What is a good hit rate for trading signals?
For short-horizon directional calls, roughly 53% to 58% is a genuinely good hit rate. Below about 52% there is usually no edge at all once costs are counted. Above about 60%, sustained over hundreds of independent positions and out-of-sample data, is rare enough that it should be treated as a measurement error until proven otherwise. Cashora measured 58.8% across 2,289 positions on 30 US stocks and 57.3% on tickers held back from all tuning.
Last measured 2026-09-22 · Thomas Solutions LLC
Why isn't a higher hit rate better by definition?
Because hit rate alone says nothing about size. A method that wins 70% of the time with tiny gains and occasional catastrophic losses is worse than one winning 55% with balanced outcomes. Hit rate has to be read next to average return per position, drawdown and Sharpe ratio. Cashora publishes all four per ticker on the measured performance page.
Why do advertised hit rates of 85% or 90% exist?
Almost always one of four reasons: the result was tuned and measured on the same data; only winning periods were shown; the "win" is defined loosely, such as any favourable move at any point rather than the close at a fixed horizon; or overlapping positions were compounded as though each were separately funded. Fixing that last error alone turned an apparent +19,812% in our own early testing into a realistic figure.
What does out of sample mean and why does it matter?
It means measured on data that played no part in choosing the rules. We used 15 tickers to select the filters and held 15 back entirely, measuring those once, afterwards. Those numbers are lower — 57.3% versus 58.8%, Sharpe 0.38 versus 0.59 — and they are the honest ones. Any tuning that improves in-sample results while hurting out-of-sample results is discarded, which is what happened to an earlier strategy re-weighting attempt.
How many trades before a hit rate means anything?
Dozens at minimum, ideally over a hundred, spread across different market regimes. A 60% hit rate over 20 positions is well within chance. Every per-ticker row we publish shows its position count so the reader can judge that themselves; thin samples are labelled in the app rather than presented as proven.
What hit rate does Cashora require before calling a signal actionable?
A signal is marked actionable only if it is a long call, the price is above its 200-day average, the confidence score is 60 or higher, and the ticker's own measured history in that direction supports it. Everything else is shown as information only with a grade of C or D. Direction is never reversed — an unreliable signal becomes a low-confidence signal, not an inverted one. The grading system is described on AI trading signals.
What about Sharpe ratio?
Single-ticker Sharpe measured 0.59 on average and 0.38 out of sample. Reaching the 1.0–2.0 range that professional portfolios target requires holding many names simultaneously rather than one position at a time; our tests of a diversified version measured 0.62–0.75 out of sample, and we do not claim it as a product feature because it is not built into the app yet.
All figures are hypothetical back-tested results excluding commissions, slippage, taxes and dividends. Past performance does not indicate future results. Cashora is a market probability signal predictor operated by Thomas Solutions LLC — not a broker, no trades are executed, and nothing here is investment advice.