Is a higher win rate better?

Not necessarily. A strategy can win 90% of the time and still lose money if the rare losses are large, and win 40% of the time and compound if the winners run far larger than the losers. Be more suspicious of very high advertised win rates, not less.

How accurate are quantitative stock signals? Why hit rate misleads

Q: How accurate are quantitative stock signals?

Accuracy (hit rate) is the wrong metric. Many durable systematic strategies are right only about 40-55% of the time and still profit, because the size of the average winner versus the average loser, the payoff, matters more than how often they win.

Q: What should I look at instead of accuracy?

Expectancy, the average outcome per signal across all of them, and profit factor (gross profit divided by gross loss). A profit factor above 1 means the winners more than pay for the losers, regardless of hit rate.

This article is educational and explains how to think about signal “accuracy.” It is not personalised investment advice and not a claim of any specific accuracy or profitability for any individual.

“How accurate are quantitative stock signals?” is the most common question and the least useful one, because accuracy, how often the signal is right, is not what makes a system profitable. Plenty of systematic strategies are right well under half the time and still compound; plenty of high-accuracy ones lose money. What matters is expectancy: accuracy multiplied by how much you win when right, against how much you lose when wrong.

Accuracy is not the same as edge

Hit rate, the share of signals that end up profitable, feels like the headline number, but on its own it tells you almost nothing. A strategy that wins 90% of the time by risking $10 to make $1 will be destroyed by the occasional loss. A trend-following strategy that wins 40% of the time but lets winners run several times the size of its losers compounds steadily. Very different accuracies, opposite outcomes, and the higher-accuracy one is the loser, because the payoff, not the hit rate, does the work.

The number that actually matters

The honest metric is expectancy (average profit per signal across the whole population) and its cousin profit factor (gross profit divided by gross loss). A profit factor above 1 means the winners more than pay for the losers, regardless of how often you win. This is exactly why a high-quality breakout system can be right less than half the time and still be strongly positive, the logic is laid out in how breakout setups work and shown trade-by-trade in the Suzaku engine.

So how “accurate” is a good system?

Often less accurate than you’d guess. Many durable systematic strategies sit somewhere around 40 to 55% winners; the great ones are defined by the size of the right tail, not the frequency of being right. A service advertising a 90%+ “win rate” is usually either cutting winners early and letting losers run (great accuracy, terrible expectancy) or simply not counting honestly. Be more suspicious of high advertised accuracy, not less.

Accuracy you can’t reproduce isn’t accuracy

Any accuracy figure is meaningless if it came from a curve-fit, survivorship-biased, or non-reproducible backtest. Before trusting a number, check how it was produced, the questions to ask are in how to vet a track record. Our own realised hit rate, payoff, and the full distribution are published, not asserted, on the track record.

Sources & further reading

Jegadeesh, N. & Titman, S. (1993). “Returns to Buying Winners and Selling Losers.” Journal of Finance, 48(1), 65 to 91., a profitable edge with a sub-50% hit rate.
Bailey, D. H. & López de Prado, M. (2014). “The Deflated Sharpe Ratio.” Journal of Portfolio Management, 40(5), 94 to 107., why a headline performance figure must be discounted for how it was found.

Shishin Research