The folly of advanced stats

this fanpost makes me feel like I’m reading about the TimeCube.

Oh I expected this would ruffle your feathers.

I’ll respond to your tripe tonight but the Delmon reference is topical regarding the proposed trade and not used to refute advanced scouts the article referenced does it just fine.

Any time a person makes a conclusion based upon one stat, that’s a mistake. Likewise, any time an untrained (and well-trained, in many cases) eye is making conclusions based upon a small sample and then trying to find an accurate comparison between players of the same position league-wide, that’s a mistake.

In order to make qualified decisions regarding statistics, you have to look at the full picture. When you look at BABIP, you have to look at the batted ball data. When you look at WAR, you have to look at UZR. When looking at UZR, you have to compare it to DRS and Total Zone. When looking at a slash line, you should also check out walk rates and ISO.

Unless you commit to a comprehensive evaluation, there’s no one statistic that can conclusively tell you much of anything, other than what has happened in the actual ballgames. In order to make proper conclusions, you should also include as much visual data as well. It all helps.

I think the problems don’t lie with the statistics…they’re doing their job. The problem is that a wide majority of people do not understand how to properly evaluate them comprehensively, which leads to some misinformed opinions.

I also disagree with your assessment that BABIP is an indicator of luck. In certain cases, I suppose it can be. I think it’s more of a quick-and-dirty way to tell whether a batter is hitting the ball effectively, rather than a luck determinant. By knowing the values at which batted ball types typically fall for hits (which, as stated in your piece, is a conclusion based upon thousands upon thousands of PAs), you can tell whether the presence of luck may be affecting the outcome, or not.

Many people merely look at the BABIP and make an evaluation without delving into this area, which is yet another fundamental mistake that I spoke of.

-C

I hate advanced stats . People can argue that this guy is good because of this stat and someone else will argue that he’s not because of a different stat . It’s enough to give someone a HEADACHE .

I guess I am just old school . But if a guy has a decent batting average and a good OPS and doesn’t kill himself with a lot of errors in the field , then he is probably a good enough player to have on the field .

So much is made anymore of how good a player is due to all these advanced stats . And they are getting so detailed that they are really hard to keep up with and sometimes even understand .

Maybe I’m just lazy . But I’ll rely on my own eyes and how he plays and produces to jufge how good of a ballplayer he is . In my opinion only , of course . I’m no proffesional .

That is, a guy who really sucks on defense in a corner outfield spot, and is a below league average hitter. He doesn’t walk, doesn’t get on base enough, and the only thing he offers to compensate for that is the possibility that he MIGHT hit 20 home runs. Surely, somewhere, there’s an article about Fregosi making an assessment of Edgar Renteria, before we traded for him, or screw it, SOMETHING where he was right?

I mean, scouting is extremely important, especially for players who are not in the major leagues. There’s no conclusive analysis of even minor league stats, which fair to be a relatively stable level of competition, with well known park dimensions and league trends. And for all we know, having a .400 BABIP in AA MIGHT be a representation of skill, even though we feel confident that a regression will come for a major league hitter.

But to choose Jim Fregosi saying that he’d take Delmon Young, who isn’t even as good a player as Jeff Francoeur (who sucked so much that nobody at all wanted him, twice) seems to unprove his point.

Is this guy intending to be an obvious troll?

is absolutely ridiculous. I don’t think any advance scout would say that about a hitter’s ability in three games — and players have much, much more control over their skill set hitting as opposed to fielding. What if an outfielder is playing before a scout for a few days and the pitchers are mostly groundball pitchers? How could you possibly make an accurate presumption?

UZR and DRS try to objectively gauge a player’s fielding skills. They can be very useful in tandem with one another.

Stopped reading at “WAR has emerged as the most accepted advanced stat metric.” Complete nonsense. Even the creators of it and those at FanGraphs know that is not even close to the truth.

Well i will defend the post, because its fun to do shit that flies in the face of what all other reasonable opinions have thus far supported.

He is completely right that An unsophisticated forecaster uses statistics as a drunken man uses lamp-posts – for support rather than for illumination. He goes to far in saying stats are useless, but he recanted in the comments.

The truth is there are a lot of people who run over to Fangraphs see that player X has a OBP of 370 and then think he is great. they completely ignore other indicators which suggest he isn’t that good (for example an BABIP of 450).

It has been stated you must look at the whole profile of statistical analysis, not just cherry pick stats. Additionally, you need to look at the player and understand the human elements that are effected him which could possibly explain why his numbers aren’t as pretty.

Before i read the sequel to this post, the folly of the eye test, the problem with the eye test is that there are approximately 2250 games a season and only one of you. which means you can’t possibly see every game. even if we were to assume you watched every game, human beings are not well suited for observation because of various physiological and psychological factors. we tend to overlook certain things and cling, sometimes irrationally, to other events. a good example of this would be Greg Norton. everyone (or at least a supermajority) hated Greg Norton in his last season because of how poorly he played (and he was bad), it just so happens that everyone forgot how great he was the year before and had no context for the fact that he essentially went through a bad stretch, but as a pinch hitter his bad streak last a whole year.

UZR in particular is the kind of metric that requires multiple years of play time to really paint an accurate picture of a player’s defensive performance. I understand the importance of scouting in that case.

But other than that, I have to straight up disagree with everything else. Your arguments against WAR and BABIP are trivial at best.

that advanced stats are far from a science…at least it’s not a physical science where you do need to have something that’s a few standard deviations (4 to 5 at least) from the mean before the community would accept it as a discovery rather than just a random fluctuation (I do think the social “sciences” may be more lenient but they are “science” in name only)

However, sports stats can only be studied in the same ways and methods of the so called social “scientists” (economists, etc.). Based on their studies, advanced stats have BETTER CORRELATION than the traditional stats to the success of the the team and future performance. There should be no denying that.

The key point is that in more specific cases, where most fans would like to make a thoughtful argument and thus look toward the advanced stats, one almost always run into the problem of small sample sizes or having to over generalize (in order to increase the sample size). That is why even the best laid argument over at Fangraphs have to be taken with a grain of salts. As more and more people try to dwell into sabremetrics with various previous statistical understandings, there are many coincidences that have been interpreted as trends. This is not so much the “folly of the advanced stats” but the “folly of the advanced stats user”.