So, after losing my first draft of this to a Windows crash, I'm going to be much more concise this time around. I decided to group qualifying starting pitchers into buckets 1 (ace) through 5 (bottom of the rotation) by FIP. I took all qualifying seasons by starting pitchers between 2007 and 2012 and found univariate percentiles for 20%, 40%, 60%, and 80%. This allowed me to group those observations into 5 buckets according to the following table.
p20 |
p40 |
p60 |
p80 |
|
FIP |
3.356 |
3.77 |
4.06 |
4.46 |
So, any observation with a FIP below 3.356 falls into the first bucket (ace). Out of the remaining observations, if they have a FIP below 3.77 they fall into bucket 2, and so on. This led to the following distribution of buckets.
|
1 |
2 |
3 |
4 |
5 |
Players |
87 |
86 |
86 |
87 |
88 |
Qualifying starting pitchers will typically be be better than the typical starters. Those who qualify are good enough to accumulate the innings necessary to meet the inning threshold to make it into the sample (assuming good and bad pitchers injured at similar rates). This will skew the results slightly as pitchers we would perceive to be better than average may be shifted down a bucket as our sample is heavy with the better pitchers. This also will have the effect of making the top buckets a bit more exclusive when we compare them to all starting pitchers in the league.
Now, let's drop the qualifying observations of starting pitchers for this season (as of games completed 7/5/2013) into the buckets we created earlier. The 93 qualifying pitchers thus far are then placed into the following buckets by team.
Team |
1 |
2 |
3 |
4 |
5 |
- - - |
0 |
0 |
1 |
0 |
0 |
0 |
1 |
1 |
1 |
0 |
|
0 |
1 |
0 |
0 |
1 |
|
0 |
1 |
0 |
1 |
2 |
|
0 |
0 |
0 |
1 |
1 |
|
0 |
2 |
3 |
0 |
0 |
|
0 |
0 |
0 |
2 |
1 |
|
3 |
0 |
0 |
0 |
0 |
|
1 |
1 |
1 |
0 |
0 |
|
1 |
0 |
0 |
2 |
1 |
|
1 |
1 |
0 |
0 |
0 |
|
0 |
2 |
1 |
1 |
0 |
|
0 |
1 |
0 |
0 |
1 |
|
1 |
1 |
0 |
0 |
1 |
|
1 |
1 |
1 |
0 |
0 |
|
1 |
0 |
0 |
2 |
0 |
|
2 |
1 |
0 |
0 |
0 |
|
0 |
0 |
0 |
1 |
2 |
|
0 |
1 |
2 |
0 |
1 |
|
1 |
2 |
0 |
0 |
0 |
|
1 |
0 |
1 |
0 |
0 |
|
2 |
0 |
0 |
0 |
0 |
|
0 |
0 |
2 |
0 |
1 |
|
0 |
1 |
0 |
1 |
1 |
|
2 |
1 |
0 |
1 |
0 |
|
1 |
1 |
0 |
0 |
0 |
|
0 |
1 |
1 |
1 |
1 |
|
3 |
1 |
0 |
0 |
0 |
|
0 |
0 |
0 |
0 |
2 |
|
1 |
0 |
1 |
0 |
1 |
|
0 |
1 |
2 |
1 |
0 |
|
Total |
22 |
22 |
17 |
15 |
17 |
The observation without a team is a player who has been traded. So far, it seems the qualifying pitchers this year are pitching slightly better than we would expect as 47% of the qualifying pitchers are in the 40th percentile of the previous years. I would expect some regression as we are looking at a half season worth of innings thus far.
Would anyone like to take a stab at which Phillies fall into their respective buckets so far? I imagine you guys won't be surprised by the following table.
Name |
FIP |
Bucket |
2.67 |
1 |
|
3.65 |
2 |
|
3.75 |
2 |
So, their three starters who haven't been hurt are pitching great to fairly well, with Kendrick just sliding under the threshold for the 40th percentile. (Note: as I'm writing this during the game, he surely will drop one bucket.)
How are all starters across the majors doing? Well, these numbers may not mean a whole lot due to incredibly small sample sizes, but let's take a look anyway.
Team |
1 |
2 |
3 |
4 |
5 |
- - - |
0 |
0 |
1 |
0 |
1 |
Angels |
0 |
3 |
2 |
1 |
4 |
Astros |
0 |
2 |
0 |
0 |
5 |
Athletics |
0 |
2 |
0 |
2 |
2 |
Blue Jays |
0 |
0 |
1 |
3 |
7 |
Braves |
1 |
2 |
3 |
0 |
0 |
Brewers |
0 |
0 |
0 |
3 |
7 |
Cardinals |
4 |
2 |
1 |
2 |
0 |
Cubs |
1 |
1 |
2 |
1 |
1 |
Diamondbacks |
1 |
1 |
0 |
4 |
1 |
Dodgers |
1 |
1 |
1 |
3 |
3 |
Giants |
0 |
2 |
1 |
2 |
2 |
Indians |
0 |
2 |
0 |
1 |
5 |
Mariners |
2 |
1 |
0 |
1 |
4 |
Marlins |
3 |
1 |
0 |
2 |
3 |
Mets |
1 |
0 |
2 |
2 |
2 |
Nationals |
3 |
4 |
0 |
0 |
2 |
Orioles |
1 |
1 |
0 |
2 |
8 |
Padres |
1 |
2 |
1 |
1 |
3 |
Phillies |
1 |
2 |
1 |
1 |
2 |
Pirates |
4 |
1 |
2 |
3 |
1 |
Rangers |
3 |
1 |
0 |
3 |
2 |
Rays |
0 |
2 |
2 |
1 |
3 |
Red Sox |
1 |
1 |
1 |
1 |
4 |
Reds |
2 |
1 |
1 |
2 |
1 |
Rockies |
3 |
1 |
0 |
0 |
4 |
Royals |
0 |
1 |
1 |
1 |
3 |
Tigers |
4 |
1 |
0 |
0 |
1 |
Twins |
1 |
1 |
1 |
1 |
5 |
White Sox |
1 |
1 |
1 |
1 |
3 |
Yankees |
1 |
1 |
3 |
2 |
0 |
Total |
40 |
41 |
28 |
46 |
89 |
Looking at the distribution of players across the majors, we see roughly what we would expect. The fifth bucket is extremely heavy due to the introduction of the fringe players whose performances wouldn't be represented in the original five year sample.
Let's take a look at how our Phillies starters on the whole are doing.
Name |
FIP |
Bucket |
GS |
Cliff Lee |
2.67 |
1 |
18 |
Cole Hamels |
3.65 |
2 |
18 |
Kyle Kendrick |
3.75 |
2 |
17 |
3.78 |
3 |
7 |
|
4.09 |
4 |
6 |
|
4.49 |
5 |
14 |
|
6.27 |
5 |
7 |
I included an additional column to show how small the sample sizes are both as a whole and in the cases of the players who didn't qualify for the previous table of Phillies players. Things to take from this table are that Kendrick and Lannan are hovering either side of the 40th percentile, Cloyd just missed the 60th percentile, and Pettibone just missed the 80th percentile. Don't put too much stock in those values yet, though. The season is only halfway over and these samples are still fairly small.
One last thing I looked at was how well all teams were doing as a whole. As we can see with the Phillies example, things can look drastically different when we are considering only qualifying pitchers to all starters.
Team |
1 |
2 |
3 |
4 |
5 |
Phillies (qualifying) |
1 |
2 |
0 |
0 |
0 |
Phillies (all) |
1 |
2 |
1 |
1 |
2 |
A simple mean of the buckets won't tell the whole story. I chose to weight each starter by the proportion of games he started for this team to try to make it more reflective of the starts each team has gotten so far.
Team |
Mean bucket qualifying |
Weighted bucket qualifying |
Mean bucket all |
Weighted bucket all |
- - - |
3 |
3 |
4 |
3.316 |
Angels |
3 |
2.980 |
3.6 |
3.465 |
Astros |
3.5 |
3.541 |
4.143 |
3.931 |
Athletics |
4 |
4.014 |
3.667 |
3.713 |
Blue Jays |
4.5 |
4.5 |
4.545 |
4.471 |
Braves |
2.6 |
2.612 |
2.333 |
2.593 |
Brewers |
4.333 |
4.340 |
4.7 |
4.553 |
Cardinals |
1 |
1 |
2.111 |
1.812 |
Cubs |
2 |
1.961 |
3 |
2.377 |
Diamondbacks |
3.5 |
3.478 |
3.429 |
3.337 |
Dodgers |
1.5 |
1.486 |
3.667 |
3.106 |
Giants |
2.75 |
2.768 |
3.571 |
3.141 |
Indians |
3.5 |
3.417 |
4.125 |
3.756 |
Mariners |
2.667 |
2.623 |
3.5 |
3.326 |
Marlins |
2 |
1.936 |
3.111 |
3.047 |
Mets |
3 |
2.941 |
3.571 |
3.086 |
Nationals |
1.333 |
1.353 |
2.333 |
2.233 |
Orioles |
4.667 |
4.694 |
4.25 |
4.488 |
Padres |
3.25 |
3.239 |
3.375 |
3.310 |
Phillies |
1.667 |
1.660 |
3.143 |
2.736 |
Pirates |
2 |
2.097 |
2.636 |
2.576 |
Rangers |
1 |
1 |
3 |
2.884 |
Rays |
3.667 |
3.627 |
3.625 |
3.368 |
Red Sox |
3.667 |
3.74 |
3.75 |
3.432 |
Reds |
2 |
2 |
2.857 |
2.326 |
Rockies |
1.5 |
1.529 |
3.125 |
3.046 |
Royals |
3.5 |
3.485 |
4 |
3.759 |
Tigers |
1.25 |
1.235 |
1.833 |
1.376 |
Twins |
5 |
5 |
3.889 |
4.398 |
White Sox |
3 |
3.082 |
3.571 |
3.349 |
Yankees |
3 |
3.031 |
2.857 |
2.919 |
This looks similar to what we would expect as the weighted mean of qualifying starters is very close to the standard mean of qualifying starters and we can see a much larger difference between the mean across all starters and the weighted mean across all starters.
By this metric, the Phillies starters thus far have been fairly balanced, in regards to the last several years, with a weighted bucket of 2.736. A weighted bucket score of 2 would put their performance in the 20th to 40th percentile and a score of 3 would place them in the 40th to 60th. Using a crude approximation, I would estimate them to be in the range of the 35th to 55th percentiles. This was found by (bucket percentile width)*(weighted bucket all - 1) = 34.72 and (bucket percentile width)*(weighted bucket all) = 54.72. This seems reasonable based on the how close their weighted bucket is to 3. It also seems on par with their performance so far on the season.
Over the next week, I'll look to use this metric to analyze team performances at this point in the season.