Matt Cain as an Example in Beating the Spread

The discussion a month ago about Matt Cain and his –and teammates’—home run suppression had me thinking about park factors and the nature of comparing unadjusted stats like RA to ones like xFIP. It feels at times that people will point to the gap between a pitcher’s ERA and his FIP or xFIP as proof that said pitcher is a statistically significant outlier at having control over one of the results of pitching that the DIPS theory rejects. Where I think a disconnect forms is that those assertions ignore that controlling for “luck” factors like BABIP and HR/FB is not the only thing that metrics like FIP do. They also attempt to control out the pitcher’s home park and the defensive skill behind him.

If we want to examine the question of whether a metric like xFIP underrates Matt Cain (or anyone) because it ignores some aspect of Cain’s skill at controlling his batted balls then we need to first isolate that part of the spread between his ERA and his xFIP. That means making corrections for park and defensive skill.

I started off only intending to look at park factors for home runs in this case because a large part of the argument is that Cain (or the Giants’ staff) does better than expected at preventing keeping fly balls in the park. And I wanted to go into greater depth than simply using San Francisco’s park factor on the entirety of Cain’s outcomes so I separated every fly ball Matt Cain gave up by the park it occurred in and whether the batter was hitting from the left or right side at the time. I feel the latter is important because according to my research on handed park factors, parks (San Francisco’s is one) can have vastly different effects on right and left-handed batters. On the whole, Cain’s average fly ball has occurred in a park that decreases HR/FB rate by about 6%. Over his career, that’s about six additional home runs or about 8.5 runs.

That’s not a lot to be true, but that’s only one park factor and only one thing out of a pitcher’s control. Another is the quality of his defense and that is an area that Matt Cain has benefitted from tremendously. From 2006 (Cain’s first full season) through last season, the Giants have totaled a league-leading 193 UZR. The next highest mark is the Cubs at 116 runs. The Giants’ run of defensive excellence the past five seasons might be one of the more under talked about streaks in baseball. Assuming that Cain gets a share of the defensive bonus equal to the proportion of innings he’s pitched, Cain is in debt to his fielders at a total of just under 30 runs.

Adjusting Cain’s runs allowed for things we can all agree on, parks affect home run rates and defenses affect BABIP, raises his RA by about 0.3 runs. Other factors that xFIP glosses over that ERA doesn’t, like opposing hitter quality, shows that the average hitter Matt Cain has faced has a slugging percentage about 10 points lower than the National League average over the same time. A difference that looks to be worth about another 0.1 runs allowed average. It helps to get to face the Padres and Dodgers often.

All together, adjusting for park home run rates, defensive skill and quality of batters faced cuts the gap between Matt Cain’s ERA and his xFIP by about half. True, that still leaves a rather sizable difference of about a half run per nine innings. Identifying the cause of the remaining gap is not my intent here however. Rather, I only wanted to point out that when it comes down to such a debate, it should not be ERA versus xFIP. You can try to argue that xFIP is not valuing Matt Cain fairly because he possess some extra skill that xFIP is not capturing. But you cannot use the gap between ERA and xFIP to prove it.




Print This Post



Matthew Carruth is a software engineer who has been fascinated with baseball statistics since age five. When not dissecting baseball, he is watching hockey or playing soccer.


28 Responses to “Matt Cain as an Example in Beating the Spread”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. MrKnowNothing says:

    *You can’t use the ENTIRE gap to prove it – there’s still apparently a “rather sizable difference of about a half run per nine innings” still unaccounted for. That might be luck, it might be something unique to Cain, it might be XYZ.

    Vote -1 Vote +1

  2. fairweatherfan says:

    But i just read Eno’s analysis and he says the Giants defense is terrible!

    Vote -1 Vote +1

    • merizobeach says:

      Where is it terrible? Torres and Ross/Schierholtz cover ‘huge tracts of land’, and while Burrell doesn’t have that range, he has an adequate, accurate arm and decent glove. Huff (who did some of his best hitting as a LF in 2010 in a SSS) fielded the position adequately as well. Huff was also good at 1B, where is Ishi is also very good, and Belt has acclaim. The rest of the infield, while a little light on range, was also sure-handed and light on errors, and Posey is an excellent catcher. I can’t vouch for Tejada.

      Vote -1 Vote +1

      • Mike says:

        Eno’s preview knocked the Giants defense (like pretty much all of MSM has)

        Vote -1 Vote +1

      • AA says:

        Burrell is an awful defensive player, outside his arm. He’s basically Bobby Abreu, but worse, defensively speaking, only Abreu was actually good at one time.

        Vote -1 Vote +1

    • Jason says:

      Eno’s previews so far have been a little uneven. The Giants are “flawed” but its ok that the Padres can’t score runs because “offense hasn’t really been their stalwart.”

      Vote -1 Vote +1

  3. merizobeach says:

    Maybe the ‘missing data’ is actually indicative of a sort of small sample size: this year will be his age 26 season, meaning he very well may have a four-season physical peak *ahead* of him. When I look at his xFIP, I see what may be a trend–although there may not yet be enough data: after a generally successful debut of 46.1 IP in 2005 with an xFIP of 4.96, Cain amassed 608.1 IP over the next three seasons with an xFIP of 4.55, and notably all three seasons were within the range of 4.52 – 4.61. Since then, over two seasons, he has compiled 441 IP with an xFIP of 4.20, (4.19 – 4.22). The two sets of ranges are remarkably narrow, and if he is trending, it is toward improvement. It seems to me like a reasonably conservative estimate that Cain may post totals that again result in a 4.20 xFIP this year, though he may hit another pinnacle and plateau again in the next couple years.

    Vote -1 Vote +1

  4. Barkey Walker says:

    Obviously, you miss the point when you write, “the average hitter Matt Cain has faced has a slugging percentage about 10 points lower than the National League average over the same time.” In reality, batters who dare face Matt Cain have their slugging percentage suppressed so drastically that their season total goes down by 10 points.

    Vote -1 Vote +1

    • merizobeach says:

      Especially when they just faced Timmy and had Durty coming up.

      Vote -1 Vote +1

      • fredsbank says:

        you took that delightful post and ruined it with your suck. for shame, sir, for shame.

        Vote -1 Vote +1

      • merizobeach says:

        If there is shame in having Lincecum, Sanchez, and Cain at the top of a rotation, or even delighting in it a bit, pardon that it escapes me.

        Vote -1 Vote +1

      • fredsbank says:

        no its just that it was a great thing about matt cain, and you just…tossed those other names in there, when it was all about him. which was lame.

        Vote -1 Vote +1

  5. DrBGiantsfan says:

    I don’t have any problem with what you are saying here, but what you are saying is a whole lot different than warning people not to draft Matt Cain in fantasy baseball because he’s “xFIP shows he’s bound to regress” like some writers are saying.

    Vote -1 Vote +1

    • Nate says:

      Any fantasy player who would strip out park and defense for a player who hasn’t changed teams/parks would be a idiot.

      Vote -1 Vote +1

    • Travis says:

      So you have a problem with fantasy writers giving incorrect advice to potential competitors, when you know better?

      Vote -1 Vote +1

    • Al Dimond says:

      Who cares about fantasy baseball? That is — the post didn’t even mention fantasy baseball or any arguments over drafting Matt Cain in your league. Mr. Carruth is not entering into the “draft Matt Cain or not” debate. This post is about real-world valuation and understanding baseball performance. xFIP isn’t a fantasy predictor, it tries to put pitchers on equal footing. If some fantasy writers don’t understand that it’s because they’re doing what you’re doing: fitting articles and metrics into their world instead of understanding them for what they are.

      Vote -1 Vote +1

    • DrBGiantsfan says:

      It’s not just fantasy. There have been several writers on this site in both fantasy and non-fantasy related articles who are convinced that Matt Cain is going to regress THIS YEAR because, you know, his ERA has outperformed his xFIP for 5 YEARS IN A ROW, and you know, he’s bound to regress some time, isn’t he?

      That has softened a bit as more in-depth analysis points to his performances being sustainable, but there is a still a signficant “Matt Cain is going to regress” sentiment on this site.

      Well, yeah, he’s probably going to have a bad season in here somewhere, most pitchers do, but it’s a lot more likely to be due to injury than to “regression.”

      Vote -1 Vote +1

  6. merizobeach says:

    We have another problem with SSS in the “Matt Cain: ERA vs xFIP” debate when we look at the postseason numbers… A 0.00 ERA over 21.1 IP earned him a less-than-stellar 4.98 xFIP. Does anybody know how to quantify intimidation? Or can anyone diagnose “Orwell’s Boxer the Horse” condition?

    “I will work harder.” Matt “Boxer” Cain

    Vote -1 Vote +1

  7. AJS says:

    Doesn’t looking at the slugging percentage of the guys Cain faces in a vacuum just confound this further? The Padres and Dodgers that you mention play in pitchers’ parks. I don’t think we can say conclusively that the offensive players on those teams are worse than average – maybe the fact that their slugging percentages are kept down by their home parks just makes them look that way.

    Vote -1 Vote +1

  8. DrBGiantsfan says:

    Well, the long awaited regression might have arrived. Cainer will skip his scheduled Spring Training start tomorrow after a MRI showed inflammation of the elbow.

    Vote -1 Vote +1

  9. stigs says:

    There seems to be some double counting going on here. You state (correctly) that Matt Cain benefits from facing the Dodgers and the Padres,noting that their hitters aren’t very good -poor slugging percentage. Well, the Padres, Dodgers, and Giants all have pitcher friendly parks. The park factors weigh down the slugging percentage of the NL west players. When you use both slg and park factors to close the Matt Cain ERA FIP gap, it seems to me that there is double counting going on since the two factors are interrelated. On the other hand, I don’t have any ideas about what could be done to adjust for this.

    Vote -1 Vote +1

  10. Shane Bussmann says:

    Shouldn’t the factors discussed in this post also affect the difference in ERA and xFIP for Tim Lincecum, Jonathan Sanchez, and Barry Zito? If “adjusting for park home run rates, defensive skill and quality of batters faced cuts the gap between Matt Cain’s ERA and his xFIP by about half”, then what affect does applying the same corrections to these other Giants pitchers have?

    Vote -1 Vote +1

  11. Nathaniel Dawson says:

    It’s also good to keep in mind that Matt Cain, being an extreme flyball pitcher, gives up fewer unearned runs than other pitchers. For his career, he’s given up .17 unearned runs per 9 innings, while the average is usually around .35. If you scale xFIP to R/9 and use your adjustment for defense and park, he’d still be about .7 runs under for his career. Using that adjustment cuts the difference between his R/9 and xFIP down about a third.

    It does help explain some of the variance, but there’s still a lot left. I’d have to think part of the lot left is xFIP’s tendency to over-estimate runs allowed for flyball pitchers. That might account for another three-tenths of a run, getting the variance somewhere down around .4 runs per 9. That’s still quite a lot for over a thousand innings, but it’s a lot less than the starting point of 1.16 (Cain’s xFIP scaled to R/9 minus his R/9). What’s left could be random variation (chance) and/or his ability to control HR/FB.

    Vote -1 Vote +1

    • evo34 says:

      The only thing is if you start taking into account his flyball tendencies, you would have to say that he benefits less from his team’s good defense…so his xFIP/ERA divergence becomes slightly less explainable.

      Vote -1 Vote +1

      • DrBGiantsfan says:

        Not sure why you would say this as the Giants OF D has been the biggest part of their overall defense the last several years. I have no problem saying part of the reason Matt Cain is so good is he has a great defense behind him. It is a team sport, after all. Again, there is no reason to think the Giants defense is going to suddenly tank this year.

        Vote -1 Vote +1

  12. evo34 says:

    This article tells me that we need something in addition to xFIP that assumes same park/defense, so we can see what portion of the ERA/xFIP discrepancy is theoretically due to luck, vs. due to current environment. Right now, it’s kind of a mess, since xFIP excludes current defense, HR park factor, and HR ability of opponent, but includes park effects on BB and K, and BB/K skill of opponents. This leads to a messy combination of raw/adjusted stats.

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>