Sparrow Data

Graded prediction archive

Players

205 players have appeared on a board. Counts are GAMES, not board slots: a bat named on two boards the same night counts once. Click any column heading to sort, again to reverse. Type to filter by name.

Minimum graded games: 0 — drag right to drop thin samples before sorting by rate.

Sorting by more than one column: click a heading to sort by it; shift-click a second heading to use it as a tie-break. Active columns show their priority (▼1, ▲2). Note that a tie-break only bites among rows that are EQUAL on the first column — to find bats picked often that homered rarely, the question is really “how did he do against what we predicted”, so sort by Expected descending, or by HR rate ascending with the minimum-games slider raised.

Expected is the sum of our own probabilities across his graded games — what the model said he would hit. A bat with 12 picks, 1 home run and 1.4 expected did roughly what we predicted; the same bat with 3.0 expected did not. Raw counts alone cannot tell those apart, which is why the column is here.

HR rate is home runs divided by graded games — descriptive only. It is greyed for the 136 players with fewer than 5 graded games, which is most of them: the median player has 2. At those sample sizes a 100% rate means a bat was picked once and homered, not that it is a better pick than one at 30%.