Hit Rate (Accuracy)
Key Points
- The forecast value is compared with the previous realised value to set the predicted direction; if the realised value moves the same way the forecast counts as a hit.
- 50% is coin-toss level; a meaningful model must stay above that with a sufficient sample (at least 30–60 forecasts).
- a high hit rate on few samples can be luck.
Formula / Calculation
Hit rate = Number of hits / Total forecasts × 100. Days with zero realised movement are excluded. Mean absolute percentage error (MAPE) = Mean(|Actual − Forecast| / Actual) × 100 measures magnitude error regardless of direction.
Interpretation
50% is coin-toss level; a meaningful model must stay above that with a sufficient sample (at least 30–60 forecasts). Yatırımcı.AI reports its forecasts with error measures instead: the AI track record shows stock price forecasts by mean absolute percentage error, and the live fund estimate accuracy page shows fund estimates by mean absolute error in percentage points.
Pitfalls
a high hit rate on few samples can be luck. Directional accuracy says nothing about the size of moves; small correct moves may not offset large incorrect ones. Past performance is no guarantee of future results and forecasts are not investment advice.
Pages Where This Term Appears
Related Terms
Frequently Asked Questions
What is Hit rate (accuracy)?
The hit rate is the share of predictions in which the direction a forecasting model predicted (up or down) matched the realised direction.
How is Hit rate (accuracy) calculated?
Hit rate = Number of hits / Total forecasts × 100. Days with zero realised movement are excluded. Mean absolute percentage error (MAPE) = Mean(|Actual − Forecast| / Actual) × 100 measures magnitude error regardless of direction.
Prepared by: Yatırımcı.AI Research TeamLast reviewed: Method: MethodologyEditorial policy