The NPB King is a research tool for NPB player props. We project a realistic number for every player on the slate, compare it to the posted line, and surface the spots where the two disagree the most. Here's the whole approach, in plain English.
For each player and each stat, we estimate what we actually expect to happen tonight — not a guess, but a number grounded in recent form, the matchup, the role, and the conditions of the game. That projection is the foundation for everything else.
Every prop comes with a line. We show our projection and the model's win probability beside it and leave the comparison to you. We used to publish the gap as an edge and sort it into Elite / Strong / Lean / Pass tiers; we no longer do.
We graded them. This ledger carries no repaired-inputs marker, so no period is held back from the figure and none is separated out of it: across all 84 graded slates the board is 55.0% — 946-775 on the 1,721 picks that were decided, plus 81 pushed, out of 1,802 picks graded in all. The count and the rate do not share a denominator: a push sits inside the graded total and outside the hit rate. That is short of the 57.7% a two-pick entry has to clear. The three tiers we used to surface — Elite, Strong and Lean — were retired on 2026-07-27. On the record as it stood that day they came to 57.4% (290-215), but the slate-clustered 95% interval on that was [53.3%, 62.0%], whose lower bound sat below even the ~54% break-even of a single -119 leg; the record could not separate the tiers from no edge at all, and the Lean tier (54.3%, 101-85) was flat on units. Those are the numbers of 2026-07-27, kept here as the reason for the decision and deliberately not restated as a current figure — the current figures are the ones above, and they are derived, not typed. A tier we cannot show works is not a conviction, so we stopped printing one.
Worth knowing how the ledger is built: the nightly grader re-derives every result from the full snapshot history, and it keeps every one of them. What it publishes is the most recent 300 rows, so the row-by-row ledger you can inspect is a rolling window that does not accumulate. The totals quoted above are computed over the whole graded history, not over that window — which is why the counts above are larger than the table below. What you can check row by row is what there is.
Most plays are submitted as multi-pick slips, so a single legs page isn't enough. We assemble the day's combinations — balancing the chance of hitting against the payout on offer, and how the picks relate to one another. We no longer publish an EV number or a recommended stake for them.
Once games finish, every pick we posted is graded — win, loss, or push — and added to our track record. You can see our hit rate over time. Nothing gets quietly deleted after the fact.
See our model accuracy → — the reliability curve and Brier score proving our published probabilities match real-world frequencies. Calibration, not a betting record.
No — and we do not claim to beat the price. Player props are inherently uncertain, our projections are often wrong, and the graded record on this site does not clear the pick’em break-even. Treat everything here as research, which is exactly why we publish the whole track record.
The board refreshes through the day as lines and information move, so the projections you see reflect the latest available picture before first pitch.
Power slips pay more but every leg must hit. Flex slips pay less per win but still return something if one leg misses. We build and size both so you can choose the risk profile you're comfortable with.
We do not recommend a stake. Only ever risk money you can afford to lose. This is an information tool, not financial advice.
The NPB King is for entertainment and informational purposes only. 21+. If gambling is no longer fun, it's time to stop — please play responsibly.