today’s largest companies, so survivorship flatters them). Merging QQQ into SPY in the sample nudged Sharpe from 0.96 to 0.97 and cut the worst year by 3.6 points. The step-by-step version is in how to reduce risk without lowering returns.
The holding with the largest share of risk relative to its weight is the usual first trim. In the sample, Nvidia was 15% of the money and an estimated 30% of the volatility (see which holding adds the most risk).
In hindsight the cap cost a little Sharpe (0.96 to 0.94), because Nvidia’s own 5-year Sharpe was 1.16 versus 0.69 for SPY. That is the honest catch: capping is a decision about risk you can see today, made without knowing which stock will run. Judge it by the worst year it removed, not by the winner it trimmed.
This is the lever that most often raises the ratio, because a low-correlation asset can cut volatility by more than it cuts return. Gold did that here: Sharpe 0.96 to 0.99. Over the same five years, a 5% slice of gold (GLD) lowered an S&P 500 portfolio’s volatility from 15.7% to 15.1%.
Bonds did the opposite in this window. Long Treasuries (TLT) returned −31.2% in 2022, the same year stocks fell, and their 10-year Sharpe was −0.32. In a backtest from July 2016, a monthly rebalanced 60/40 mix had a Sharpe of 0.75 against 0.88 for the S&P 500. Whether bonds help depends on the inflation and rate backdrop; the do bonds improve Sharpe guide goes deeper.
Costs are the one lever that only works in your favor. A fee or tax drag lowers return one-for-one and leaves volatility unchanged, so every point saved raises Sharpe. Check expense ratios, trading costs and the tax cost of turnover; see how to check fund fees.
Rebalancing keeps the risk you chose from drifting upward as winners grow. How often matters less than people think: in the same backtest a 60/40 mix rebalanced monthly had a Sharpe of 0.75 and once a year 0.76. The benefit is control of risk, not extra return.
| Idea | What happened | Source |
|---|---|---|
| Hold more cash | Return and volatility fell together; Sharpe stayed 0.96 | Sample backtest above |
| Use leverage | Scales return and volatility up together; borrowing costs then lower the ratio | Definition of Sharpe |
| Max-Sharpe optimizer (36-month lookback) | Sharpe 0.83, same as SPY; max drawdown 30.5% vs 23.9% | Backtest, Jul 2019–Sep 2026 |
| Minimum-variance optimizer | Volatility 5.9%, but annual return 0.7% and Sharpe −0.27 | Backtest, Jul 2019–Sep 2026 |
| Buy the past Sharpe leader | Leaders change; ranking on history rewards luck | See backtest overfitting |
Optimizers chase the inputs they are fed, and past returns are noisy inputs. See the max-Sharpe backtest and backtest overfitting.
More methods are collected under portfolio optimization.
Compare it with a benchmark over the same period rather than a fixed number. In Exent’s data the S&P 500 (SPY) had a Sharpe ratio of 0.69 over the five years and 0.86 over the ten years to September 30, 2026.
Historically it often did, but not in the decade to September 2026. Exent’s 60/40 backtest had a Sharpe of 0.75 versus 0.88 for the S&P 500, largely because bonds fell sharply in 2022.
No. Cash earns the risk-free rate, so adding it lowers excess return and volatility in the same proportion and leaves the ratio about unchanged.
Yes, when the added asset has low correlation with the rest and a positive return above cash. In a sample backtest, adding 10% gold raised the Sharpe ratio from 0.96 to 0.99.
Not on its own. A high ratio from a short or lucky period says little about the future, and it hides how deep the worst loss was, so check drawdown alongside it.
Upload a screenshot of your holdings and get concentration, overlap, crash exposure and a risk-matched benchmark in about a minute.
Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.