Thursday, 5 September 2013

W - L for Hitters

Baseball fans learn to understand win loss records pretty quickly. At least until they become the dreaded sabermetricians who doubt everything they ever knew about win loss records.

But, win-loss records are still a thing, and given a win loss record you probably have an 80% chance of understanding what kind of season that pitcher had. When I was a kid, Jack Morris came to play for the hometown blue jays and WON 20 GAMES. Well, he was worth 2.9 WAR that year, and his 4.04 ERA was not special. So that is a misleading example. I have already written about Cliff Lee and how much his 2012 6-9 record means.

Anyways, yesterday I decided to turn RE24, which can transform into REW, into an equivalent win loss record that depends a little less on how the pitcher's team does in individual games and a little more about how the pitcher actually, you know, pitches. I can't go into what RE24 means again, so please go read that one first.

Today I realized that all players have RE24 (base our runs added). For each change of base/out state, equal and opposite RE24 are added or subtracted from the pitcher and hitter. If I can turn starting pitching REW (base out wins added) into a pitching win loss record, then why can't I turn hitting REW into a similar win loss record? I think there are at least two good reasons to try this:

1. Pitching is no more than half the game in most cases - each team spends about half the time hitting, and pitching isn't pitching without defensive support. For starting pitching, it has become normal to throw 6 or 7 innings in a start, and consider that a good start. Fine. But then 2/3 of half the game is... 1/3? A starting pitcher may influence the game more than other players, but it's only tradition for a pitcher to get a win.

2. Remember back in 2011 when Justin Verlander was in the middle of being really awesome and winning the Cy Young and MVP award? Well, one of the arguments went like this: JV pitches 250 innings, and each inning has more than 3 batters, so he is influencing more than 750 plate appearances - probably more like 1000 plate appearances - more than an everyday player. So why can't his impact be as large? Touche. However, in the context of RE24, there are some subtleties I would like to bring up.

Using 1993-2010 data from Tom Tango, let's go through a few examples. Each time a pitcher comes into the game, the run expectancy is 0.544.

1. When the pitcher gets a three-up-three-down inning, the pitcher gets +0.544 and the hitters get -0.544.

2. Now, suppose the pitcher loads the bases with no outs. The RE is now 2.390. The pitcher has gone up from 0.544, so has accumulated -1.846. The batters have +1.846.
a. The next batter hits a home run. 4 runs score, and the run expectancy goes down to 0.544 again. The pitcher has -4.00 so far, and the hitters have +4.00.
b. The next three hitters get out without scoring a run. The pitcher goes up 2.390 now, bringing the total to +0.544 for the pitcher and +3.456 for the hitter.

If you're with me, then you just had that feeling that may have happened when you were doing algebra homework as a kid, and suddenly everything cancelled each other out and you were left with x = 4. It's not quite that simple, I think because of errors and other stuff maybe? I'm still not sure who gets blamed for stolen bases.

Anyways, the point remains that in most cases, the best a pitcher can do is to get through the inning without allowing any runs. That's good for about +0.544 RE24.

For a batter, it's a different story. Grand slam with two outs? That's +3.186 in a single at bat. Solo shot with out out? That's worth +0.709. Double to score both of the men on first and second with no outs? That's worth +1.614. Batters get more opportunities to fail, but they can also put up numbers more quickly.

I think both of these are reasons to actually do this comparison, because I want to see what batters are worth in a pitching W-L record context. Of course, this leaves out defence and some other stuff, so Miguel Cabrera is going to be ahead of Mike Trout in this stat.

Now for the mechanics. I decided that plate appearances and innings pitched are similar, and gave them the relationship:

IP = (PA - 200) * 0.5

This makes a 600 PA season equivalent to 200 IP.
A 750 plate appearance season is equivalent to 275 IP.
At or below 200 PA, this doesn't make any sense.

Then it follows the same structure as pitching Wx and Lx:

Wx = (IP / 9.0 IPS / 2) + REW
Lx = (IP / 9.0 IPS / 2) - REW

I will post a bunch of tables of stats including Wx - Lx records and "IP" equivalent for batters. I only have included players with at least 300 PA.

Let's start with some 2013 stats. For perspective, Clayton Kershaw leads MLB with a 16-7 Wx-Lx record. Here are the current MLB leaders in batting REW:

Here are the middle 20, the .500 starters:

And here are the worst 20:

Here are the top 20 REW seasons since 1950. It is quite evident that in the early 2000s, Barry Bonds broke baseball. He went 23-0 twice and become the only player to go over 10 REW in a season. But remember how great skinny Barry Bonds was! His 1993 and 1996 seasons made this list. Of course, this stat only goes back to about 1950, so we miss all of Babe Ruth and the best of Ted Williams, Ty Cobb, Rogers Hornsby, Lou Gehrig, Jimmie Foxx, and others.

Here are the worst 20 REW seasons since 1950. It really is amazing that Neifi Perez was allowed to reach 585 plate appearances in 2002.

Here are the 20 seasons with the most plate appearances since 1950, the "highest workload" seasons. I find it amazing that 4 of the top six seasons occurred in only two seasons, 1975 and 2007.

I think that the next part of this should be in its own post, so I will do that instead. Stay tuned!

No comments:

Post a Comment