T-Mobile Park, solar powered

Proposing a new variable

EDITOR’S NOTE 3/10: Swings & Takes has moved! This post will remain here, as I understand it's been linked a few places around the internet, but you can read it on the new site as well at the following link. Thanks for reading.

—

Batters who say they struggle to see the ball in Seattle are probably onto something.

In September, I introduced a premise. If something about T-Mobile Park is messing with batter vision, then we should see performance change with the position of the sun. Batting is a visual experience, and light (via the sun) changes the way we see objects. As I wrote in the original post:

When the sun is high in the sky, the stadium will look one way. When the sun is close to the horizon, it will look another. When the sun is fully set and replaced by LEDs, it will look another still.

Each position of the sun creates a unique backdrop within T-Mobile Park. If performance changes with those backdrops, we can infer 1) some backdrops are better for hitting than others, and 2) T-Mobile Park is creating an inconsistent visual reference for batters. You can think of the function like so:

performance ← vision ← light ← sun

In the first post, I found whiffs go up about 10% after sunset, and the effect appeared most prominent during the hours of twilight. My conclusions today aren’t much different from then—the effect still exists at about the same size and location.

But I hedged a bit in the original post and listed a series of limitations preventing a stronger conclusion. Since then, I’ve gathered some feedback, improved my data, and learned quite a bit more. I now believe there’s sufficient (statistical) evidence to support my claim, that batter vision is a real issue in Seattle. But as I will discuss, understanding the sun isn’t so simple. There are many environmental factors driving performance at T-Mobile Park, and isolating their influence is decidedly tricky.

So with this post, I’d like to reaffirm why I believe this affect exists, why I know this affects exists, and why sometimes I wonder if I’ve been staring at the sun just a bit too long.

“House Of Horrors”

Park factors, at their core, are simply a comparison of home and road performance. These should be about the same over a large enough sample. When they aren’t, we assume the stadium (the most notable variable) is creating that difference. The average park factor at T-Mobile Park for its existence is 94, meaning there’s been about 6% less offense (by wOBA) in Mariners home games than Mariners road games over the last 25 years. In fact, T-Mobile Park has never recorded an overall park factor greater than 100 since opening in 1999.

One reason is the notorious “marine layer.” Seattle is cool and close to sea level, and it has the densest air on average across MLB. Batted balls travel less distance in Seattle than most other places, and many would-be hits wind up as outs. This is well known.

What isn’t well known is that T-Mobile Park seems to create extra strikeouts. There’s been about 7% more strikeouts in Mariners home games than road games since the stadium opened. These extra strikeouts appear to be going up, too. In 2024, Mariners home games had 22% more strikeouts than Mariners road games.

T-Mobile Park is now nearly as extreme as Coors Field, albeit in the opposite direction. It’s become a "House of Horrors,” as coined by John Trupin at Lookout Landing.

The marine layer has gorged itself on hitters and become something new, fearsome, and untamed. If the Mile High Stadium is the zenith for hitters, T-Mobile Park is its Mariana Trench. When you gaze into the abyss, sometimes strikeouts stare back.

Again, it’s well understood why stadiums affect performance on batted balls, but it’s not so easy to come up with a clean narrative for why stadiums create (or limit) strikeouts.

I contend some of these extra strikeouts must be related to batter vision. Stadiums look different. They’re built with different dimensions, materials, and colors; they incorporate different types and amounts of light. What a batter sees from the box at T-Mobile Park is fundamentally different from what they see anywhere else. I don’t think it’s a stretch to assume there are better and worse backdrops for picking up an object moving 100 mph. In fact, players have been telling us this for years:

Batter vision at T-Mobile Park has been embattled since the stadium opened. Newspaper clippings referencing vision complaints date as far back as 2000, when Edgar Martinez told the Associated Press he was doing daily sight exercises to keep his eye muscle strong.

“With the glare so bad, you can’t see any red stitches on the ball,” Martinez said. “All I can see is white. There is no way to tell rotation, so any pitch that is coming in high near you, you tend to jerk away from.”

Over the last 25 years, team officials have coated the scoreboard with a non-glare surface, covered aluminum seats with green padding, tilted the angle of the batter’s eye “to cut reflection of the sun from the west,” planted and later removed trees beyond the center field fence, and consulted the aerospace industry to develop a batter’s eye material that was described as “very, very, very” black. None of this has satisfied players, as complaints have come and gone for years.

The most important testimony on the matter came this summer. Teoscar Hernández told The Seattle Times he felt crooked while standing in the batter’s box at T-Mobile Park:

“I talk to a lot of players around the league, and they feel the same thing when they go to Seattle and play two or three games over there,” Hernandez said. “They had the same feeling. So it was not only me.”

Hernandez said he never felt like he was lined up in a straight line with the pitcher’s mound at T-Mobile Park.

“It was a little crooked,” he said. “I didn’t feel really straight with the pitcher, for some reason. I moved everywhere in the batter’s box and tried to fix it, but I couldn’t figure it out.”

This is an interesting departure from what we’ve heard in the past. Assuming the field dimensions are accurate, what Teo described sounds a bit like an optical illusion—a function of light. We’ve since heard from former players, notably Trevor May, who corroborated this claim as a legitimate and widely held belief.

Weather or not

Performance literally changes with the sun in Seattle.

One thing I regret about my original post was I didn’t emphasize this relationship can be seen in an average. It’s convenient as analyst to wave some fancy statistics in your face and say, “I’m right, please believe me.” But if the sun were a split on Fangraphs, this relationship would be readily apparent:

The fancy statistics are only necessary to test for causal inference—i.e., whether the sun’s position is directly responsible for changes in performance. For this, we need to find all the things that might influence performance, plug them into the appropriate stats machine (GAM), and test the size and reliability of the sun’s effect holding all else constant.

These are the things I found to matter the most:

• Sun level: Altitude and azimuth

• Environment level: Air density, wind speed, and wind direction

• Pitch level: Pitch type, location, and quality via aStuff+

• Situation level: Handedness, count, times through the order, pitcher “position” (SP or RP), and whether the Mariners are batting or pitching

• Time level: Month and year

• Player level: Underlying batter skill via random effects

In the original post, I tested the effect of the sun on whiffs, and I still think that’s the most conceptually pure test variable out there. An issue with vision would presumably affect timing, and whiffs seem like a logical function of timing—swing late, miss ball. Pitchers always want to generate a whiff; batters never want to generate a whiff. Umpires have no say. Fielders have no say. Batter approach and team philosophy have no say.

Here are all swings at T-Mobile Park from 2021 to 2023. The dots represent the estimated probability of a whiff given the above variables, and the line shows the average whiff rate as the sun sets from right to left.

This plot helps illustrate two things.

The first is the 0-altitude threshold (i.e., sunset) appears crucial. Whiff rate holds mostly steady through the early part of night games at around 25%. At sunset, whiff rate leaps and holds around 28%. This is notable, as sunset represents the most significant change in light throughout the day. What a batter sees after sunset is quite different from what they see before, and this change in performance is the foundation for my claim.

The second thing you might notice is most of the variation in whiffs is left unexplained by this model. The r^2 here is about 7.33%. This is not inherently disqualifying, but keep that in mind as you read this analysis.

Still, to the extent these variables can be used to estimate whiffs, the sun’s position does have some say. We’re looking for p-values less than .05 to determine statistical significance:

This shows the change in whiff rate we see at sunset is unlikely to have occurred by chance. The sun’s position is, definitively, a predictor of performance in Seattle. Whether that’s because of sunlight itself is the next question to address.

You’ll notice the p-values for air density and wind don’t present a significant relationship with whiffs. Digging a bit further, however, shows it’s not so simple. Data from OpenWeather isn’t measured in the 60 feet between the pitcher and the batter. It’s measured at some abstract space above the stadium. As research from Weather Applied will tell you, wind doesn’t move directionally at field level—it whirls and swirls and hurls through the stadium. I can’t account for that.

What I can account for is pitch quality. There are several “stuff” models that aggregate pitch speed, spin, and movement to score a pitch by the effectiveness of its properties. Adam Salorio, who wrote one of the essential accounts of T-Mobile Park for Pitcher List, was kind enough to send me some data from his model, dubbed aStuff+. This was a crucial addition because aStuff+ is a strong predictor of whiffs, and because it helps address our issue with weather data. If the environment is affecting pitch movement (or something else that matters), aStuff+ should help eliminate some of that bias.

The second concern here is that air density and wind speed both change with the sun, meaning it’s possible the sun’s position is simply riding on their coattails. Here’s how air density and wind speed are distributed:

There’s a couple ways to wrestle with this and isolate the sun.

The first is to match variables—that is, create samples on either side of sunset with identical (~99%) distributions of air and wind. Doing so drops total swings from 19,524 to 13,384 in the matched data, which should be fine.

The second is to create a series of new terms to capture the way these variables interact. There are so, so many ways to do this, from the overly simple to the overly complicated, and it’s unlikely I’ve exactly nailed the total effect of the environment here. But I think the terms I settled on are good enough to help us broadly understand what matters.

I ran the model again over the matched data with the new terms included. The r^2 jumps to 8.11%.

The sun is a large, statistically significant, and now independent predictor of performance in Seattle. Its effect remains, even after adding better controls for weather. Again, there were many choices to make in building this model, but I found the sun’s position to be a remarkably consistent predictor across various tests.

This table also shows no global effect of air density on whiffs within T-Mobile Park. The same goes for the new interaction term between air density and wind speed. I ran many versions of this model and did not find any results where air density (or its components) was a significant predictor of whiffs. Now, this isn’t to say air density doesn’t matter. Seattle has MLB’s highest average air density, and pitches there will move a bit different relative to, say, Denver. But I suspect the effect of air density is only observable in its distribution between parks, rather than in its relative changes within parks. In other words, Seattle’s air density might be effecting baseline performance within T-Mobile Park, but changes in air density are probably not the culprit for the time-series variation we see with whiffs.

Wind tells a different story. The table above shows near-significance for the global wind speed term. The other term, used to capture wind’s interaction with aStuff+, shows an even stronger relationship. This suggests changing wind patterns may amplify (and dampen) pitch quality in a way that meaningfully influences whiffs.

There’s one more term to look at here:

This is, I think, the most interesting piece of evidence here. Basically, what this term does is measure the effect of wind patterns on whiffs from either side of sunset independently. Again, keep in mind the distribution of wind speed and direction is identical in both samples.

When the sun is up, changes in wind patterns at T-Mobile Park do not change the probability of a whiff. When the sun is set, changes in wind patterns are an exceptionally strong predictor of whiffs. In other words, yes, the wind is subtly affecting pitch quality, and that matters most within the suspected negative vision environment after sunset.

The sun is, definitively, a predictor of performance in Seattle. The question addressed in this section is whether that’s because of light and vision, or because the sun is obscuring another, hidden relationship. In my best attempt to isolate these variables, the sun does appear to have its own independent effect.

Light is a variable in Seattle.

A glaring omission

I’ve only discussed night games to this point.

Day games appear to have one, mostly even, baseline performance level. This makes sense because day games don’t have nearly as much change in light, whereas there are real, astronomical definitions for the stages of light we see in the evening. The sun’s movement in day games doesn’t obviously change what a batter sees, and crucially, those changes don’t affect performance.

The issue, however, is that day games show a baseline performance level similar to what we see after sunset in night games, not before.

The only way this doesn’t undercut my theory is if there’s a separate negative vision environment during the day. Is that a stretch? I don’t think so, given glare has been a common complaint from players for years.

This isn’t easy to prove, however. I tried looking at performance against other weather descriptors (clear skies vs. overcast vs. roof closed), but I didn’t find any reliable results against these variables. I tried looking into specific east-west positions of the sun by azimuth, and while there were some nuances, I ultimately could not attribute the higher baseline in day games to anything specific.

Here’s my very broad theory:

• When the sun is high in the sky, light reflects off the surfaces of the stadium to create glare for batters;

• When the sun is close to the horizon, the stadium is cast in a soft, natural light that makes it easier for batters to pickup the ball;

• When the sun is set, something about the darkness mixes with the stadium to create an optical illusion. I can’t pinpoint the exact cause, but I suspect it’s related to the “very, very, very” black batter’s eye. Perhaps at night the wall disappears into the darkness, depth perception is altered, and the ball looks as if it’s coming from nowhere.

Trying to rationalize the mechanics here is difficult without more direct data. I think that’s OK. The original premise for this project was not that there would be a specific effect, but that there might be an effect at all. It’s not that darkness or brightness are the enemy, but that light (via the sun) is simply a tool we might use to look for inconsistencies in batter vision. And I think that’s what we see at T-Mobile Park.

Why you might care

The next place to go in this research is to do a formal, less napkin-y version of something I attempted in October. Unless you’re here for the subplot (a tour of statistical modeling), then you probably care most about contextualizing all this information—how it affects Seattle’s run environment, the Mariners, and MLB. But doing so requires understanding the sun at every other stadium, and that opens a mind-bending matrix of comparison.

In an informal look, I found what’s happening at T-Mobile Park to be somewhat unique in its size and reliability. But it’s also not the only place where the sun is a variable to some degree. Grappling with all this will take time, and figuring out how to present it will take even more. That’s to say, there are no firm conclusions to be found in the chart below, but this is where we’re headed:

About one-third of the time, T-Mobile Park performs close to neutral by our favored “true talent” metrics. Seattle isn’t one run environment—it’s at least two and probably three.

If these splits are indeed driven by batter vision, then a solution may exist. The Mariners can’t change the air density or wind speed in Seattle. They can’t “fix” climate-related run suppression without building a new stadium or heavily modifying the existing structure. But they could absolutely change how the stadium looks to a batter.

That’s why this matters. That’s why I’ve felt the need to spend so much time on this subject. That’s why I’m encouraging others to think about this, too. Because if we think T-Mobile Park could or should always perform at its best, there’s a good chunk of “action” waiting to be clawed back into games.

And that applies everywhere.

I keep coming back to this article about the Astros painting “the right side” of the batter’s eye in the middle of the 2023 season. I was surprised they could just, like, do that. Nothing in the rule book implies they can’t. In fact, the section governing stadiums is surprisingly light. The color of a pitcher’s glove is regulated, for instance, but most everything else in view is not. I’ve since found many similar stories of players complaining of vision issues, and teams coming up with arbitrary (perhaps desperate) sounding solutions. In the era of efficiency and standardization in MLB, it feels a bit shortsighted to leave one of the game’s truest fundamentals up to team staffers and a can of paint.

If MLB is looking for non-invasive, administrative solutions to a run environment in decline, maybe it should explore what can be done for batter vision. These could be as unobtrusive as massaging game times and travel schedules to play in more favorable settings. (Notably, this is true, even if you disagree about vision and light.)

A more direct approach would be to understand and regulate the batter’s eye, or the light bulbs, or the seat colors, or the whatever. This wouldn’t be simple, as each stadium likely requires its own solutions. But, say, adding visual depth to the centerfield concourse at T-Mobile Park wouldn’t fundamentally change baseball itself (and it might be more effective).

Or maybe it’s time to explore better gear. A few NHLers have taken to wearing pink visors to reduce glare from LEDs after concussions. I can’t speak to the science there, but I assume there must be some workable parallel. Heck, maybe Enrique Hernández was on to something when he hit that homer in the NLCS with pink tinted frames. Hernández, of course, found out midway through the season he should be wearing glasses at all.

The players are telling us the way a stadium looks matters. Maybe more runs are just out of sight.

####