One honest number,
and how we get there.
Every home run prop on the board runs through the same five steps. No hidden layers, no score mash-ups, no rules that exist because they used to.
- 01
Who is eligible
A bat has to be in tonight's lineup and never scratched from it, carry a real posted price, be in a game that has not finished, and not have been dropped several spots in the order since the lineup posted. That is the whole eligibility test. We also measure whether he has a genuine power profile — ISO of .190 or better, ten or more home runs, or an elite barrel rate when ISO is missing — and a bat without one is flagged and capped on his card, because over thirteen slates those bats homered at about 6% against a board average near 10%.
- 02
A projected home run number for tonight
Every bat that clears the gates gets a projected home run total for tonight, and the letter is the band that projection falls into — with the bands cut at hit rates we measured on our own graded slates, not at round numbers. Bats the projection does not carry go through a checklist of the same power and matchup inputs instead, so a strong card is not missed just because the projection skipped him.
- 03
Signals are measured, not assumed
Every composite signal is scored against what the model already expected of that bat, and recomputed nightly from our own graded slates. Most of them do less than they look like they do: a worn bullpen measures 1.42x, high volatility 1.22x and HR-or-Bust 1.15x, while Double Edge sits at 0.94x and Barrel Match at 0.87x — both below neutral. Barrel Match read like a 1.3x signal until we divided out what the model had already priced in. These numbers are shown to you and are not quietly multiplied into the grade; they are context for your own read, and a signal that stops working gets published at its real number rather than retired quietly.
- 04
The letter is a seat, not a threshold
Eligible bats are ranked by their projected home run total and the seats go out in order — roughly seventeen S, twenty-six A, thirty B and thirty C on a normal fifteen-game night. That works out to about the top 6% of the board for S, 16% through A, 27% through B and 38% through C. The number of seats scales with how many games are still alive, so a four-game night hands out fewer letters rather than promoting weaker bats to fill a quota. Below the last seat a bat carries no letter at all, which is the model saying 'not today' instead of padding a bucket.
- 05
Where the price does come in
The price never adds a point to a bat's score — that number is power, matchup, park and bullpen, and nothing else. But the price does decide who is shown a letter. A bat with no posted number cannot be graded at all, and the full range of letters is reserved for bats priced in the middle of the board; outside that lane only the very top grades survive. So two bats with the same score can end up with different letters if one of them is priced somewhere we would not send you.
What the price does and does not do
No market data adds a single point to a bat's score. Line movement, sharp money and our own mispriced flag are measured and shown to you, and none of them make a hitter look better on his own merits than he is — the score is power, matchup, park and bullpen.
The price does decide who carries a letter. A bat with no posted number is not graded at all, and the full range of letters is reserved for the middle of the board; outside that lane only the top grades survive. So two bats with an identical score can end up with different letters when one of them is priced somewhere we would not send you. We would rather say that plainly than claim a separation we do not keep.
Results for this model are published in full on the track record page, graded from predictions recorded before first pitch. For entertainment purposes only.