Three Numbers, Three Different Worlds
Ask three investors how they evaluate an asset and you may get three different answers. One looks at total return — the number that compounds wealth. Another looks at the Sharpe ratio — return adjusted for volatility. The third looks at the Sortino ratio — return adjusted for downside volatility specifically.
Each is right, in its own context. Each is misleading when applied outside that context. The art is knowing which metric to look at now, given your time horizon, risk tolerance, and what question you are actually trying to answer.
This post breaks down the three metrics, the math behind them, the situations where each shines, and which one you probably want to optimize for given your investor profile.
Table of Contents
- The three metrics defined
- The math, lightly
- When Sharpe is right
- When Sortino is right
- When raw return is right
- A worked comparison across three tickers
- Matching the metric to the investor
- How Dipsern handles all three
The Three Metrics Defined
Raw return is the simplest: the percentage change in price over a period. If you bought at $100 and sold at $115 ninety days later, the return is +15%. No adjustment for risk, no normalization. Just the number.
Sharpe ratio divides excess return (return above a risk-free rate) by the total standard deviation of returns. It answers the question: "How much return am I getting per unit of total volatility, including the up moves?"
Sortino ratio divides excess return by the downside deviation only — the standard deviation of returns that fell below the risk-free rate. It answers: "How much return am I getting per unit of bad volatility, ignoring the good kind?"
The intuition: Sharpe treats all volatility as bad. Sortino acknowledges that an asset jumping +20% is qualitatively different from one falling -20%, and only penalizes the second.
The Math, Lightly
For a series of returns r_t, with a risk-free rate rf:
Sharpe = (mean(r_t) - rf) / stdev(r_t)
Sortino = (mean(r_t) - rf) / stdev_downside(r_t)
where stdev_downside is the standard deviation computed only on returns less than rf.
Dipsern uses an annualized version. With daily returns, both ratios are typically multiplied by sqrt(252) (the number of trading days in a year) to put them on an annual scale. The default risk-free rate in Dipsern's engine is 7.5% annual — a conservative choice that makes the ratios harder to clear and avoids flattering low-yield environments.
A Sharpe above 1.0 is generally considered acceptable; above 2.0 is excellent; above 3.0 is rare and typically a sign of either genuine alpha or data mining. The same rough scale applies to Sortino, though Sortino tends to read higher than Sharpe for the same asset because it ignores upside variance.
When Sharpe Is Right
Sharpe is the right metric when you care about total volatility — that is, when large positive surprises are also unwelcome.
This sounds strange in retail context (who minds an unexpected windfall?), but it matters in two real situations.
Tactical position sizing. If you size positions based on volatility (a common professional approach — see "vol targeting" portfolios), you want your sizing model to anticipate all future jumps, not just downside ones. An asset that surges +30% unexpectedly is exhibiting variance, and that variance affects how much you should allocate.
Cash-equivalent and bond-like assets. For an asset whose role in the portfolio is stability — a short-duration bond ETF, a Treasury proxy — total variance is the relevant risk, because the whole point of the position is not to move much. Sharpe captures that correctly. Sortino, by ignoring upside variance, would flatter a bond that occasionally spikes.
When you're benchmarking against a peer group. Sharpe is the lingua franca of the institutional world. If you're comparing your portfolio to a hedge-fund index or a manager's track record, Sharpe is the metric on the page.
When Sortino Is Right
Sortino is the right metric when only downside volatility is risk to you — when upside surprises are pure benefit and only downside surprises are pain.
For most individual investors, this is the more honest framing. If a stock you own jumps 30% unexpectedly, you do not complain about the variance. You just have more money. The only "bad" outcome is loss.
Sortino is most appropriate when:
You're building a long-term, growth-oriented portfolio. You're holding for years; you do not care if your portfolio occasionally jumps. You care about avoiding deep drawdowns. Sortino prices that priority directly.
You're comparing volatile assets to less-volatile ones. Crypto and small-caps often look terrible on Sharpe because their total volatility is huge. But much of that volatility is upside. Sortino strips that out and gives you a fairer comparison.
You're risk-averse about losses, not about gains. Behavioral economics calls this loss aversion. Most people have it. Sortino is the metric that respects that asymmetry mathematically.
When Raw Return Is Right
Sometimes the risk-adjusted ratios are the wrong question. Raw return is the right metric in three situations.
Very long horizons. Over 20-30 year holding periods, the variance of annual returns averages out. What you ultimately care about is the compound growth rate, which is driven almost entirely by raw return. A portfolio with Sharpe 0.8 and 9% annualized return beats a portfolio with Sharpe 1.2 and 6% annualized return — over 30 years, by a lot.
Small position sizes relative to net worth. If a position is 1% of your portfolio, you do not really care about its volatility. You care about its expected return. Use raw return for the screen; use Sharpe/Sortino for the sizing.
When you genuinely do not need the money for the period. If you have a fixed amount you can lose entirely without changing your life, the volatility-adjusted ratios are a luxury concern. Raw expected return is what's left.
The mistake is using risk-adjusted ratios in contexts where they do not apply, then ending up with a portfolio that "looks safe" on paper but compounds at 4% per year — when raw return on a different mix would have delivered 9% with drawdowns the investor could have stomached anyway.
A Worked Comparison Across Three Tickers
The table below shows stylized values for three hypothetical assets over the same period. The numbers are illustrative; they make the point about how the three metrics disagree.
| Asset | Annualized return | Sharpe | Sortino | Max drawdown | |---|---|---|---|---| | Large-cap ETF | +9% | 0.85 | 1.15 | -22% | | Growth tech stock | +18% | 0.72 | 1.40 | -45% | | Bond ETF | +4% | 1.30 | 1.40 | -8% |
By raw return, the growth stock wins handily. By Sharpe, the bond wins — its lower return is more than compensated for by its much lower total volatility. By Sortino, the growth stock leads narrowly because most of its volatility was on the way up, and Sortino does not penalize that.
There is no "right" answer in the abstract. There is only a right answer given the investor.
- A retiree drawing income for the next ten years probably picks the bond — Sharpe is the relevant metric, and the bond's low max drawdown matters more than the equity premium.
- A 30-year-old saving for a 30-year horizon probably picks the growth stock — raw return matters more, and the volatility will average out.
- A balanced 40-something probably picks the large-cap ETF — moderate return, manageable drawdown, defensible on Sortino.
The metric you optimize for defines the portfolio you end up with. Choose deliberately.
Matching the Metric to the Investor
A rough mapping, useful as a starting point:
| Investor profile | Primary metric | Secondary metric | |---|---|---| | Conservative, capital preservation | Sharpe | Max drawdown | | Income-focused, near retirement | Sharpe | Sortino | | Balanced, mid-career | Sortino | Raw return | | Growth-oriented, long horizon | Sortino | Raw return | | Aggressive, very long horizon | Raw return | Sortino | | Speculative, small allocation | Raw return | — |
This mapping is not law. An aggressive investor with sizable positions in volatile assets might still want to monitor Sortino just to make sure no single position is producing the bulk of the downside. A conservative investor might still check raw return to make sure their "safe" portfolio is actually keeping up with inflation.
The mapping is a starting point. The discipline is to pick a primary metric before you build the portfolio, then evaluate every position against it consistently.
How Dipsern Handles All Three
Dipsern's engine exposes all three metrics as switchable options. When you run an analysis, you choose whether the forward-looking signal is computed in units of:
- Return — simple forward percentage change (default).
- Sharpe — annualized excess return divided by total volatility of the window.
- Sortino — annualized excess return divided by downside-only volatility.
The same drawdown segmentation and rolling-median machinery applies under each metric. What changes is what you are forecasting: the typical return from this drawdown bucket, or the typical Sharpe or Sortino of holding from this drawdown bucket.
That matters because the optimal entry can look different under different metrics. An asset might have a high median forward return at -30% drawdown but a mediocre median forward Sharpe because the path from here is volatile. A conservative investor would pass; an aggressive one would take it.
The switch is one click. Use it.
Key Takeaways
- Sharpe penalizes all volatility, including upside. Use it for stability-oriented assets and institutional benchmarking.
- Sortino penalizes only downside volatility. Use it for growth-oriented portfolios and loss-averse investors.
- Raw return is the right metric for long horizons, small positions, and capital you can truly afford to lose.
- Choose the metric before you build the portfolio, then evaluate every position consistently.
- Conservative investors should lean toward Sharpe; aggressive ones toward raw return; most others toward Sortino.
- Dipsern lets you switch among all three with a single toggle — try the same asset under each metric and notice how the picture changes.
Try It Yourself
Pick an asset you know well — something like QQQ or a volatile name like TSLA — and run the analysis under all three metrics. Compare what the signal says about the same drawdown level when measured by return, Sharpe, and Sortino. The disagreement (or agreement) is often the most useful thing the analysis tells you.
Educational content. Past performance does not guarantee future results. This is not financial advice.