Win Values: Updated for 1969, 1974-1977 - Rob Wood (January 6, 2004)
One of the gems that has been underdiscussed, in my view, is Rob Wood's Win Values. It's a rather fascinating way to determine the impact of a starting pitcher's performance. The reason I'm bringing this forward is the case of Bert Blyleven. By this measure, his team did not underachieve, relative to his overall performance.
Here's a cut/paste of the relevant information from Rob's original article, that discusses his fascinating methodology:
==========
Consider a team that scores 7 runs in a 9-inning game. Suppose I find a team that scores 7 runs in a game will have scored 5 runs at the conclusion of the 6th inning 10% of the time. I also know the distribution of final scores of every team that scored 5 runs at the conclusion of the 6th inning. Say they wind up with 5 runs 12% of the time, 6 runs 20% of the time, 7 runs 25% of the time, …, and 15 runs 1% of the time.
I can calculate this bootstrapped distribution of final scores for every possible number of runs scored at the conclusion of the 6th inning. To find the ultimate “could have been” distribution of final scores, I would then weight these probability distributions of each possible runs scored outcome by the respective probability of having that many runs scored at the conclusion of the 6th inning (10% in the case for starting with 5 runs in the example above).
The result is a “smearing” of the run support provided in a game. For example, this method may find that a team that actually scored 7 runs “could have” scored runs with the following probabilities: 0 runs (1%), 1 run (2%), 2 runs (4%), 3 runs (6%), 4 runs (7%), 5 runs (9%), 6 runs (12%), 7 runs (15%), 8 runs (10%), 9 runs (8%), 10 runs (7%), 11 runs (6%), 12 runs (5%), 13 runs (4%), 14 runs (3%), and 15 runs (1%). I would then use this “could have been” smeared probability distribution for the pitcher’s possible run support in evaluating his outing.
Now that I have answered some questions that you may have had, let me try to summarize the conceptual approach I take. I am introducing a method that evaluates a starting pitcher’s contribution to his team’s chance of winning the game if the score is RS to RA when he leaves the game at the conclusion of the Zth inning. I will first “smear” the run support based upon RS and Z using a backwards Bayesian bootstrapping method. That will give me a probability distribution that the team could have scored X runs at the conclusion of the Zth inning, where X ranges from 0 to 25, say.
Next, using the smeared run support distribution, I will estimate the probability that the team would win a game when giving up RA runs at the conclusion of the Zth inning. Then, using the smeared run support distribution, I will estimate the probability that the team would win this game with league average pitching. I then will subtract these two probabilities to derive the pitcher’s win contribution for that game. For those readers interested in a mathematical representation, all the formulas are presented below.
===========
So, if I understand this correctly, Rob doesn't care what the score was when the pitcher is taken out (as almost all win probability measures do), but he does this reverse process in order to account for the "run in the first is worth as much as the run in the 9th" situation.
--posted by TangoTiger at 11:08 AM EDT
Posted 9:43 p.m.,
January 6, 2004
(#1) -
Charles Saeger(e-mail)
I have all of Bert Blyleven's starts from 1972-1977 (his Pythagorean win short period) in a spreadsheet, and have looked at them in depth, though I'm sure I have a start or two in error somewhere in there. A few notes:
* Blyleven's support, relative to his team, was pretty bad. For this period, he was -30 runs of support below expectation, which actually is pretty good compared to 1970-1, when he was -42 runs. He was a pretty typical hitter for a pitcher, so that can't be it ...
* Blyleven's record by run support is eerily consistent. When shutout, Blyleven went 0-17 with 2.59 ERA in 18 starts. When supported by 1 or 2 runs, Blyleven went 15-43 with a 2.53 ERA in 63 starts; he won 10 1-0 games. When supported by 3 to 5 runs (Bill James's favorite range), Blyleven went 37-27 a 2.97 ERA in 80 starts. When supported by 6 to 9 runs, Blyleven went 30-1 with a 2.81 ERA in 40 starts. When supported by 10 or more runs, Blyleven went 14-0 with a 2.43 ERA in 15 starts. I would expect the ERA to match the support more, since the run support would have been influenced by the environment, but I really don't know what this should look like.
* When you break his stats down by Win, Loss or No Decision, however, is when the data become really bizarre. When he won (96 starts) he had a 1.38 ERA. When he lost (88 starts, I'm short one, my mistake), he had a 4.24 ERA and he had a 3.47 ERA in 40 No Decisions. But, when we look at "why?" things become bizarre. His support (everything hereafter W-L-ND) went 5.76-1.89-4.50 R/G. His $H was .241-.306-.308, his GDP/Opp (an Opp is 75% of (H-HR), BB and HBP less WP and Balks times BiP/BFP) goes 16.2%-9.4%-10.8%, his UERA (that's Unearned Runs/9) was 0.23-0.58-0.41. Blyleven's pitcher-only stats are, oddly, consistent: HR/9 going 0.28-0.88-0.82, his BB/9 going 1.91-2.41-2.95 and his K/9 going 7.44-7.11-6.94. If I figure his FIP and add 3.2 we have ERAs of 2.62-3.76-3.84; his ERCs go 1.56-3.83-3.91. He pitched worse in his losses and no decisions, naturally, but not as bad as one would expect.
I somehow suspect the friendly voice coming so often on my television in the summer was an absolute terror in the clubhouse in his 20s. I'd expect his team to have performed worse, but this much?
Incidentally, when I compute custom PythW for each GAME (i.e., if the Twins scored 5 runs and Bert allowed 2, I'd plug those two numbers into the Pyth formula and multiply the result by the number of decisions Bert had in the game, 0 or 1) and add them together, I come up with -1.6 -- no real shortfall. I'm not sure it's a good idea to do this, but it was interesting.
Posted 11:52 p.m.,
January 6, 2004
(#2) -
Tangotiger
Charlie great stuff!
Your second bullet point is the most telling. Essentially, Bert did not pitch "well enough to lose". He essentially pitched the same, regardless of his offensive support.
Your third bullet point is not bizarre at all. When he wins, he gave up 1.61 RA and his team had 5.76. When he lost, his team scored 1.89 and he gave up 4.82 RA. I don't remember what the average RS/RA in wins and losses, but 5.5 RS and 2.5 RA seems about right. So, when he wins, he gives up less runs than an average pitcher that wins. When he loses he gives up less runs than an average pitcher that loses.
All your other numbers in the third bullet are because of your sampling... because he lost, we expect him to have had bad H, HR,BB rates.
So, Blyleven's W/L record does NOT show any "underachieving". Charlie's analysis, and Rob Wood's analysis match here.
Posted 1:54 p.m.,
January 7, 2004
(#3) -
Charles Saeger(e-mail)
All your other numbers in the third bullet are because of your sampling... because he lost, we expect him to have had bad H, HR, BB rates.
I know that. However, without being able to pull specifics out of my ass, pitchers usually don't hold up as well as Blyleven did. Really, his only difference was the home run rate. When he didn't win, he was a league-average pitcher who went 0-89. Somehow, I find that remarkable.
Posted 3:26 p.m.,
January 7, 2004
(#4) -
tangotiger
My guess is that *all* great pitchers post a league average ERA and go 0-whatever in his losses.
Posted 4:33 p.m.,
January 7, 2004
(#5) -
Charles Saeger(e-mail)
I would have expected a great pitcher to still be noticeably worse than league average in his losses. He did lose, and I would suspect giving up more runs than a normal pitcher would be the reason.
Posted 4:52 p.m.,
January 7, 2004
(#6) -
tangotiger
From 1974-1990, the average team (4.3 RPG), scored 2.7 runs in its losses and 5.9 in its wins, or a little more than twice runs scored in wins than losses.
If you have a good pitcher, say one who allows 3.0 RPG, I would expect such a pitcher to give up 4.64 RPG in his team's losses (which would happen 35% of the time), and 2.12 RPG in his team's wins (which would happen 65% of the time).