Keeping Score: Lee’s Value to Phillies Can Be Measured in Zeros – NYTimes.com
Posted by Sean Forman on September 10, 2011
Keeping Score: Lees Value to Phillies Can Be Measured in Zeros - NYTimes.com.
My latest in the Times. My favorite snip from the article is that Cliff Lee has 20% of all shutouts in the National League. This is the highest percentage since Ed Morris of the 1886 AA Pittsburgh Alleghenys.
September 10th, 2011 at 11:23 am
That's some interesting stuff
September 10th, 2011 at 11:53 am
For a second, I read that as "strikeouts" and I thought to myself, "Does he even have 20% of the strikeouts on the Phillies?!?!" (He probably does, but certainly not of the league! Speaking of, what is the highest percentage of a league's Ks that an individual pitcher has amassed? I'll see if I can find it but not quite sure I know how to search for it...)
September 10th, 2011 at 7:09 pm
@2
As a first guess, I went to Old Hoss Radbourn in 1884. He had over 10% of the league's Ks that year (441/4335)!!! Didn't bother checking any other seasons, but frankly it seems unlikely to me that anything would top that.
September 10th, 2011 at 7:27 pm
Had to check.
Matt Kilroy, pitching for Baltimore (Yes, the Orioles), in 1886
had 513 Strikeouts. The AA as a league had 4,730 total.
As a 20 year old, Kilroy completed 66 of 68 starts that year.
Next year he completed 66 of 69 starts, throwing 583 and
589.1 innings respectively.
He must have been hurt in 88 as he completed only 35
of 40 starts.
Then in 1889 Matt Kilroy completed 55 of 56 starts.
It was a different world back then.
September 10th, 2011 at 9:13 pm
The last pitcher with more than 11 starts with zero runs was Bob Gibson in 1968 with 13 games (all shutouts) out of 34 starts. The two most recent at 11 were Gooden (35 starts, 8 SHO) and Tudor (36 starts, 10 SHO). Cliff Lee has 11 starts with zero runs (28 starts, 6 SHO). 5 of the non-complete game were team SHOs.
He has the most total shutouts from 2008-2011.
Counting this year, he's been in the top 7 in the league in SHOs 5 times in the last 4 years.
September 11th, 2011 at 12:11 am
"Top 7" is a little screwy with shutouts, as without checking I assume it just means he's gotten *a* SHO each of those years.
September 11th, 2011 at 8:12 am
@6
You're right. But the funny part is "5 times in 4 years".
September 11th, 2011 at 10:24 am
As I was reading Sean's piece on Lee's shutouts, a passage from Jack Moore's recent Fangraphs/ESPN article about Doug Fister was in my mind:
[comparing Fister and Max Scherzer:] "Although the chances of a poor start from Fister are slightly higher, the chances of a truly shutdown start are also higher. This is a trade the Tigers (or any team) should be more than willing to make -- a pitcher giving up zero runs will never lose, whereas the Tigers' offense could bail Fister out of a four-, five- or even six-run outing behind the powerful bats of Miguel Cabrera, Victor Martinez and crew."
http://insider.espn.go.com/mlb/story/_/id/6946849/mlb-doug-fister-become-reliable-no-2-starter-detroit-tigers
(sorry, it's an Insider piece; couldn't find a different location)
Could any of the math smarties out there point me towards understanding the truth (or not) of that claim? It's a question that's often raised, but I've not seen a definitive answer: In an average scoring context, will a pitcher generate more win value by allowing 2 or 3 runs every time, or alternate shutouts with 5-run games?
September 11th, 2011 at 10:26 am
I mangled my last sentence @8 -- it should have ended:
"... will a pitcher generate more win value by allowing 2 or 3 runs every time, or by alternating shutouts with 5-run games?"
September 11th, 2011 at 5:57 pm
I hope this can be understood without further detail. Without going into all the statistics, based on the ML probability of a team scoring 0-2, 3-5, 6+ runs, it's slightly in favor of the 0,5 pitcher. That's assuming the pitcher pitches a complete game and I put all the shutouts as wins, no 0-0 ties. It makes a difference, if the game is incomplete and goes to the relievers because there's a probability that they will give up a run and my analysis was based on only the starting picher giving up runs. 0 and 5 run games are very likely blowouts and if the relievers give up a run or two, it won't change the outcome of many games. Relievers are more likely to change the outcome of a 2,3 run outing, because those are more likely to be close games and the winning percentage can only be lower than what I calculated based on total runs for the pitcher's team for the entire game.
September 11th, 2011 at 9:01 pm
I found this newspaper article about Kilroy in Nov. 1889. It said he was an outstanding fielder, 0.274 batting average, and that he had a 7 inning 0-0 nohitter, but he lost a run when he failed to touch 3B when he scored. The article said he pitched 59 games and gave up 104 runs for a 1.76 average. Maybe they were talking about earned runs.
Boston Daily Globe
Saturday Nov. 26, 1889
page 7
I saw this game June 24, 1990
Chicago 22, Brooklyn 3
1 earned run each team
Brooklyn had 20 errors
"Johnny Ward's folks played ball like harvest hands today. In the fourth inning the White Stockings piled on 13 runs , none earned".
September 11th, 2011 at 11:56 pm
JA, I know the "consistency" issue with pitchers was addressed a couple times here in the past couple months. I think it was stated that consistency (at least, at a higher level) is better, but I don't know that we had any hard numbers behind it.
I chose your examples from #8. Team winning % in games allowing either 0 or 5 runs, or when allowing 2 or 3 runs.
............0 or 5....2 or 3
2011:...581......666
2010:...621......693
2009:...574......723
There is a clear advantage to consistency at this level. If you raise the average runs scored, I don't think it would be so clear. The results might also change somewhat if you filter for runs allowed only by the SP.
September 12th, 2011 at 12:57 am
@8
The pitching team's batters R/G distribution is the same as the major league R/G distribution
We will throw out the 0, 2, 3 , 5 run games when necessary
In 2011 there were 4053 games where the team scored 1 run or more
SHO pitchers would be 4053-0 or 100%
In 2011 there were 3002 games where the team scored 3 runs or more and 767 where the scored 1 or less.
2 run pitchers would be 3002-767 or 80%.
In 2011 there were 2369 games where the team scored 4 runs or more and 1344 where they scored 2 runs or less.
3 run pitchers would be 2369-1344 or 63%
In 2011 there were 1260 games where the team scored 6 runs or more and 2584 where they scored 4 runs or less.
5 run pitchers would be 1260-2584 or 33%
So if the 0,5 run guy pitches 100 0-run games and 100 5-run games he would be expected to win 133 (66%)
So if the 2,3 run guy pitches 100 2-run games and 100 3-run games he would be expected to win 143 (72%)
The 2,3 pitcher has the edge.
September 12th, 2011 at 6:48 am
@13
That data was based on league run distribution on the probability that a team would be above or below the runs given up by the pitcher in 2011
Actual win loss record of teams
0 runs 293-0 100%
2 runs 428-149 74%
3 runs 378-255 60%
5 runs 169-333 34%
Normalize to same number of games pitched.
The 0,5 pitcher wins 67%
The 2,3 pitcher wins 67%
September 12th, 2011 at 4:51 pm
Normalize to same number of games pitched.
Ahh, I don't think I did that in my post #12, and obviously I should have.
September 12th, 2011 at 5:13 pm
0 run pitcher. Bizarre sense of humor pitcher who likes the beach.
5 run pitcher. Identical twin that no one knows about.
My French teacher told us he had a pair of twins in his class. One would always do well on a test and on the next one the other would do well. He figured out how to tell them apart and the pattern changed.