How to read a scan's hit rate

Every scan page here carries a panel that replays the query across the last 250 sessions and reports what the matches did next. It answers one question: has this setup been worth a look? Read it the wrong way and it will happily flatter a scan that has nothing.

What the columns measure

For each past session where the scan matched, the replay takes the close-to-close return over the next 1, 5 and 20 trading days, then aggregates. Three windows, because setups have different natural lifespans. A hammer at support is a claim about the next few days. A golden cross is a claim about the next few months. Grade either on the other's clock and you get noise. Read the column that matches how long you actually hold, and ignore the other two.

Misreading one: hit rate without payoff

A 65% hit rate sounds like an edge. Consider a scan that is right 65% of the time for +1.2% and wrong 35% of the time for −3.1%. It loses money, at a hit rate that reads like a win.

The reverse trips people up more often. Breakout scans like the 52-week high breakout frequently show hit rates near or below 50% and are still interesting, because the whole premise of the setup is a few large continuations paying for a lot of small failures. Read the pair. How often it worked, and how much working and failing each paid.

Misreading two: the average without the count

The other trap is a beautiful average sitting on eleven matches.

Eleven matches across 250 sessions means the scan fires about once a month, and one lucky cluster carries the entire average: a single results week, one index rally. Before you trust a number in that table, look at how many signals produced it. A modest average over three hundred matches tells you more than a spectacular one over a dozen. Scans like volume shockers sit at the far end of that scale, with plenty of signals and weaker meaning in each, and the panel shows you the trade-off instead of letting you assume it away.

What the replay deliberately is not

The panel calls itself a sketch and I chose the word carefully. Close-to-close returns assume you entered at the close on the day the signal appeared, which is plausible for an end-of-day workflow and generous for anything else. There are no transaction costs in it, no slippage, no position sizing. Universe membership is measured as of today, which leans survivorship-positive: a stock that has since dropped out of the liquidity tier is not in there to hurt the numbers.

That does not make the sketch useless. It makes it a filter. A scan that cannot look good under assumptions this friendly has nothing to offer the unfriendly ones, and you found that out in ninety milliseconds rather than ninety trades. The longer walkthrough covers the method end to end.

The habit

Before you act on any scan, whether it is a preset here or something you translated from Chartink at eleven at night, open the panel and ask three questions in order. How many signals? What did my actual holding window pay? Does the payoff survive the hit rate? It takes thirty seconds and I have talked myself out of plenty of scans in those thirty seconds.