Posted by MikeT23 on 6/26/2012 8:14:00 AM (view original):
I have no doubt that at least one person believes that. I'll refer to the ol' saying of "The cream always rises to the top".
Lohse has proven, over the course of his career, to be a quite average pitcher. So, in this particular game, he might have recorded 19 outs via lasers hit right at someone. However, if history holds true, some of those 19 outs will become hits and Lohse will produce a line that's in line with his career numbers. Some of those hits may be of the blooper variety where Lohse threw a perfect pitch, even mediocre pitchers are capable of it, and the batter simply stuck his bat out.
For me, it's much easier to believe that things will always "even out" than for me to disregard, in this case, 86% of the game.
Looks like I'm pretty late to this party but I love FIP. Your last sentence - I think that's all FIP is trying to do: account for "evening out" of balls in play events by assigning a league average BABIP to each pitcher. It generally works pretty darn well because pitchers' career BABIPs tend to be pretty tightly clustered.
FIP is just a measure of performance on outcomes that don't involve a fielder making a play on a ball. It's scaled to ERA for context/ease of reading. Strikeouts only measure one kind of out. AVG only measures certain plate outcomes and ignores others. It's not new for a stat to not cover every single outcome. Most are designed to focus on one thing and ignore others. FIP ignores batted balls touched by fielders (and so any resulting impact of luck and defensive differences, where they do exist, are then also removed) by highlighting specific outcomes that have been proven to be more stable for a pitcher (the things that are so called "more in their control"). The fact that it really does generally measure performance in a similar but more stable way than ERA does while not counting actual balls in play data is amazing (to me).
Given a long enough career, I think even saber nerds agree that ERA is a useful measuring stick (given normal caveats of adjustments to era, league, park, etc., I guess). FIP and ERA converge well over time for most pitchers. Because FIP isn't subject to all the non-pitcher controlled events that ERA is, FIP is more stable over time and probably provides a better measure of talent at any given point in time than ERA does because ERA is much more variable year to year.
Data below is for all pitchers from 2001 - 2011 that pitched at least 150IP in consecutive years (621 instances). Columns that end in 1 are the ERA, FIP and BABIP for pitchers in first year of pair and the other columns are results from following year. Pitchers are grouped by amount of over- or under- performance in year one ERA relative to year one FIP. Every group (but not every pitcher within group) saw ERA regress toward FIP in year two and BABIP regress toward the mean (.291) while FIP stayed pretty constant. The groups with the biggest over- or under- performance saw the biggest moves toward FIP. FIP was a better predictor of year 2 ERA than year 1 ERA was. Not every pitcher moves this way every year but over time almost all of them do. Isn't that pretty cool?
| Y1 FIP-ERA |
# |
ERA1 |
FIP1 |
BABIP1 |
ERA2 |
FIP2 |
BABIP2 |
| >= 1.00 |
15 |
3.33 |
4.54 |
0.257 |
4.40 |
4.56 |
0.283 |
| 0.99 >= 0.75 |
39 |
3.39 |
4.24 |
0.272 |
4.03 |
4.21 |
0.287 |
| 0.74 >= 0.50 |
76 |
3.65 |
4.26 |
0.274 |
4.11 |
4.23 |
0.287 |
| 0.49 >= 0.25 |
97 |
3.71 |
4.06 |
0.283 |
4.00 |
4.12 |
0.291 |
| 0.24 >= 0.01 |
119 |
3.89 |
4.01 |
0.288 |
4.06 |
4.10 |
0.291 |
| |
|
|
|
|
|
|
|
| 0.00 |
5 |
3.65 |
3.65 |
0.293 |
3.89 |
3.99 |
0.293 |
| |
|
|
|
|
|
|
|
| -0.24 <= -0.01 |
110 |
4.04 |
3.92 |
0.295 |
4.01 |
4.03 |
0.291 |
| -0.49 <= -0.25 |
75 |
4.38 |
4.02 |
0.301 |
3.95 |
3.94 |
0.291 |
| -0.74 <= -0.50 |
46 |
4.71 |
4.10 |
0.311 |
4.27 |
4.00 |
0.303 |
| -0.99 <= -0.75 |
23 |
4.80 |
3.95 |
0.320 |
3.97 |
3.89 |
0.292 |
| <= -1.00 |
16 |
5.43 |
4.21 |
0.327 |
4.33 |
4.32 |
0.289 |