MLB stats: Which ones are relevant to handicapping?

[Beyond the Bets]

Well, man, it's a big week for guest bloggers. First, BTB writes for Deadspin on Monday.

Today, brilliant forum member BetATL writes for BTB. Surely BetATL will be swamped with media and interview requests all afternoon, but I figured we could discuss his post until he returns.

Quite honestly, it's a good one. A breakdown of relevant stats for handicapping MLB and how they should be applied or used.

Read it here.

[Wiseguy]

I thought this was fantastic.

[SkyBlues42]

Agreed. ^ ^

Good stuff, BetATL.

[mount187]

I actually have an argument going on in my head about how much value I give stats like FIP, xFIP, SIERA, etc. in relation to sports betting. I love those stats for discussing how good players are in relation to their peers, but do you really want to use those stats for betting? Does it matter if a pitcher's real ERA should be 3.20 instead of 4.20? It's not like their defense is changing from start to start to lower that number. If the same left side of the infield couldn't get to a ball in the hole five days ago, are they going to be able to get to it now? And, should we even be looking at an earned run metric of any kind? Does it matter if pitchers are giving up earned or unearned runs? If you're a ground ball pitcher, and your infield is terrible, isn't that going to be reflected by the number of total runs you give up? Some question could be asked for a fly ball pitcher with a slow OF.

Also, I don't think you can just look at BABIP in a vaccum; you have to compare it against their career averages to know if they're giving up/getting an abnormally high volume of "fluke" hits. Joey Votto's BABIP from year to year looks ridiculous, but it's been that way for four seasons now. He squares the ball up like crazy, and hits a lot of line drives, which increases his likelihood of getting hits. I think he also went something like 1.5 seasons w/o hitting a popup to the infield; less "easy outs" leads to a higher BABIP.

I think the stats have value, I'm just trying to figure out how to use them as it relates to sports betting.

[BetATL]

Good points Mount.

I often have the same argument myself. Handicapping MLB can be a little overwhelming sometimes because there are so many stats you can use. I think the stats listed in the articles are good ones to keep an eye on throughout the season but are also ones to be used in conjunction with handicapping tactics that take a more macro approach.

[SkyBlues42]

I still think the stats mentioned in the article are useful. Even if you don't use them for team vs. team match ups, you can still use them to find value in totals.

[mount187]

Yeah, like I said, I haven't pieced together exactly how I feel. I know what we're doing in our model has been successful up to this point, but we're still only ~35 days into the season, so the sample size isn't the best. The same model was used last year, and was profitable, but we're using different stat combinations this year, so I'm being a little cautious with my enthusiasm.

Anyways, I liked the article, and I wasn't trying to discredit what you wrote. It was just interesting timing, as I had just been kicking the same ideas around.

[ChiefZilla]

Overall, awesome article BetATL. Well done.

And mount, I've wondered the same thing about guys like Votto. I don't watch many Reds games, but given his decently high K numbers (125+ the last two seasons), he seems like the type of hitter who waits for his pitch, even if it means sometimes you strike out. And when you wait for your pitch, you'll drive the ball more. And when you drive the ball more, I assume you'll have a higher BABIP than guys who aren't as patient.

So if Votto is an exception, aren't there others?

[JSpauny]

(05-08-2012 10:41 AM)ChiefZilla Wrote:  Overall, awesome article BetATL. Well done.

And mount, I've wondered the same thing about guys like Votto. I don't watch many Reds games, but given his decently high K numbers (125+ the last two seasons), he seems like the type of hitter who waits for his pitch, even if it means sometimes you strike out. And when you wait for your pitch, you'll drive the ball more. And when you drive the ball more, I assume you'll have a higher BABIP than guys who aren't as patient.

So if Votto is an exception, aren't there others?

BABIP is a very interesting stat. It absolutely can point to good/bad luck, but as an end all number it's poor. It always needs context.

It's a mistake to compare it purely against the normal league average (not that BetATL said that). Speed guys and LD hitters will have a higher BABIP almost every time. The reasons behind that are simple. Speed guys can leg out a shot to the hole or a slow roller to 3rd that most hitters would be out on. Line drives have been proven to be the best type of hit so hitters who hit more LDs than the average player will naturally have a higher BABIP.

The opposite is true of GB hitters. GB hitters without speed (I had a good example but completely forgot who it was) will naturally have a lower BABIP under regular circumstances.

For Votto in particular, his LD rate in his career has been 24.4% which is pretty high. The last two years it's been 27.5% and 36.2%. Those are pretty insane LD numbers, but since LDs are typically hits and not outs, his BABIP will be high. His BABIP last year was .349 and this year it's .388. Compared to his career of .353, those numbers seem pretty normal, especially if you factor in his high LD%.

That's why BABIP isn't a vacuum stat like mount says. Have to have context behind it. But with context, it's an extremely valuable tool.

[JSpauny]

Let me add that are going to exceptions to the rule everywhere, but generally speed, more LDs, and less GBs should equate to a higher BABIP.

[ChiefZilla]

Are there any exceptions to the rule in regards to pitchers?

i.e. Do groundball pitchers generally have lower BABIPs?

[BetATL]

Also good points J,

With a number like BABIP, I usually try to maintain a macro approach. I try not to get too involved with crunching individual numbers like BABIP because when I do I usually find myself spending hours upon hours doing it and as Mount mentioned, I'm not sure how much return I'm getting for the time I'm putting in. A good working knowledge of what teams and key players might have abnormal variances is what I aim for and if I can pick something deeper up along the way, even better.

I'm liking the discussion on this topic. Lot of good view points.

[JSpauny]

(05-08-2012 12:23 PM)ChiefZilla Wrote:  Are there any exceptions to the rule in regards to pitchers?

i.e. Do groundball pitchers generally have lower BABIPs?

I'm not entirely sure it works that way with pitchers. If you look at the leaderboard in BABIP last year, Hellickson, Weaver and Tomlin were all under .255 for BABIP with a GB% less than 40. Defense is huge for pitchers BABIP. I think it's one of the biggest determinants. Does that shot to the hole become an out or a hit? Same with the dribbler to 3rd. I'd factor in defense before GB%. It's why you'll notice that the Rays had some of the lowest BABIPs in the MLB last year, it's because their D was top notch.

[iKnowNothing]

(05-08-2012 12:23 PM)ChiefZilla Wrote:  Are there any exceptions to the rule in regards to pitchers?

i.e. Do groundball pitchers generally have lower BABIPs?

Generally they have a lower BABIP, ERA, FIP and xFIP. BABIP is strongly correlated to LD% and should rise or fall accordingly. Guys with high GB rates usually have lower stats due to the fact that most ground balls will be outs. Ground balls are also good for double plays (getting out of jams and raising the LOB%) and ground balls never leave the yard as home runs. All put together, ground ball pitchers are usually more successful than fly ball pitchers. A look at the career of Brandon McCarthy is a prime example of this.

However, another thing to think about is strikeouts. The thinking is that once a ball is put in play, a pitcher has no control over the game. Pitchers with high K% (and more specifically a high K/BB showing great stuff plus control) can control the game more due to less balls in play, reducing the randomness of BABIP.

I also wanted to mention what Mount touched on a bit as it's highly useful. Context is key. Just because there is an average BABIP does NOT mean that everyone should be in there. Some guys just consistently outperform expectations or underperform expectations. You should look at guys in respect to their career averages. That is the key mistake many people use with BABIP. Regression to the mean does not imply everyone will regress to league average; it implies everyone will regress to their own personal average.

For pitchers, I like to look at:
ERA vs. FIP and xFIP
BABIP and LD% to career norms
LOB% to career norms
K/BB to career norms

A combination of these will help you quickly understand how the pitcher is doing relative to his past performance and find the regression to the mean.

Sorry for my rambling. I can go into more detail if you ask. I'm pretty fond of sabermetrics.

[MrLomez]

(05-08-2012 09:01 AM)mount187 Wrote:  I actually have an argument going on in my head about how much value I give stats like FIP, xFIP, SIERA, etc. in relation to sports betting. I love those stats for discussing how good players are in relation to their peers, but do you really want to use those stats for betting? Does it matter if a pitcher's real ERA should be 3.20 instead of 4.20? It's not like their defense is changing from start to start to lower that number. If the same left side of the infield couldn't get to a ball in the hole five days ago, are they going to be able to get to it now? And, should we even be looking at an earned run metric of any kind? Does it matter if pitchers are giving up earned or unearned runs? If you're a ground ball pitcher, and your infield is terrible, isn't that going to be reflected by the number of total runs you give up? Some question could be asked for a fly ball pitcher with a slow OF.

My two cents on Defense Independent Pitching (DIPs). In short, yes they are extremely useful for sports betting. They are substantially more accurate in predicting future performance than regular old ERA (the usual caveats about pitchers who routinely defy their DIPs aside).

But, this is only true, if your betting model ALSO takes into account a team's defense. Jeremy Hellickson will always outperform his DIPs, but so long as you factor in that Tampa's defense will prevent x number of runs per game, the discrepancy will be accounted for.

This is especially true when you're dealing with a smaller sample size.

[gynecologist]

where do you find OPS+?

[JSpauny]

(05-08-2012 09:18 PM)gynecologist Wrote:  where do you find OPS+?

baseball-reference.com calculates OPS+ if you go to the team stats pages or individual player pages.

[DSzymborski]

(05-08-2012 12:36 PM)JSpauny Wrote:  

(05-08-2012 12:23 PM)ChiefZilla Wrote:  Are there any exceptions to the rule in regards to pitchers?

i.e. Do groundball pitchers generally have lower BABIPs?

I'm not entirely sure it works that way with pitchers. If you look at the leaderboard in BABIP last year, Hellickson, Weaver and Tomlin were all under .255 for BABIP with a GB% less than 40. Defense is huge for pitchers BABIP. I think it's one of the biggest determinants. Does that shot to the hole become an out or a hit? Same with the dribbler to 3rd. I'd factor in defense before GB%. It's why you'll notice that the Rays had some of the lowest BABIPs in the MLB last year, it's because their D was top notch.

The mean's the biggest determinant.

Essentially, if you know nothing about a pitcher other than his BABIP, the expected break-even point (the point at which his actual BABIP passes mean BABIP as the better predictor of future BABIP) is between 3000 and 3500 batters faced (the limited ability to exceed BABIP doesn't generally seem to age significantly).

Assuming Tom Glavine could theoretically come back and pitch at even a passable level tomorrow, if all you knew was his career BABIP, your best guess at future BABIP, without knowing anything additional, would be in the neighborhood of 85% actual BABIP, 15% mean BABIP.

With Josh Tomlin, that would almost flip, 75% mean BABIP, 25% actual BABIP.

With team defense, that gets better, but not dramatically. BABIP above team still leaves you with a break-even point somewhere in the 2000-2500 batters faced range.

[mount187]

(05-08-2012 04:47 PM)MrLomez Wrote:  But, this is only true, if your betting model ALSO takes into account a team's defense. Jeremy Hellickson will always outperform his DIPs, but so long as you factor in that Tampa's defense will prevent x number of runs per game, the discrepancy will be accounted for.

I should have made this point and didn't. If you're going to use a model that values pitchers based on "fielding independent" stats, then you have to factor in fielding in some other way. Good catch, Mr. Lomez, and thanks for bringing this up.

[gynecologist]

(05-08-2012 09:55 PM)JSpauny Wrote:  

(05-08-2012 09:18 PM)gynecologist Wrote:  where do you find OPS+?

baseball-reference.com calculates OPS+ if you go to the team stats pages or individual player pages.

thanks spauny :)