Chalk of the Day, August 16: Alvarez to homer at +250

The chalkiest home run prop on the Sunday board, checked against 5,000 play-by-play simulations of the Mariners at the Astros. The +250 price implies 30.0%. The simulations say 28.6%. Roughly Fair; it becomes a play at +249 or longer.

Written by Jesse, NegativeEV. Last updated 16 August 2026.

• Sample: 5,000 pitch-by-pitch simulations of one game
• Window: 16 August 2026 slate
• Priced at: +250 at Bovada, as quoted off the board at publication
• Measured: 2026-08-16

The bet

The check:

This is the most-bet home run prop on Sunday's board, and the simulations land close enough to the price that there is nothing in it. 15 games had odds, 9 of the chalkiest selections were graded against the sims, and the verdicts came back 3 Negative EV, 2 Roughly Fair, 4 Positive EV. Nobody here is picking anything. This is a price the public is already taking, checked.

The board ran from +200 to +250 on this prop across 4 books at the time of the check, against the +249 this needs to break even.

This prop is quoted over-only, so there is no two-sided market to strip the book's rake out of. The honest comparison is the simulated number against what the quoted price implies, and some of that gap is the rake itself.

What the simulations did

Across 5,000 simulations Alvarez's average is 0.33 home runs and the median is 0. The most common outcome by a distance is none at all.

The shape of 5,000 simulations:

The anytime line pays the same for all of the multi-homer games, and there are 4.2 in every 100 of them.

At 5,000 simulations that 28.6% carries a standard error of 0.64 points, so 95 runs in 100 would land between 27.4% and 29.9%. The gap to the price is about 0 points, roughly 0 times that error bar. Said narrowly, and it should be said narrowly: that means only that re-running the simulator would not move the number far enough to reach the price. It is not a claim that we are right and the book is wrong. A number that is systematically off is off to as many standard errors as you like.

How many cracks he gets

Alvarez bats on the home side, which cuts the other way here: his last trip only happens if the game needs a bottom of the ninth, and it does not always need one. Across the runs he comes up 4.30 times on average, at-bats plus walks, and the home run number moves with that count.

Trips to the plate, and what each one is worth:

Pooled across every simulation he goes deep on 7.7% of his trips, so whether he bats one more time or one fewer is worth more here than any read on his swing. That extra trip is not his to take either. It depends on the eight hitters around him keeping the order moving, and in the simulations they do not always do it.

What works for him

Two arms clear their own noise on this side, and together they are the story.

• His own career profile, who he is as a hitter, props the number up by about 5.8 points. Neutralize it and 28.6% becomes 22.9%. HR rate .074, 13 of 304 hitters, top 10%. BABIP .317, 60 of 304 hitters, top quarter. K rate .179, 79 of 304 hitters, middle of the pack. Exit velo 93.9, 6 of 304 hitters, top 10%. Launch angle 17, 253 of 304 hitters.
• The weather props the number up by about 4.3 points. Neutralize it and 28.6% becomes 24.3%. That arm neutralizes temperature, wind and conditions at once, so the cause is not separable. It is 94 degrees with a 10 mph wind blowing out to centre.

What works against him

Two arms clear their own noise on this side, and together they are the story.

• His own recent form takes about 4.1 points off. Neutralize it and 28.6% becomes 32.7%. HR rate .070 against .079 on the season, .074 for his career, 28 of 304 hitters, top 10%. BABIP .340 against .324 on the season, .317 for his career, 48 of 304 hitters, top quarter. K rate .050 against .178 on the season, .179 for his career, 7 of 304 hitters, top 10%. Exit velo 95.9 against 94.4 on the season, 93.9 for his career, 16 of 304 hitters, top 10%. Launch angle 25.4 against 19.5 on the season, 17 for his career, 277 of 304 hitters.
• Playing at home takes about 2.0 points off. Neutralize it and 28.6% becomes 30.6%. His last trip only happens if the game needs a bottom of the ninth, and it does not always need one.

The pitcher

A right-hander, 2,043 pitches into the season and ranked against the 558 pitchers with enough of them to compare. He releases from a low three-quarters slot, 4.98 feet up (8th percentile) and 1.57 feet across (37th), with 6.41 feet of extension (48th percentile).

The ballpark

Daikin Park ranks 6th of 30 for home runs, with a park factor of 1.20 where 1.00 is neutral. It is one of the friendlier parks to hit one in. In the 2026 data home runs land on 5.9 of every 100 balls in play in the Astros' home games against 4.9 in their road games, 2,612 batted balls at home, 3,009 away, the same club on both sides of the split, so what is left is the park. Retractable roof, grass, 315 feet to left, 409 to centre, 326 to right, 45 feet of elevation. Tonight it is 94 degrees with a 10 mph wind blowing out to centre.

Yordan Alvarez, recent against season:

• usual lineup spot: 2nd (over 120 games)
• plate appearances this season: 507
• plate appearances in the recent window: 20

Those are the inputs the simulator was served, not a separate stat source.

The rest of his night

The home run does not arrive alone. Split across the 5,000 simulations on whether it lands:

:

The game itself averages 9.37 total runs. His side scores 6.23 when the homer lands against 4.48 when it does not, so this is not an isolated event - it travels with the whole offence.

What this does and does not say

It says that at +250 the price and the simulations are close enough that there is nothing to claim either way. He goes deep in 29 of every 100 simulated games, and the price is asking for about the same. Every play is relative to its price, and the two numbers meet at +249.

Method

The game is simulated pitch by pitch 5,000 times from the projected lineups, and the prop is graded on how often it lands. 37 factors were neutralized one at a time, each arm a separate 4,000-simulation re-run, so one-at-a-time swings do not sum to the projection. Those 37 cover everything that could bear on this hitter rather than every factor in the game: the other hitters' own career, season and recent tiers were skipped on purpose, because they are most of the cost and none of the answer for a prop about one man. 13 of the 37 came back with a direction we can stand behind. The price is +250 as quoted at Bovada at publication, and the check is against that quoted price, rake included.

Inside the noise floor

24 of the 37 arms sit inside their own noise, including Bryan Woo's season form (+1.0), what Bryan Woo has been throwing lately, his pitch mix (+0.8), Bryan Woo's career profile, the pitcher he faces (+0.6), the rest of the order, the other eight, without him (+0.5), his own season form (+0.3), the ballpark (-0.6), how long Bryan Woo stays in (-0.1). Those numbers are real measurements and they are printed rather than dropped, but at this simulation count the arm cannot tell a small real effect from no effect. So none of them does nothing, and none of them can be named the thing driving this.

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