- FanGraphs Baseball - http://www.fangraphs.com/blogs -

Crunching the Numbers on Letting Lewis Hit

Posted By Dave Cameron On October 27, 2011 @ 10:49 pm In Daily Graphings | 26 Comments

It’s the World Series. Your team hasn’t played since Monday. You have a deep bullpen and the availability of a starting pitcher working on three days rest. It’s the World Series.

Top 5: Texas 4, St. Louis 3

Bases loaded, 2 outs: Colby Lewis struck out swinging

Ron Washington sent his pitcher up to bat with three men on base in a one run game. It’s hard to put into words just how bad of a decision that was. Let’s try anyway.

Lewis has only 17 Major League place appearances, since he’s an AL pitcher and all, so his .188/.188/.250 line is a small sample. But he’s a pitcher – an AL pitcher – so there’s no real reason to believe he’s significantly better than that. After all, the average pitcher hit just .143/.175/.183 this year, so Lewis belongs to a population of guys who hit about as well as he has. Even if we just take Lewis’ line (remember, better than an average pitcher’s line), there was something like an 81% chance that Lewis would make an out in that situation, ending the inning and leaving all three runners on base.

Yorvit Torrealba was the fake pinch-hitter, so let’s assume he would have been the guy to take Lewis’ place had Washington lifted Lewis. He’s a career .260/.318/.390 hitter, and his 2011 line was nearly identical to that mark. You have to discount his performance a bit to account for the pinch-hitter penalty, so let’s knock 10% off his of career line, so as a PH, that would make him a .234/.286/.351 hitter. Not great, but way better than Lewis.

In that situation, Torrealba has approximately a 5% better chance of getting a hit (likely scores two runs) and a 5% better chance of reaching base via a non-hit (likely scores one run). The cumulative probability of not making an out there is worth .15 runs, and that’s before you account for the fact that not making an out also gives Ian Kinsler a chance to expand the lead even more. Given the chances for a big rally where Torrealba reaches and Kinsler also gets a hit, the average gain from pinch hitting for Lewis there is around 0.2 runs.

That might not sound like a lot, but when you compare the expected gain from keeping Lewis in the game, it’s enormous.

First off, nearly any reliever that the Rangers brought in would be even less likely than Lewis to allow runs going forward. His 4.40 ERA/4.54 FIP/4.10 xFIP line in 2011 is worse than nearly everyone else in the bullpen, and he was about to face the Cardinals line-up for the third time. Hitters perform demonstrably better against a pitcher the more they face them in a game.

You just can’t make a rational case for Lewis being likely to outperform any of the Rangers bullpen guys going forward unless you dramatically overweight the value of his last few batters faced, ignore the first inning, and believe that Lewis had somehow reached a new level of enlightenment somewhere in inning #2. Pinch-hitting for Lewis not only makes you far more likely to score, it makes it less likely that the Cardinals score in the bottom of the fifth inning.

But, just for fun, let’s see how much better you’d have to believe Lewis was than the alternative to justify the decision. Remember, you’ve created a .2 run hole for yourself by letting Lewis hit.

If you plan on getting one more inning from Lewis, you’d have to believe that his true talent ERA at that point was 1.80 runs lower than his replacements. If you believe he was likely a 4.50 ERA pitcher in the next inning (basically, his 2011 ERA/FIP), then the alternatives would have had to have an expected 6.30 ERA to make it a break even decision. Pitcher who had a 6.30+ ERA in 2011: John Lackey.

If you plan on getting two more innings from Lewis, you’d have to believe that his true talent ERA was 0.90 runs lower than his replacements. If you again believe that he’s a 4.50 ERA pitcher (this is really stretching it, given his pitch count and the fact that he’d be facing the middle of the Cardinals order the third time through the order), then you’d have to believe his replacements would have had an expected 5.40 ERA over the next two innings. Pitcher who had a 5.40+ ERA in 2011: Andrew Miller.

And remember, if you’re asking Lewis to go two innings, you’re asking him to get Pujols/Berkman/Holliday out in that second inning of work. Both Lewis and Berkman have massive platoon splits, and there’s simply no way you should be willing to let Berkman face an RHP in that at-bat. So, if you factor in the match-ups, making the case that Lewis would provide any kind of advantage in that second inning of work is nearly impossible.

Can you imagine the reaction Rangers fans would have had if Washington went to the mound and called for Lackey or Miller to preserve a one run lead in Game Six of the World Series? That’s basically what he did by letting Colby Lewis hit.

As I type this, Adrian Beltre and Nelson Cruz have hit back to back home runs to give Texas the lead. The Rangers are now 83% favorites to win the game and take home their first World Championship. Within an hour of posting this, Washington will probably be holding a trophy, and this will just be a footnote in history that ended up not affecting the outcome of the game.

But, we gave Tony LaRussa a lot of flak for the decisions he made in Game 5, and what Washington did in the fifth inning of this game was worse than anything LaRussa did during phone-gate. It was one of the worst managerial decisions you’ll ever see. Ron Washington should be extremely thankful that he’s managing a fantastically talented group of players.


Article printed from FanGraphs Baseball: http://www.fangraphs.com/blogs

URL to article: http://www.fangraphs.com/blogs/crunching-the-numbers-on-letting-lewis-hit/

Copyright © 2009 FanGraphs Baseball. All rights reserved.