clock menu more-arrow no yes mobile

Filed under:

Roy Oswalt Gets the Loss: Why Wins and Losses Don't Properly Evaluate a Pitcher's Performance

Aside from some first inning bad luck and a second inning HR to Brian Schneider, Roy Oswalt was dominant. He looked like Roy Oswalt, not that chump we saw at the beginning of the season. However, he got the Roger Clemens treatment, which I guess we can write off to facing one of the best pitchers in baseball (who, ironically, has been the subject of a lot of criticism because of his meaningless W-L record, this must be God's will). The question is though: Did Roy Oswalt lose this game?

There are a number of tools we can employ to discern where the burden of the loss falls and while I'm sure most of you can easily jump to the conclusion "No, he didn't, please stop you're boring me," I'm going to proceed nonetheless for the simple fact that I have a soap box. Suckers. Although I guess you could just stop reading.

First, and possibly the most novel to me, is the Bill James Game Score. For those of you old enough to possible be my father, and even my father too (though this might not be up on Monday when he checks it, but "Hi, Dad."), might have actually read this in a Baseball Abstract, but I didn't stumble across this until this summer. I like it because it's simple, but very telling, and I think it accounts for a variety of different factors adequately. How it works is (and I'm lazy so I'm just going to quote this from another website -- hopefully it caught your attention and compelled to keep reading this drivel) is

Start with 50 points. Add 1 point for each out recorded, (3 points per inning). Add 2 points for each inning completed after the 4th. Add 1 point for each strikeout. Subtract 2 points for each hit allowed. Subtract 4 points for each earned run allowed. Subtract 2 points for each unearned run allowed. Subtract 1 point for each walk.

Like I said, it's simple, but somewhat telling and useful. Roy's start last night scored an 68. If you're like me, you said, "not to shabby." I don't have my copy of Will Carrol's Saving the Pitcher on hand, but if I recall, that's a pretty good score although, it is a ways behind the best score ever tallied of 105. Don't worry though, it was by some guy named Kerry Wood -- apparently the dude struck out 20, gave up a single to some no namer, and then managed to hit a guy who was a magnet for HBP's (this was all against some team that was supposed to win the World Series, but broke their hometown's collective heart). My digressions aside, what Roy's score of 68 last night tells us, is that he pitched a good game.

One novel measure down, my favorite to go. is easily one of the greatest sites out there on the net (just behind this site,, and For each game, it tracks the probability of each team winning the game based on a comprehensive model of probability devised by Tom Tango (Tango Tiger). They look something like this:



Looking at this, we see that once Roy gave up the 2 run HR, things just got worse for the Astros. However, we can also tell, thanks to Fan Graphs, how each event in the game marginally contributed to the result of a win or a loss. The sum total of every event always works out to 1. This is convenient, because if you'll note, to start each game both teams are assigned a 50/50 shot a victory. The totals of each player's "event" can be seen here. For those of you still reading, but too lazy to click, I'll save you the click. Roy's impact on the game, as a pitcher, was -7.5% (i.e. his efforts as a pitcher "lost" 7.5% of our share 50% share of the win). Our offense, however, lost the winby (can anyone guess it?) 42.5%. That's correct, our lack of any clutch hitting resulted in a loss at 42.5% of our win. Given that as a team, the Astros, only possessed 50% of a win to start the game, they really contributed to an astounding 85% of a loss with Roy chipping in the other 15%.

A few remarks on WPA and I'll stop boring the crap out of you. LIke ERA or Wins and Losses, it does a poor job of separating the pitching from the defense. So it alone cannot be used to say that offense or pitching lost a single game. It can, however, suggest an answer which needs to be validated with evidence. I chose to throw out Bill James' Game Score. From that, or from watching the game, we can deduce that if Roy Oswalt threw a 4 Hit, 6K to no BB, 8IP game, then it's really hard to hang the loss on Oswalt's shoulders. The best we can do, is give him 15% of the credit and the focus our ire on the bats. For this reason, among a whole host of very rational and credible arguments, I think it's safe to say, assigning pitchers wins and losses is arbitrary and useless, because they have only a small stake in determining the outcome of a baseball game.

And with that, I'll shut up (you can add your "AMEN"s to the comments section, but feel free to simply click and let me know how you feel: