Often a team will bring its closer into a tie game in the ninth inning at home on the theory that if he can hold the opposition in the top half of the inning, his team has a chance to win in the bottom half. When the Padres try this (as they did recently against the Dodgers) and it doesn’t work, fans and analysts alike are quick to note that Trevor Hoffman isn’t effective in non-save situations. Frequently this observation is accompanied by calls to hold him out of such situations in the future.
But here’s a question to consider: Is it true? Does Hoffman really struggle in non-save situations, or is the power of perception at work?
To answer this question, I ran through Hoffman’s game logs at Baseball-Reference. Using B-R’s Play Index, it’s possible to choose only those games that represent save opportunities. I did that for 1994-2007 (no 1993 because Hoffman wasn’t a closer yet) and dumped everything into a spreadsheet. And then I added everything together and came up with the following:
Three things jump out at me here. First, Hoffman’s ERA is indeed higher in non-save situations. Second, Hoffman appears to be less concerned with putting runners on base in non-save situations—his strikeouts are down 4%, his hits allowed are up 10%, and his walks are up a staggering 52%.
So, yeah, it looks like the naysayers are right. End of story.
Wait, I forgot to mention the third thing that jumped out at me: the difference in innings pitched. Hoffman has worked only half as many innings in non-save situations as in save situations.
By their nature, relievers are subject to the whims of small samples. This effect should be magnified when we are examining a further subset of those already-small samples.
Maybe looking at Hoffman’s career on a year-by-year basis will shed additional light. Okay, then, let’s try that. First up, here’s how Hoffman has fared in save situations as a full-time closer:
As you might expect, we see a lot of consistency—excluding 1994, 1995, 1998 and 2007, Hoffman’s ERA in save situations has been between 2.03 and 3.40. That’s a sizable range, but again, we’re dealing with small samples, generally 45-50 innings, so we would expect some fluctuation. His hit prevention has been good—usually around 7 H/9—as has his walk rate.
You’ll notice that Hoffman had no save opportunities in 2003. He missed most of the season following shoulder surgery and got into only nine September games, all in low-leverage situations. You’ll also notice that his strikeout rate has plummeted since then. Some of this may be due to his shoulder, some to the natural effects of aging. Whatever the case, even in save situations, Hoffman has not been nearly as dominant in terms of being able to put the ball past hitters as he was before his surgery.
This is important to bear in mind when we look at his numbers in non-save situations:
Taking pieces of each of these last two tables, we see that there have been seasons in which Hoffman pitches appreciably better (very imprecisely defined as a full point of ERA) in save situations:
So that’s five times in 13 seasons (we’ll throw out 2003 because nine games don’t tell us much) where conventional wisdom holds. The non-save situations also represent a little over 90 innings of work, total, so again we’re dealing with a tiny sample.
Oh, and that article I linked to in the first paragraph? It was written in 2005, which as you can see was a historically bad year for Hoffman in non-save situations. This might help explain why people were complaining about his usage that season.
How about the opposite scenario? Presumably if Hoffman’s skill level improves when he enters a save situation, we’ll never see him post superior numbers in non-save situations. Is this the case? Not exactly:
Four seasons meet our criteria here, and this time we’re looking at about 80 innings of non-save situations.
Returning to our original question—Does Hoffman struggle in non-save situations?—the answer appears to be a definite maybe. Sometimes he does, other times not so much. In other words, the only identifiable pattern is the absence of an identifiable pattern.
Of course, we looked at only one pitcher. The purpose here was to investigate a commonly accepted truth and see if it held up to the scrutiny of actual facts. It did not—at least not to a degree that matches the decibel level of objections to the tactic in question.
An interesting future study, if anyone has the time and inclination, would be to compare large groups of closers and see how their performances vary depending on situation. My suspicion is that such inquiry might lead us down the “clutch hitter” path that only leaves everyone frustrated and confused.
Hm. When I put it like that, it sounds kind of fun…
References & Resources
As always, Baseball-Reference proved invaluable and generally rocked my world.