How Accurate Are The Stats?

[Flip217]

One great benefit I've seen of Orioles Hangout is how I've become much more educated on, and comfortable with, the various measures and statistics that are employed to figure out exactly why we suck so bad.

But, as I'm still pretty green when it comes to the field of statistics, I'm wondering if somebody can tell me which are the really good stats, by which I mean, which statistics are the most reliable. I don't mean to start some sort of meta-analysis discussion examing the evaluation of the evaluations, but I figure some stats must be better than others.

If RBIs and Win-Loss records are at one end of the spectrum -- the "not-so-useful / accurate end", as I understand it -- what would be at the other end? OPS? WHIP? Some other acronym I'm wholly unfamilair with?

Thanks in advance for increasing my meagre understanding a bit.

[bgfield]

Well there are lots of stats some people can give you that are very telling, if you care to get into all of them (stats like WARP or RC/27 or whatever). If you want to limit it to what you can find on ESPN.com's stat pages, OPS is a good indicator for batters (.750 is typically an average OPS). ERA is still pretty good for pitchers, though that can be affected one way or the other by the defense behind the pitcher. K/9 and K/BB ratios are also important stats for pitcher, though not as important if you're an extreme ground ball pitcher like Chien-Ming Wang. Speaking of, GB/FB ratios are also good stats for pitchers. The more ground balls a pitcher gets, the better. The more fly balls a pitcher gives up, the more that are likely to turn into homers, etc.

Other people can get a LOT more in depth than what I've just done, but if you just want to know slightly more than your casual fan, the stats I gave you oughtta help.

EDIT: Oh yeah, WHIP is also a good one for pitchers.

[Mashed Potatoes]

Check out the Super Stat Sticky at the top of this forum, lots of good stuff there.

[Migrant Redbird]

I figure some stats must be better than others.

Of course they are, but good luck on getting any kind of a consensus.

The traditional stats (BA, OBP, SLG, ERA, Fielding percentage, etc.) have the advantage of being completely objective. There was no analysis involved in developing the stat; you merely tally up the relevant events and perform a simple calculation. A 3rd grader could do it.

The more "modern" stats are what I like to refer to as "complex stats", in that someone has taken some of the traditional stats and combined them together in a way that imposes a subjective judgment as to the relevant importance of each. OPS and WHIP are good examples. They are among the simplest of the "complex stats", and have the virtue of being easily calculated and comprehended. It's trivial to add OBP and SLG together to produce OPS, which makes it perhaps my favorite stat of all. I like OPS because I can very easily determine how the two components affected it, even if few people agree with the implicit equality of OBP and SLG that comes with OPS. It's still easy to figure out and understand.

But OPS does not take into account things like base running, defense, "giving oneself up for the team", the relative impacts of different ways of making an out, or the effects of differing ballparks and competition. "Sabermetric" stats attempt to combine a number of the traditional stats in a way which provides a more balanced assessment of a particular player's performance. Not everyone agrees upon the best way to do that, and so you end up with a number of competing complex stats, each of whom have their own proponents.

I appreciate that the more dedicated baseball statisticians are making the effort to provide us with better tools to use in evaluating players, but I'm not yet convinced which of those are the best. Something like Win Shares is nice because it takes into account offense, defense, and pitching, allowing you to compare the relative values of any players in the league, regardless of whether they play the same position. However, I've never taken the time to look at how Win Shares are computed and, even if I had, there's no way that I would be convinced that Win Shares weights everything optimumly.

I try to stay out of the arguments over which stat is best. I appreciate how they help to compare different players, but I tend to take them with a grain of salt. For example, I don't accept that Brian Roberts, Derek Jeter, Torii Hunter, and Mike Lowell had equal performances in 2007 simply because their Win Share totals were equal, nor do I accept that they were all better than Placido Polanco or B.J. Upton simply because the latter two players scored one Win Share less. I don't believe that Win Shares are accurate to that level of granularity. They're informative, but not determinative.

If 2 players are 10 Win Shares apart, it's pretty safe to say that the one with the higher score is a more valuable player, but I could probably make that judgment on my own by looking at their traditional stats, without resorting to Win Shares. Where I think that Win Shares probably help most is when trying to compare the careers of a couple players. They might have been too close together during most of their careers to easily differentiate between them, but a look at their career Win Shares may help to illuminate the comparison.

[SteveA]

As someone else pointed out, the Super Stat Sticky is a good place to look.

It also depends on what you are trying to evaluate. We all know that RBIs is not a good stat to evaluate the talent of a hitter. On the other hand, the purpose of the game is to score runs, so if you want to, after the fact, say who had a better season, then RBIs are important. If you want to predict how he will do next year, it is an awful stat.

It's been shown that a batter's BABIP (batting average on balls in play) cannot be consistently higher or consistently lower than the league average for a long period of time. In other words, there is a significant luck factor involved in it. So if a batter has a high BABIP you might conclude that he will regress in the future; if he has a low BABIP he has been unlucky and will likely get better. So that stat has high PREDICTIVE value.

But if trying to decide who deserves the MVP, or Hall of Fame, you don't care about BABIP. Luck or not, he got the darn hits and he deserves credit for them. So BABIP is useless in that context.

[Flip217]

I think you guys have raised my IQ in a mere 4-post thread.

Much thanks to all!

[RShack]

I think you guys have raised my IQ in a mere 4-post thread.

Much thanks to all!

It is very interesting. The hard part is dealing with all the different ways to look at things without getting lost in them. In the end, some folks just pick a given single stat and go with it, just because it's easy to do.

IMO, these days OPS is treated like people used to treat BA: the one thing they look at. Personally, I think OBP as the thing people should have meant back when everybody just talked about BA. It's clearly more informative than BA. OPS is more informative still. But IMO, OPS also stacks the deck against little singles-hitters in a way that I don't always agree with. That's what's happening when people talk about somebody hitting an "empty .300" (or whatever the number is). I think it depends on how a guy fits in with the rest of the lineup. I think that somebody with an "empty" (but high) BA/OBP is a great guy to have if there are good hitters coming up right after him. This is a normal idea that many people share. So, I think context matters, and that you should look beyond just OPS. Some folks do that, no question about it, but some other folks just look at OPS as if it's all they need to know. I don't think it is, but then others think I'm naive about it, so be your own judge.

I think the main problem with stats is that baseball teams and baseball games involve so many interdependencies. In general, these are way to complex to have good models of. IMO, the inability of stats to model these things is what gets missed a lot. I think stats are great tools, but I think it's also important to be mindful of all the stuff that stats just can't see. It's not the fault of stats, it's the fault of how complex baseball is. Baseball is the real thing, and stats are the secondary thing. They simplify and clarify, but they also lose important things too, mainly because they can't see them. It's a trade-off.