How we calculate hit rates

TargetWatch compares recorded price targets with later share-price data. This page explains what each measure includes and where the data has limits.

Hit rate, direction accuracy and median error

A target is counted as hit if a daily closing price reaches it during the 365 days after publication. For a target above the reference price, a close at or above the target counts; for a target below the reference price, a close at or below it counts. The reference price is the latest available daily close on or before the target date.

Direction accuracy is the share of evaluable targets for which the stock’s return at the end of the 12-month period had the direction implied by the target. Median error is the median absolute difference, in percentage points, between the target’s implied return from the reference price and the stock’s return at the end of that period. These measures describe past outcomes; they do not rank skill or predict future results.

Which targets are counted

A target is evaluated only after its full 365-day period has passed.
For the same stock and source label, an unchanged target repeated within 90 days of the previously counted target is skipped. Different source labels are handled separately before house-level aggregation.
A target is excluded if the stock cannot be matched to a listing, there is no daily close on or before its publication date, or fewer than 50 daily price observations are available in the 365-day outcome period.
A house’s percentages are shown only when it has at least 30 evaluable targets. Sector hit rates are shown only for sectors with at least 30 evaluable targets; stocks with no sector classification are omitted.

Consensus and agreement on stock pages

For the 90-day consensus, TargetWatch takes each source’s latest target dated within the last 90 days, applies a price-level filter, and shows the median of the remaining targets. Targets above three times or below one third of the latest share price are filtered out. If fewer than two sources have targets in that window, the page falls back to each source’s latest target regardless of age and marks the data as older.

Agreement is based on the coefficient of variation (population standard deviation divided by the mean) across the included targets: below 8% is “Aligned”, 8% to below 16% is “Some disagreement”, and 16% or more is “Divided”. The label is not shown when there are fewer than two targets.

Difference from peer targets

The bias measure compares each target with the mean of the latest targets from other source labels for the same stock dated within the previous 30 days, including the target date. It is calculated as the target divided by that peer mean, minus one. Individual differences are capped at +40% or -40%; the house figure is the median of the compared targets from the last 365 days. It is displayed when at least 30 targets can be compared.

A difference above +2% is labelled “Optimistic”, below -2% “Conservative”, and the range between them “In line”. These are labels for the difference from peer targets, not a measure of hit rate or a rating. Repeated targets are not removed from this measure. Source labels are combined into house names for the displayed aggregate, but the peer comparison itself distinguishes the recorded source labels.

Data and limitations

Price targets are collected from Placera’s published target-price overview. Outcome calculations use stored daily closing prices from EODHD. The prices are not adjusted for dividends or stock splits. A split during the outcome period can therefore affect whether a target appears to have been reached. The results depend on the available target records and price history; they are historical summaries, not forecasts or guarantees of future performance.