Methodology
Everything on ArchivePicks is one of three things: a recorded statistic, a price that a sportsbook or prediction market has posted, or arithmetic on those two. This page spells out the arithmetic so you can check it and judge how much weight a number deserves. For an overview of the site, see About ArchivePicks.
The hit rate
A hit rate is the share of a player's games in which a stat finished over a posted line. If the line is 64.5 receiving yards and the player went over it in 7 of their last 10 games, the L10 hit rate is 7 of 10, or 70%.
- L5 uses the player's last 5 games, L10 their last 10, and Season the games played this season.
- On a player's page the chart can also cover the last 15 games, and you can drag the line to see what the hit rate would have been at any other number.
- A rate is only shown when the player has played enough games to fill the window. Otherwise the cell is left blank rather than computed from a shorter sample.
Missing games are never zeros
If a stat was not recorded for a game (a player was inactive, or the data source did not report that stat), the game is treated as missing and left out. It is not counted as a zero. Counting it as a zero would make a player who sat out a game look like he went under every line, which is wrong: he never had the chance to clear it.
Sample size
A hit rate is a count over a small number of games, so it moves a lot. Always read the "of N" next to the percentage: 3 of 3 is 100% but says much less than 14 of 20. Early in a season the Season rate rests on only a few games, and if a player has not yet played this season the chart falls back to their most recent games so it is not empty.
Head-to-head history works the same way, with a smaller sample. A player usually meets a given opponent only about once a season, so those results are shown with their size and should be treated as context, not a pattern.
Where the inputs come from
| Input | Source | Freshness |
|---|---|---|
| Game logs, rosters, team stats | api-sports.io | Loaded into our database every night |
| NFL lines | Courtside, Underdog, Kalshi, Polymarket | Refreshed every couple of minutes |
| NBA lines | Courtside, Underdog | Refreshed every couple of minutes |
| College football lines | Underdog | Refreshed every couple of minutes |
A price on the board can be slightly out of date and may already have moved at the source. The statistics behind a hit rate update overnight, so a game played this evening appears the next day.
Prices from prediction markets
Kalshi and Polymarket quote probabilities rather than American odds. ArchivePicks converts those to American odds so they sit next to sportsbook prices, and each side is shown at the price someone could actually buy it at.
Fantasy values
Fantasy player values are built from recorded NFL season statistics, scored under PPR, half-PPR or standard formats, then ranked into value tiers by position. "Trending" flags players moving up or down over recent weeks. Values are a way to compare players, not a forecast of how any one of them will perform.
What these numbers do not do
- They are not projections. ArchivePicks does not predict how a player will perform.
- They do not account for injuries, matchups, weather or role changes beyond what the box score shows.
- Past results do not predict future ones. ArchivePicks does not take bets or tell you which side of a line to take.
Spotted something that looks wrong? Tell us on the contact page.