Onward with some musings on two topics.
1. Astros Pythagorean Record
The Pythagorean formula for projecting teams' W-L records is an outgrowth of sabermetrics. The formula, as originally devised by Bill James:
Win% = (Runs Scored^2)/( Runs Scored^2+Runs Allowed^2)
The formula has been tweaked by sabermetricians, modifying the exponents to a number slightly below 2, but the modifications are relatively minor. The Pythagorean Record (or Pythag, for short) has been shown to be reasonably accurate in predicting teams' actual win-loss records. However, each year there are teams which deviate significantly from the Pythag Record. Generally the difference is viewed as "luck." But sometimes the deviations are large enough to spur analysts to search for causes other than "luck." But finding explanations has become vexing for analysts.
The Astros' Pythag projection has been higher than the teams' actual record for several months. The Astros' deviation from their Pythag W-L record actually increased since mid-season: the Astros Pythag is now 4 wins higher than the actual wins. At first glance, it seems surprising that the Astros' continue to under-perform their Pythag, despite a historically bad string of losing over the last 30 days. After all, the Astros have suffered a number of blow out losses during this losing skid, which should reduce the Pythag. But, on second thought, it's not surprising that the Astros' actual record continues to trail the Pythag Projection. The Astros have 4 wins in the last 30 days--- a .182 win percent. The winning percent is so staggeringly low that you assume that the team probably is underperforming its Pythag.
Let me speculate on two possible causes for the Astros' underperformance of their Pythag: lack of clutch hitting and poor bullpen performance. The depth and leveraging of relief pitchers is frequently hypothesized as contributors to over and under performance of Pythag. And there is some evidence to support that idea. In a previous article, I suggested that the fangraphs' clutch statistic may partially explain deviations from Pythag. Un-clutch hitting, for instance, reflects a poor distribution of hitting between leveraged situations, like close games, and unleveraged situations, like blow outs.
Let's check out the Pythagorean win discrepancy for the top and bottom five teams in clutch hitting and bullpen ERA. A positive Pythag "luck" is over performance and a negative is under performance.
Clutch Hitting |
||
Top Five Clutch |
Pythag "Luck" |
|
1.Mets |
-1 |
|
2.Reds |
4 |
|
3.Dodgers |
3 |
|
4.Pirates |
3 |
|
5.Mariners |
-4 |
|
Worst Clutch |
Pythag "Luck" |
|
30.Astros |
-4 |
|
29.Cards |
-6 |
|
28.Twins |
1 |
|
27.Padres |
-2 |
|
26.Royals |
-2 |
|
Relief Pitching ERA |
||
Top Bullpen |
Pythag "Luck" |
|
1.Reds |
4 |
|
2.Pirates |
3 |
|
3.A's |
1 |
|
4.Rays |
0 |
|
5.Braves |
1 |
|
Worst Bullpen |
Pythag "Luck" |
|
30.Astros |
-4 |
|
29.Mets |
-1 |
|
28.Brewers |
-4 |
|
27.Cubs |
-1 |
|
26.Rockies |
-5 |
The best and worst bullpens appear to follow the expected Pythagorean deviation. All of the worst bullpen teams, including the Astros, have fallen short of the Pythag projection--and mostly well short. All but one of the best bullpen teams has won more games than the Pythag projection.
The pattern is not quite as consistent for the best and worst clutch hitting teams. But three of the five best clutch hitting teams exceeded the Pythag projection, and all but one of the worst clutch hitting teams have won fewer games than the Pythag projection. The fact that the worst clutch hitting teams generally fall short of their Pythag projection at least suggests a connection.
As a side note, the Pythag deviations provide an interesting context for the NL Central race. The Cardinals, the second worst clutch hitting team and the seventh worst bullpen in baseball, have won six fewer games than the teams' Pythag Projection. If the Cardinals had played up to their Pythag, that team would lead the NL Central. Instead, the Cardinals have fallen behind the PIrates and Reds, two teams which have out performed their Pythag projections. If the NL Central teams tend to regress toward the Pythag, don't be surprised if the Cardinals knock the Pirates out of the race and battle the Reds for the division title. On the other hand, if the teams' bullpens continue performing as they have, perhaps we won't see those teams regress toward their Pythag winning percentage.
2. Lucas Harrell's WAR
Wins Above Replacment (WAR) is another product of sabermetrics. As calculated by Fangraphs, Lucas Harrell is the 8th best rookie starting pitcher based on WAR. Harrell's current 1.8 WAR is better than the performance of notable rookies like Matt Moore, Tommy Milone, and Randall Delgado.
There are also some well regarded rookie starters ahead of Harrell so far---Miley, Lynn, Parker, and Darvish, for example. So it's an uphill battle for Harrell to get in the middle of the rookie of the year competition. But Harrell's pitching is surprising for a guy claimed off waivers last year.