With the Pittsburgh Pirates’ off season additions of Cory Dickerson and Colin Moran, along with Josh Bell and Gregory Polanco already on the roster, the team looks primed to hit a lot of homeruns from the left side of the plate.
Given that the Clemente wall sits just 325 ft. down the first base line, many fans were encouraged that the Pittsburgh Pirates were finally looking to take advantage of PNC Park’s short porch in right. Given the Pirates well noted home run woes last season, the Pirates can and should take every advantage available to increase their power numbers.
There were, however, some local sports media reporters who took exception to this rather intuitive belief that lefties can poke more homers over the Clemente Wall. The skeptics pointed to a metric called “Homerun Park Factor”, which attempts to measure the ease with which a batter hits homeruns in a given venue. “Park Factors” attempt to give us a basis with which we can compare players who play in different stadiums and could be used to compare the stadiums themselves.
The classic example of this is a pitcher in Colorado who has elevated numbers as a result of his home park being spacious and the air being thin being compared to a pitcher with better numbers playing at a sea-level park like Dodger Stadium. If we want to compare them with any kind of fairness, we have to control for the differences in their respective home parks and Park Factors do provide this to an extent.
How they work
The way park factors usually work is that 100 is average, above 100 means it is easier to hit, and below 100 is harder to hit. Moreover, if a stadium has a Park Factor of 105, then it is 5 percent easier than average to hit and if the park factor is 95, it is 5 percent harder than average to hit; the park factor minus 100 is the percentage easier or harder than average to hit in a given park.
What skeptics have correctly pointed out is that PNC’s Park Factor for home runs has been below average for lefties for most of its history. According to Fangraph’s Left Handed Home run Park Factors, in 2017 PNC’s Lefty Factor was 98, or 2% more difficult than average for a lefty to hit a home run; in PNC Park’s history the lefty Park Factor has never been higher than 103 and has gone as low as 92.
Park factors were largely created in the early 2000’s and were a bit ambitious. Sabermatricians wanted to create a metric that could be calculated back through baseball’s history, so that they could better compare historical players. Given the data available when they were created, and for what their purpose is, Park Factors work well enough. The issue is that they make pretty heroic assumptions, which are necessary for historical park factors and using the data from those times, but are no longer necessary given modern baseball data .
While there are park factors for every kind of hit, I’ll just use HR Park Factor (HRPF) as the example. HRPF uses, in part, Homeruns per Game Played in each stadium, to estimate that stadium’s HRPF. The obvious flaw in this thinking is when you take a team like the 2017 Pirates who hit the 2nd fewest homeruns in the whole league, correspondingly had the second fewest homeruns hit in their home park. This means that the overall HRPF is highly dependent on the on the ability of a home team to hit homeruns, as their going to be the team that is most responsible for the quantity of homeruns hit at their home stadium. In other words, the Homerun Park Factor makes a big assumption that all teams hit the ball more or less the same, and it is differences in the park that cause variation. If you want to make a metric that can be calculated across baseball’s long history, this is an assumption that you have to make; however, in the Statcast era, we do have data on the actual contact made, so there is no longer a need to make that assumption when comparing modern ballparks.
Park factors were largely created in the early 2000’s and were a bit ambitious.
The key metric I’ll be using is Expected Weighted On Base Average (xwOBA), which is a measure created by the MLB to estimate the wOBA value of a ball in play, based on the ball’s Exit Velocity and Launch Angle. So, based on the actual contact made, xwOBA makes a prediction about how much offense that ball will generate. If we take the average xwOBA of homeruns hit in a given park by all lefties we start to get a good idea of how much easier or difficult it is for those lefties to hit a homerun in that park. The smaller the average xwOBA is, the easier it is for a player to hit a homerun in that park.
The thinking goes like this, if it is easier to hit a homerun, the required quality of contact to hit a ball beyond that ballpark’s fence will be less, and thus the average xwOBA for a homerun in that park will be reduced. While there may be a specific hitter or two on a team with a lot of power who might shift that xwOBA up, the easier it is to hit homeruns means more “lower quality” contact will be able to sneak out too, keeping the overall average xwOBA lower.
If we take the difference between the actual wOBA value of a homerun, 2.0, and the average xwOBA on homeruns at each of the 30 parks, then find the percent difference from the average, and put that difference on the Park Factor scale, we get a “new Homerun Park Factor”; we’ll abbreviate it nHPF. We’ll also take the average nHPF at each park over the past two seasons, since 2016 and 2017 have largely marked a new homerun environment.
Left-Handed New Homerun Park Factors 2016-2017
|Park||Total Lefty HRs||Avg Lefty HR xwOBA||Lefty nHRF|
The Pittsburgh Pirates’ nHPF over the past two seasons was 108, so based on the actual contact of homeruns in PNC, it was 8% easier than average for a lefty to hit a homerun at PNC Park. This makes PNC Park the 6th easiest ballpark in the league for a lefty to hit a homer in. This is great news for the Pirates who should have a significant amount of power from the left hand side. This would estimate that they will hit about 8% more homeruns next season than if they played in an “average” stadium. Additionally this just confirms the initial intuition that having the Clemente Wall so close down the line in right field should mean good things for left handed power hitters.
The key takeaway is this; there are lies, damn lies, and statistics. Someone looking for a contrarian position to take, can always parse up, misapply, or misinterpret some metric to bolster their argument. In this case, while Park Factors do have value in giving us some basis to compare most players in baseball’s history, it’s an admittedly weak basis. Since we can now control for the assumptions made in the original Park Factor metrics, we can get a much better sense how easy it is to hit homeruns in different parks. This approach doesn’t rely on the heroic assumption that all teams effectively have the same contact, and thus it is much better metric as it better reflects reality. The resulting conclusion is this; the intuition that the short fence in PNC’s right field is good for lefty power numbers is completely valid. Moreover, sometimes it’s just better to go with the reasonable intuition, rather than using a difficult to grasp statistical inference to be contrarian.