In order to answer this question I looked at all starting pitchers that threw at least 100 innings for a given team from 1935-2014 and put them into quartiles based on FIP. Then I looked at all teams from 1935-2014 and put them into quartiles based on their Fangraphs fielding metric.
The first chart shows fielder and pitcher performance order by fielding ratings. E/F stands for ERA/FIP while RA9/F stands for RA9/FIP. I include these stats to measure the impact of defense based on a proportion rather than just a total. Here it is below.
This chart shows that the value of having a good defense decreases when you have good starters. For good fielding teams the difference between ERA and FIP is greatest for the worst pitchers. For poor fielding teams this is still the case but the difference is minimal. It is still better to have excellent pitching than having average pitching but teams have limited resources and may not be able to easily afford top pitching.
The second chart shows pitcher and fielder performance ordered by pitcher ratings.
This chart shows that good pitchers (those in the second quartile) that have the best fielders prevent more runs than the best pitchers with the worst fielders. Bad pitchers that have the best fielders prevent as many runs as good pitchers with bad fielders. Bad pitchers with the best fielders behind them give up only 3 more runs per 200 innings than good pitchers with good fielding. The best pitchers do prevent more runs on average than worse pitchers but good fielding is able to significantly limit the gap.
The chart also shows that even the best fielders are unable to do much for the worst pitchers. It is true that the worst pitchers with the best fielders have an RA9 nearly .7 points lower than the worst pitchers with the worst fielders. It’s also the case that teams with the worst pitchers and best fielders give up 4.7 runs on average per 9 innings and are still ineffective.
This next chart shows how it worked in practice for the 2014 Orioles.
The chart shows that the Orioles have seen this happen first hand. The only starter we had that had an above average FIP was Gausman and he was the only starter who underperformed his FIP. The other starters had an FIP that was either in the third of fourth quartile and mostly outperformed what their FIP suggested. One could argue that the Orioles actually did have a rotation of below-average starters and were able to succeed due to superior fielding. It’s not that Chris Tillman or Wei-Yin Chen are good necessarily but that they have an excellent defense supporting them.
Miguel Gonzalez outperformed his FIP by a substantial amount. There have been 8,487 starters that have thrown 100 innings for one team in a given year from 1935 to 2014. In 2014, Miguel Gonzalez has the tenth highest E-F and the fifth highest RA9-F. Players with similar seasons include Jorge Sosa in 2005 and J.A Happ in 2009. It seems likely that Miguel Gonzalez should expect significant regression if used as a starter next year.
It appears reasonable to argue that the Orioles have been able to prevent runs because they have a large number of decent starters that are able to succeed because they have a strong defense behind them. They may not have the best pitchers in MLB but good defense can fix a lot of problems.
In order to continue being successful, the Orioles need to consider pulling a trick out of the Rays playbook. The Rays typically offer pitching prospects long term deals before they are proven in order to save money and then trade them a few years before they become free agents. Of the Rays 10 starters that have thrown the most innings for the Rays from 2005-2014, they traded seven, had two get injured and currently control only Alex Cobb. In contrast, the Rays only traded four of their ten offensive players with the most PAs from 2005 to 2014. Three of those four were traded this offseason in order to save money.
The Rays were able to do this because their fielding has been ranked #1 by Fangraphs from 2008 to 2014. Having good fielders supporting their pitchers allows them to risk giving guaranteed money to unproven pitchers and risk trading their proven pitchers to other teams. They weren’t that aggressive when it came to position players because they needed to focus on offense and fielding in order to win.
The Orioles should consider a similar strategy. Signing Gausman and especially Bundy to long-term contracts later this year will save the club money if they are successful. Trading a starter like Wei-Yin Chen or Chris Tillman could potentially net a reasonable return while solving our rotation questions.
Excellent fielding is more helpful for decent pitching than excellent pitching. That means that teams shouldn’t focus on trying to find the best fielders and starters at the cost of offense. Spending money in free agency to improve pitching is a luxury that a team with limited resources simply can’t afford. The best plan is to use most available resources for offense and fielding.
Hasn't Gonzales outpitched his FIP every year? If a pitcher does so consistently, whether because of fielding or not, why is there reason to expect a regression?
ReplyDeleteThere are two reasons.
ReplyDeleteIt's not whether Gonzo can outperform what his FIP predicts. It's whether Gonzo can continue doing it by the extent that he did. Historically, his RA9-FIP was the 5th largest ever in 2014 and the 1421st largest ever in 2013. If he outperforms his FIP by the same extent as he did in 2013 then he would be expected to give up at least another run per nine innings.
If starters could outperform their FIP by a large extent consistently then we'd see it in the data. The two starters that have outperformed their FIP by the most (five year minimum) are Jim Palmer and Jered Weaver at .37 R/9 and .31 R/9.
If Gonzo is similar to those two guys then he should have allowed 4.5 R/9 last year instead of a 3.38 R/9.
Yes, you would still expect a regression. When you consider players in a historical context, Gonzalez is exceptionally unique. It is as if you flipped 10 heads in a row. You expect the coin flip results to even out based on what you know rather comprehensively about coins.
ReplyDeleteAt one level, reducing runs should not be a focus of good-pitching teams should be pretty intuitive. Fielders affect the rate at which events turn into hits. So, if you reduce Kershaw's ERA by 10%, that is less than reducing some AAAA scrub's ERA by 10%.
ReplyDeleteThese first approximations don't really work because even the Pythagorean theorem is not linear, and even it is only an approximation of final results.
Still, the problem is knowing what defense is. It is one of the hardest things to do. It is also variable with ballpark.
On the other hand, increased batting linearly increases runs, approximately on a batter-by-batter basis. If you have the Washington Nationals' pitching staff, do you really need Ozzie Smith? Or would you rather have an equal quality run-scoring shortstop? Probably the latter. Thus, Ian Desmond!
I wasn't expecting a linear relationship. Since most hits are valueless by themselves (unless they're followed by another hit or a sacrifice) I would have thought that reducing the number of hits in a game would have a larger percentage impact on a good pitcher while having a larger absolute impact on a bad pitcher.
ReplyDeleteI mean suppose a good pitcher only gives up 1 hit per game. In theory, he should have a strand rate flirting with 100%, no?
But it doesn't seem to work that way. Unless I simply used overly large categories.
The main reason fielding impacts good pitchers less is because of strikeouts - there are fewer balls put in play.
ReplyDelete