New Yankee Stadium Home Runs Not Necessarily Related to Weather

Yesterday AccuWeather via their blog put forth a hypothesis that the well documented increase in home runs at the new Yankee Stadium was due to a change in the stadium’s shape, allowing for west winds to enter and exit the stadium at a high speed and the proper angle to turn lazy fly balls into home runs. More specifically:

The wind on Saturday during the Yankees' 22-4 loss against the Cleveland Indians was mostly from the west at 15-20 mph. Given the layout of the stadium, the wind could have had an effect on fly balls in right field. Six home runs were hit by the Indians in the second inning alone.

Although the field dimensions of the new stadium are exactly that of the old stadium, the shell of the new stadium is shaped differently. AccuWeather.com meteorologists also estimate that the angle of the seating tiers in the new stadium could have a different effect on wind motion across the field.

Note the lack of scientific explanation for the statements in the article. Notice how many times words like mostly, could, estimate are used in the paragraphs. No modeling, or observational studies were used. Let me translate that for you into plain English, “hey we watched the game on Fox on Saturday and it sure looked like the wind was blowing the ball hard, maybe it’s the new stadium?” Look it seems like a reasonable hypothesis and all, especially watching some of those pop flies end up as home runs on Saturday, but it’s just a theory and the accuweather story struck me as being less science and more guessing. I was frustrated that a weather company would present this theory with only one games worth of wind data as as motivating observation.

So I wanted to test their theory with a bit of weather data to see if it holds any water. The methods are explained below and the study includes only six games the four versus the Indians this past weekend and two exhibition games versus the Cubs in early April. The chart shows the number of home runs hit each game, the mean direction of the wind and the wind speed. First note, only three of the six games hosted at Yankee Stadium so far had a west wind. So three of the games played so far had winds not out of the west, and often out of the opposite direction (east), which would by the accuweather theory suppress home runs. In each of these games where the wind was not out of the west, 3 home runs were hit.


Table 1. The above table shows the average wind direction, and speed for games at Yankee Stadium so far in 2009. See the methods section below for details.

During games with a west wind, 8, 6 and 5 home runs were hit, each a significant amount of home runs for a MLB game. But might these home runs be due to the players involved in the game and not the weather? I’m not a saberatician so excuse my clumsy attempt to convince you that there is just as good a chance that the HR’s were due to good offences hitting against struggling pitching as there is a chance that the wind was directly responsible for the home runs. Cleveland lead the American League in runs scored after the all-star break last year and is #2 in runs scored in this young season. During the west wind games this high powered offense was coupled with some exceptionally poor pitching. Of those three games with a west wind, two of them were started by Chien Ming Wang who has for a lack of a better word been awful both at Yankee Stadium and away so far this year. Other pitchers starting in these high home run games were Fausto Carmona (5.85 ERA in 2008, 7.87 ERA in 2009) served up 4 HR, Anthony Reyes (5.73 ERA in 2009, was lights out in second half of 2009) served up 3 HR while Joba Chamberlain (5.60 ERA in 2009 and 2.60 ERA in 2008 and arguably the best pitcher of the crew) only allowed 1 HR. Relievers who served up home runs include minor league pitcher Anthony Claggett who appeared to be overmatched, Jose Veras, Edward Ramirez, Zack Jackson and Jenson Lewis whos ERA’s range from 5.79 to 43.20 this year. Johan Santana, Tim Lincecum, Brandon Webb and Mariano Rivera these pitchers are not.

It’s too early to draw any conclusions from the weather data presented above. But it’s clear that a number of factors not related to the weather may have resulted in the home run deluge, and we can’t say for sure if the west wind is the cause of the HR deluge:

• The accuweather hypothesis appears to be at best an untested guess, not based on modeling studies or scientific observation.


• Winds have been equally out of the west as other directions, thus they can’t directly explain the deluge of runs and home runs scored over the weekend.


• More home runs have been hit when there is a west wind at Yankee Stadium than when the wind has been from the east or south.


• In the games with west winds the pitching has been poor, pitched either by pitchers who do not have demonstrated long term records of being above league average or are who are clearly struggling so far this season.


• Cleveland and the Yankees are amongst the two best offensive teams in the American League


• At this point there is not sufficient data to prove whether or not wind direction is statistically tied to home runs.

As Peter Abraham pointed out in his blog today, it’s too early to rush to judgment, the Yankees and opponents hit 20 HR in a four game series last year as well! It’s a neat theory to tie wind direction to HR production, but I think it’s irresponsible for accuweather to claim that the west wind is leading to these home runs. We’ll have to see what the numbers bare out this summer, and draw conclusions after a sufficient sample size has been drawn.

Methods. I calculated the mean wind direction, wind speed and air temperature at LaGuardia Airport from raw METAR code taken from the Plymouth State Weather Data Archive. LaGuardia Aiport was chosen to represent conditions in the Bronx due to the well documented problems with the Central Park observation site. The averages were calculated for the hours of the game, based on data taken from box scores, except during the exhibition games vs. the Cubs where length was estimated to be 4 hours.


Table 2. As per table 1, but includes standard deviation and temperature data as well.