Alligator Blood

Baseball Prospectus runs rankings in a certain one of its own stats every day on its main page. They have enough stats that they can continually update the display throughout the year, ranging from best Team Defensive Efficiency to worst VORP among NL catchers. Today, the stat ranking was the top AL relievers in terms of leverage, which caught my eye because two Twins, Nathan and Rincon, ranked in the top seven. I looked deeper, and it brought up a series of interesting questions.

The first interesting stat that jumped off the page was that eight of the top nine relievers (min 40 IP) in terms of leverage played for winning teams. That fact may not seem surprising at first, but it should once you consider the nature of reliever game-state leverage. Think of what it means to be pitching in a high-leverage state: the later and closer the game is, the higher the leverage of the situation will be. All games are close at the beginning, but the better or worse a team is, the less close the game tends to be as it gets towards the end (which is the time when a) relievers pitch and b) leverage ramps up). In other words, teams around .500 should have higher leverage situations than teams who are very good or very bad, as they tend to separate even within a nine-inning sample- the same rationale for why winning blowouts is a better measure of a good team than winning one-run games, which is that it involves less luck and more skill. It doesn’t so much matter whether the team is stronger at the plate or in the field, since game-state stats treat every run as equally important, even though it is actually probably a bit more difficult to score the first run of the game in the 7th inning than it would be to score the 10th run of the game in the 7th inning due to bullpen usage patterns and psychological concerns, but that is another issue for another time. That is to say that the leverage would be the same going into, say, the 8th inning of a game with the score 3-2 as it would be if the score was 12-11.

So having eight of the top nine relievers playing for winning teams (including six on 90 win teams) is something of an aberration; it should be roughly equally balanced between winning and losing teams who are all close to .500 (meaning they maximize high-leverage situations by playing in less blowouts either winning or losing). To see if that aberrant result was repeated in the NL, I skipped over to the senior circuit’s leverage report, which featured an astounding six pitchers from 79-83 win teams in the top nine, meaning fully two-thirds of the highest leverage pitchers in the league came from a range covering one-third of the NL’s teams. While the sample is not large enough to prove much, it confirmed my belief that the AL report smelled wrong.

That query got me on the subject of how these teams got so many high-leverage situations, since it is not realistic to think that the Yankees, Angels, Indians and White Sox were ALL playing a disproportionate number of close games. That turned out to be basically true- the AL Central teams were close to league average in one-run games, the Yankees were well below average and the Angels well above. But that raised the question of performance in close games; perhaps it was strong bullpens that made these teams winners, which would explain part of the statistical aberration. If all of the pitchers who led the league in leverage were exceptionally successful in those situations, it would go a long way towards transforming a good team into a very good one. But there is even a strong amount of variation in that category, as the Yankees and White Sox were very good in one-run games, the Indians very bad and the Angels about as good as they were in less close games.

Taking a step back, though, one-run games don’t tell us everything about how well a bullpen performed. Due to the save rule, most managers treat a one-run lead the same as they would treat a two- or three-run lead, and games where the bullpen pitched well enough to allow the offense to get close can actually work against them (which might help explain why the Indians one-run record is so bad even though the bullpen was good). In fact, all five of these teams, including the sub-90 win Twins now, have bullpens with excellent reputations. So if the leverage stat is correct and all of these teams had many close scores late in games, what made the difference? The obvious reaction would be that the Twins offense was deficient, but the nature of the question standardizes that solution away; the run scoring/run preventing combination in the early innings got them into close games just like those units for the White Sox, Indians, Yankees and Angels, but they turned that multitude of high-leverage situations into 83 wins rather than 93 or more, as the other four did.

Now, I understand that there are methodological concerns here, namely that the Twins had more pitchers near the top of the leverage list (3 of the top 14), meaning they probably had even more high leverage situations than these other teams who were represented by one or two pitchers at the top. Still, I think it demonstrates that the Twins big three of Nathan, Rincon and Crain have been slightly overrated exactly because they pitch in such high-leverage situations. Look what it did for Jose Mesa; he completed a year of sub-replacement level relief for Pittsburgh but ranked third in the league in leverage and was able to get a lucrative contract out of it with Colorado. The same effect may be at play for the Twins. If you see a pitcher pitching at big times frequently enough, he may start to seem like a big time pitcher, inflating his reputation over his true ability.

I do not mean to say that the Twins relief aces are bad pitchers. Actually, they all rank in the top 13 in the league in WX (reliever win expectancy, compared to league average, not adjusted for leverage), implying that they are solid through and through. Still, don’t look for any of them to break through in the near future, as underrated pitchers would tend to be those with high WX’s and low leverage, otherwise known as dominance in mop-up duty, a combination that allowed pitchers like Scot Shields, Scott Linebrink and Ray King to fly under the radar in 2004 before posting higher-leverage, higher-profile 2005s. Similarly, don’t be surprised if Dan Wheeler, Al Reyes, Jason Frasor and/or Justin Duchsherer step into brighter lights in the coming season.