Anytime home run props realized about -23%

Anytime home run props are the worst-priced bet measured on an MLB board so far. Graded against real outcomes rather than against simulations, 18,468 of them returned -23.1% per bet at the best posted price.

Written by Jesse, NegativeEV. Last updated 2 August 2026.

• Sample: 18,468 anytime home run props, from 1,104 games over 91 slates
• Window: 2026-04-19 to 2026-07-31
• Priced at: best posted price
• Measured: 2026-08-02

How much did anytime home run props return?

-23.1% per bet, graded against what actually happened rather than estimated.

Realized EV per bet, same pass	Figure
Anytime home run (n = 18,468)	-23.1% (± 1.7, z = -13.3)
Over 0.5 hits (n = 18,541)	-5.1% (± 0.7)
Pitcher strikeouts, over side (n = 2,061)	-4.5% (± 2.2, z = -2.1)

The home run props implied a 14.8% chance of cashing on average and cashed 11.8% of the time: 2,170 of the 18,468 hit. About 94% of the props were priced above the realized rate of their own price bucket.

Why are the three markets so far apart?

The three markets were graded in one pass over the same window, which is what makes the spread readable: the label "player prop" covers an eighteen-point range. Home run props sit at +250 and longer, in the longshot band where the margin is thickest, while hits and strikeout lines sit near even money.

Did this pass change a figure already published?

Yes, one. The same pass measured the simulator against these realized numbers and revised a public figure: the strikeout claim had been -3.2% from simulation and came back -4.5% ± 2.2 against real outcomes. On 2,061 props that interval clears zero, so the strikeout side of the comparison is now a measured cost rather than an estimate - a small one, next to the home run figure. A blanket market claim gets validated against what actually happened, never against the simulator alone.

Method

Every archived anytime home run prop is graded against whether the batter actually homered in that game, from the mapped play-by-play corpus, and reported as the mean implied probability the price charged, the realized rate, and expected value per bet, bucketed by price. Props for games missing from the corpus are excluded rather than defaulted.

Inside the noise floor

All three rows clear their noise floor on this window. The home run figure is -23.1% ± 1.7 (z = -13.3) on 18,468 props, the hits figure is -5.1% ± 0.7 on 18,541, and the strikeout row is -4.5% ± 2.2 (z = -2.1) on 2,061 - the last of these only narrowly, and it is the one to read with the most caution. What the census establishes is the spread between home run props and the near-even-money markets, not any ordering inside the near-even-money group.

What does this page not establish?

Whether any single prop is priced well. These are market-wide averages over one fixed window, and an average that bad still contains fairly priced props; nothing here says which. The window also starts on 19 April, where the archived odds start, so the opening weeks of the season sit outside it.

Written up for readers: https://negativeev.com/about/home-run-props All findings: https://negativeev.com/research