Monday, October 03, 2005

Playoff Stats Primer

Before the actual previews a brief (OK, lengthy) description of my methods. When evaluating teams last year, I relied mostly on VORP, which is a cumulative metric based on value measured in runs. It's a good tool for evaluating a player's contribution over the course of a season because it accounts for positional value and durability. When it comes to the playoffs, however, those factors don't matter as much, if at all, because of the small sample of games that we're dealing with--at the most, 19 games, which doesn't tell you anything about how good a team actually is. I decided to focus on rate stats this year, at least on the hitting side. One of the best rate metrics is EqA (explained here), not only because it has a higher correlation with runs scored than any other rate stat, but also because it is intuitively easy to understand. A batter with a .260 EqA is exactly league average. EqAs over .400 equate with the best 15 or so seasons ever. Below .200 is just like Mendoza territory. EqA takes baserunning into account as well, giving a full perspective of his offensive abilities and converting it to a scale that can be read just like that old familiar stat, batting average, allowing for easy comparison. So when you see a number after a guy's name, that's his EqA.

A player's defense is notoriously hard to evaluate because it is greatly affected by the other players on his team. A shortstop and second baseman can help or hurt each other when turning a double play, or a top notch first baseman can scoop balls out of the dirt that otherwise would have been errors. Individual defensive metrics (such as FRAR or FRAA) should not be heavily emphasized. However, when you lay out a team's defensive alignment and compare their FRAA, it can be instructive. For example, if there's a negative FRAA guy (aka, below average) on a team full of positives, you can be sure he's pretty bad. Even with all of his teammates with good gloves couldn't help him out. Same goes for the converse situation. To clarify, looking at a complete layout of FRAA allows you to pick out a team's weak spots relative to itself. Defensive efficiency, or the percentage of balls in play turned into outs, says more about the overall quality of a defense and is best used when comparing one team to another.

I moved away from pitcher VORP this year and looked at Support Neutral Value Added. Support neutral means that the pitcher's value is evaluated outside the context of both his defense and his run support. I specifically looked at SNLVAR, which also adjusts for the quality of opposition. It's a great way for quickly determining how good a pitcher is compared to other pitchers. It separates the great (Roger Clemens and his 9.4 SNLVAR was the best in baseball) from the merely good (Barry Zito's 5.5 SNLVAR was good for 20th). For arbitrary purposes, let's say that a SNLVAR above 6.5 qualifies as an "ace" starter--using that cutoff there were 12 aces in baseball this year. Sounds about right to me.

For similar reasons, I chose to use Reliever Expected Wins Added (among other things) to evaluate relief pitchers. The great thing about WXRL is that it takes into account the importance of situational pitching. Bases loaded with 1 out in the 8th in a tie game is always more important than bases empty with 2 outs in the 9th when leading by 3 runs. WXRL gives more credit to pitchers who perform well in the former situation than it does to pitchers who perform well in the latter. WXRL shows that on some teams, the best pitcher is not always the closer, and that the guy who got 1-2-3 in the 7thg is more deserving of recognition than the guy who gives up 2 runs and still gets credit for a save. WXRL also reveals a manager's usage patterns--guys with high WXRL are the guys a manager has turned to in a tight spot. Best of all, WXRL takes the emphasis off ERA when evaluating relievers. Small sample size issues abound with reliever ERA, and ERA does not factor in high leverage situations. Pitcher A has an ERA of 3.00 but most of his allowed runs result in his team losing the lead. Pitcher B has a 3.75 ERA but gave up most of his runs in blowouts where his team was already losing, while performing well in high pressure situations in close games. Wouldn't you rather have the pitcher with the higher ERA? I sure would.

Hopefully all of this will add up to better predictions, although I admit that predicting the results of a 5 game series is rather silly. Most predictions try to determine the outcome of each individual game. As last year's playoffs showed you, that is a nearly impossible task. My predictions will reflect which team is better in an Eastern thought, holistic sense, with the margin of victory reflecting the degree of difference. I won't try to say who will win, but I'll certainly try to say who should.

0 Comments:

Post a Comment

<< Home