We have moved to our new permanent home at HighHeatStats.com!

All existing content will remain, but all new content is posted at our new location.

Digging deeper: Looking at WPA+ and WPA- instead of just WPA

Posted by Andy
Nyjer Morgan / Icon SMI
Lately we've had numerous discussions about seasonal WPA and what it really means. I think it can certainly be deceptive or at least less than fully informative without more information. Following is a quick-and-dirty breakdown of WPA into WPA+ and WPA- to get a better sense of what the overall number really tells us.

I opted to choose a small set of players with similar seasonal WPA values, and I chose them all from 2011 so they were at least playing in overall similar run-scoring environments (ignoring, of course, park effects.)

Here are all the players in 2011 with a seasonal WPA between 1.90 and 2.10:

1 Nyjer Morgan 2.088 6 429 115 20 6 4 37 19 70 14 15 6 13 .778
2 Todd Helton 2.082 14 491 127 27 0 14 69 59 71 3 0 6 0 .850
3 Matthew Joyce 2.077 22 522 128 32 2 19 75 49 106 4 0 7 13 .825
4 Jose Reyes 2.027 31 586 181 31 16 7 44 43 41 0 2 5 39 .877
5 David Wright 1.967 9 447 99 23 1 14 61 52 97 3 0 5 13 .771
6 Dexter Fowler 1.932 4 563 128 35 15 5 45 68 130 6 7 6 12 .796
Provided by Baseball-Reference.com: View Play Index Tool Used
Generated 12/15/2011.

I included WAR Batting Runs to see how similar they were in this regard, and they are not very similar. Looking at some of the other columns, we can see that a player like Nyjer Morgan contributed with a lot of HBP and sacrifices, which certainly helped his WPA+.

Notice that all of these players had similar (and low) GIDP totals and, with the exception of Morgan, had a similar number of extra-base hits.

Anyway, by going to each players batting pages, I pulled out the WPA+ and WPA- numbers. These two are added together to get the seasonal WPA you see above.

The format here is:

Player   WPA+ WPA- aLI Percentage

Morgan   8.1 -6.0  .94 35.0%
Helton   9.8 -7.8 1.01 25.6%
Joyce   10.4 -8.4  .99 23.8%
Reyes   11.3 -9.3 1.04 21.5%
Wright   9.9 -8.0 1.12 23.8%
Fowler  11.0 -9.1 1.00 20.9%

Percentage is the amount that WPA+ is greater than WPA- (on an absolute value basis).

As you can see, Morgan had the biggest spread, on a ratio basis, between his WPA+ and WPA-. What does this mean? Well, he also had the lowest average leverage index, meaning that on average, the situations in which he hit were less important towards the game outcome than for the rest of these players. But it probably also means that he performed slightly better, in terms of generating a higher ratio of WPA+ to WPA- than these other players (if not, and it was just that his aLI was smaller, then his percentage would be the same as the others and his total WPA would be a little smaller.)

If we look at Morgan's season splits, he had just 71 PAs with runners in scoring position, 16.5% of his total PAs. In those PAs he hit .313/.370/.513, as compared to .304/.357/.421 overall. It seems he reserved a number of his extra-base hits for situations with RISP. His own tOPS+ with RISP was 125.

Morgan had just 73 PAs in high-leverage situations, 17% of his total. Another 32.9% came in medium leverage, and 50.1% came in low leverage. In high leverage situations, his tOPS+ was a whopping 143.

Let's compare these numbers for all of the guys in our group:

Player   PA  %RISP tOPS+RISP  %hLI %mLI %lLI tOPS+hLI

Morgan   429  16.5  125        17.0 32.9 50.1 143
Helton   491  18.1   64        20.8 34.4 44.6  76
Joyce    522  31.6  103        19.5 34.5 46.0 135
Reyes    586  20.0   78        17.1 50.3 32.6  96
Wright   447  28.6   99        24.4 37.4 38.3 115
Fowler   563  21.1  132        17.4 42.8 39.8 156

So Morgan had fewer overall plate appearances and also had the smallest fraction of them coming with runners in scoring position--but he hit just about the best going by each player's tOPS+ in that situation. Morgan also had the smallest fraction of PAs coming in high leverage situation, and also the smallest coming the middle leverage situations. But, he was one of the best hitters in those hLI situations, again going by tOPS+.

Wright had an unusually high fraction of high leverage situations (recall his aLI of 1.14 from the first table.) He hit pretty well in those situations.

Fowler is similar to Morgan--a relatively low fraction of hLI situations but he hit really well.

Joyce is interesting--he had a ridiculously high fraction of PAs w/RISP (nearly double that of Morgan) but they weren't very high leverage. So even though he hit well, he got about the same seasonal WPA as the rest of these guys.

So I think we can conclude that Morgan and Fowler had better seasons that the rest of these guys in terms of producing when in counted, but particularly in the case of Morgan it came in a very small number of such plate appearances. That means he's less likely to produce as well next year given the smaller sample size.

No comments:

Post a Comment