The front office's favorite commenter, SternFan1 offered up his first of what are sure to be his many concerns over the upcoming season in a thread last week:
My only reservation is the lasck of experience this rotation has
I think the talent is there, but will it measure up come crunch time?
At first glance, it seems a pretty easy statement to disregard. We know it was proven in The Book that pitchers over adequate sample sizes generally pitch to their true talent level in leverage splits. It concludes that you should use your best relievers in the most important spots.
It would seem the same could be said for pitching in the biggest games aka the World Series. You would think in adequate sample sizes "young pitcher nerves" and "veteran mettle" would be little more than media talking points. My intuition was to disagree with SF1, but I generally like to have some data on my side. If I am going to put in this much leg work, my findings will be posted regardless of any conclusions drawn.
The Study
The sample will consist of all World Series starts over the past 10 years (2000-2009) or 106 total starts. The median starting age for a World Series pitcher was 29 years-old. We will put the younger half of the pitchers in one group, and the 30+ in the second group to test their veteran mettle. I composed a crude aggregate FIP (Fielding Independent Pitching for each group. The term crude is used because I only used BB (did not adjust for IBB or HBP) and also used a FIP constant of 3.2. Obviously, this varies from season-to-season and from league-to-league. For reference the average margin between the leagues has revealed the AL constant to be about .09 higher from 2000-2009.
The aggregate FIP does not tell enough because we don't know what each grouping's true talent level is for comparison. I decided to use two baselines. The first is the average FIP for the group for the season. The second baseline is to weight each player's FIP for the season by his % of the group's World Series innings pitched. For example if a pitcher pitched 5 innings of a 200 inning sample, his season FIP would be weighted by 2.5%.
The major flaw of the study is the FIP constant used. However, I remain unsure of how best to adjust for league. In the World Series, some games are played under DH rules, while in others the pitcher must bat (Damn you Joe Blanton!). I certainly welcome constructive feedback.
The Data
The workbooks are available here: Workbook
The Results
Well I'll be darned if ol' SternFan didn't have shreds of truth to his blind assertion. The group of pitchers under 30 years of age posted an aggregate World Series FIP of 4.53 in 55 starts covering 308 innings. This compares to their average FIP over the season of 4.02 and a weighted FIP of 4.01. The younger group of pitchers seems to have underperformed their FIP by about 0.50. This is well beyond a discrepancy of league FIP constants.
The veterans (30+) posted an aggregate World Series FIP of 3.58. This compares to their average FIP of 3.95 and a weighted FIP of 3.88 during the season. They readily outperformed both benchmarks.
So the wheels are turning a little bit. What could be the cause? We know that the older group was a better class of ptichers based on their season numbers. What if we were to combine all pitchers above the overall median FIP of 3.97 regardless of age, and those below the median FIP?
The aggregate World Series FIP for the better pitchers was 3.52. For the same group the average regular season FIP was 3.40 and the weighted FIP was 3.35. So when the better pitchers are combined regardless of age, they slightly under-perform their true talent in the World Series.
The aggregate World Series FIP for the pitchers with FIPs north of 3.97 was 4.64. This compares to the average regular season FIP and weighted FIP of 4.59. The lesser pitchers also slightly underperformed their true talent level. This means the more skilled younger pitchers really pulled the overall skilled pitcher pool down. We can hypothesize that the evidence found supporting age being a factor in a pitcher's World Series performance is not a result of the older pitchers merely being more talented.
Finally, I wanted to test the performance of the American League starters versus the National League. Intuitively, you would expect the AL pitchers to possibly outperform their season FIP because they are now facing pitchers in some of the games as opposed to the designated hitter. Likewise, National League pitchers would have to face DH's for the first time. In reality the National League pitchers performed at their level while the American Leaguers slightly underperformed, close enough to assume statistical noise. The findings are summarized in the tables below:
wsFIP=World Series FIP, aFIP=Average FIP, wFIP= Weighted FIP
|
wsIP |
wsFIP |
aFIP |
wFIP |
|
|
<30 Years |
308 |
4.53 |
4.02 |
4.01 |
|
> 29 Years |
308.2 |
3.58 |
3.95 |
3.88 |
|
FIP < 3.98 |
323.2 |
3.52 |
3.4 |
3.35 |
|
FIP > 3.97 |
293 |
4.64 |
4.59 |
4.59 |
|
AL |
301.1 |
4.15 |
3.97 |
3.95 |
|
NL |
315.1 |
3.97 |
4 |
3.94 |
While not conclusive, there was far more evidence to support SternFan's concern that the Rays' pitching staff's inexperience could be a negative factor in the playoffs than I initially would have expected. Of the three variables tested for (age, skill, and league), only age provided any meaningful discrepancy between big game performance and true talent level. Having said that there does not mean young pitchers cannot experience World Series success, right Josh Beckett? The Rays will continue to organically grow their own pitching to maintain low costs at a position where injuries are frequent and severe. As Billy Beane emphasized in Michael Lewis's Moneyball, once you are confined to the small sample size of a seven game series, luck becomes far more prevalent in the determination of outcomes.
Finally, just for fun, I will leave you with the top 10 World Series performances as ranked by FIP:
|
Year |
Pitcher |
Age |
HR |
BB |
K |
IP |
FIP |
yFIP |
|
2001 |
Roger Clemens |
39 |
0 |
1 |
10 |
6.3 |
0.5 |
3.29 |
|
2000 |
Roger Clemens |
38 |
0 |
0 |
9 |
8 |
0.95 |
4.33 |
|
2009 |
Cliff Lee |
31 |
0 |
0 |
10 |
9 |
0.98 |
3.11 |
|
2000 |
Rick Reed |
36 |
0 |
1 |
8 |
6 |
1.03 |
4.43 |
|
2003 |
Mike Mussina |
35 |
0 |
1 |
9 |
7 |
1.06 |
3.09 |
|
2007 |
Josh Beckett |
27 |
0 |
1 |
9 |
7 |
1.06 |
3.08 |
|
2001 |
Randy Johnson |
38 |
0 |
1 |
11 |
9 |
1.09 |
2.13 |
|
2005 |
Brandon Backe |
27 |
0 |
0 |
7 |
7 |
1.2 |
4.8 |
|
2001 |
Curt Schilling |
35 |
0 |
1 |
8 |
7 |
1.34 |
3.11 |
|
2009 |
AJ Burnett |
32 |
0 |
2 |
9 |
7 |
1.49 |
4.33 |