Model prices BAL +135 at 47.9% — a 12.65% edge against a one-sided market

This is a pregame read, priced before first pitch — not a live in-game number. Baltimore moneyline at +135, model probability 47.9%, expected value +12.65%, four stars. The engine would take it.

What we found

+135 implies the market thinks Baltimore is meaningfully worse than a coin flip. Our plate-appearance simulation puts them at 47.9% — close enough to even that the price is the whole story. That gap is the +12.65% EV, and it is the second-largest moneyline read on an 8-game board, behind only STL ML +167 (model 41.6%, EV +11.1% on its own line).

What makes BAL @ TB worth writing about is not the number in isolation. It is where the money is.

The trends

The tracked money on this game is emphatically not on Baltimore. Two of the three largest campaigns on the board today live in this game: $216,028 on Over 8 at 66% one-sided, and $81,168 on Tampa Bay's moneyline at 98% one-sided. Across the full slate we are tracking $906,280 over 12 campaigns, with the biggest single ticket at $24,703 — so this one game is absorbing a disproportionate share of concentrated pregame flow, and nearly all of the moneyline share is against us.

The over money has a plausible story behind it. Tampa Bay sends the coldest arm on the board: Freddy Peralta at an 8.44 ERA over his last 5, 8.18 over his last 3, against a season mark of 5.33. That is a pitcher trending the wrong way on both windows, not a single blowup dragging an average. It is also, mechanically, an argument for Baltimore's bats — the same input that fuels a 66% one-sided over is an input to our 47.9%.

That over lean is not isolated: all 8 flagged total edges on today's board point over, led by SEA @ HOU at a fair 10.15 against a 7.5 line, a +2.65-run gap. When every total edge points the same direction, the honest read is that the run environment is doing work in the model, and some of that is going to be model, not market.

The model also leans underdog in 5 spots today — BAL ML +130, WSH ML +153, STL ML +167. Five dog leans on an 8-game board is a systematic tilt, and it is the sort of thing that either reflects genuine market overcorrection on favorites or reflects our simulation under-weighting them. We do not get to know which from one slate.

Movement elsewhere shows what a market agreeing with itself looks like: 335BOSPIT-BOS drifted 49.5c to 52.5c (+3.0) over the last 20 hours, 337NYYTOR-TOR 50.5c to 53.5c (+3.0), 515STLCHC-CHC 60.5c to 62.5c (+2.0), 340SDCLE-SD 50.5c to 51.5c (+1.0). Only 335AZATL-ATL went the other way, 54.5c to 53.5c (-1.0). Boston, for reference, is carrying $101,041 at 99% one-sided on its moneyline and moved 3 cents — money and line agreeing.

What it means

The honest framing: this is a disagreement, not a discovery. We are on the opposite side of 98% of tracked moneyline money in this game, on a number the engine says is a coin flip and the market says is not. One pregame price, one game, sample size of one — no result here confirms or refutes anything.

What has a sample is the signal book. VELO_DROP is running 25.6% ROI over 101 graded bets and TRAILING_LATE 10.4% ROI over 106 graded bets — both out-of-sample, both graded. Those are the numbers with track records attached. A single +135 dog is not evidence; it is an entry in a ledger that will be evidence in a few hundred bets.

What we are testing next: whether the 5-dog lean is a repeatable market inefficiency or a calibration artifact. If our underdog reads across a few hundred graded bets close at prices better than we took, the model is finding something. If they close worse, we are reading noise and we will say so.

What would change our mind: a Peralta scratch, or the moneyline moving toward Baltimore before first pitch — a price that closes at our number means the edge was already gone when we took it.

21+ where legal.