Tuesday, January 2, 2007
Trading RJ - Revisited
Randy Johnson is still a Yankee, although it would seem to be just a matter of time before that’s no longer true. A lot of people seem to think that it’s addition by subtraction to get rid of Johnson, but I need to reiterate again that I think it is going to hurt the Yankees in 2007, especially if the rumored package of a few minor league arms is what’s coming back in return.
There are plenty of reasons to be concerned about Johnson’s effectiveness heading into 2007. He’s going to be 43. He just had back surgery, and he’s had knee problems for the last several years. He also gave up a ton of runs last year, despite underlying peripherals that point to this being somewhat of a fluke.
All that being said, all the projections I’ve seen for Johnson have him providing 180-200 innings of above average to good quality. Focusing on only his recent season and his ERA+ of 88 is not good player evaluation. To show why, I took a look at all the pitchers who’ve put up an ERA+ of 90 or worse in 200 innings or more since 1945.
Here are the cumulative totals of all those players, Year 1 being the year they put up the ERA+ of worse than 90, and Year 2 being the following season.
In this chart, ERA+ is the league average ERA adjusted for the players’ home park, divided by the players’ ERA, then multiplied by 100. HR+ is the league average HR/Batters faced divided by the players’ HR/Batters faced and multiplied by 100. BB+ is same thing, but for BB instead of HR. K+ is the pitchers’ K/BF divided by the league average K/BF and multiplied by 100.
As you can see, the biggest difference between the first and second year is the HR rate. Research has shown that HR rate is at least partly a function of fly balls allowed and has some fluctuation from year to year. BB rate and K rate are pretty constant, which indicates that at least part of the poor performance may have been on factors not in the pitchers’ direct control. In the interest of full disclosure, the average age of these pitchers was 28 in Year 1 and 29 in Year 2, so they had relative youth on their side, something with Johnson doesn’t have.
As a group, these pitchers went from an ERA+ of 83 to an ERA+ of 95 the following season, and cut their runs allowed total by over 1/2 run per nine innings.
Now, looking at this in this way could put us in danger of selection bias, as the people who were really bad may have been culled from the Year 2 sample, which would boost that performance accordingly, as the 14000 inning shortfall may possibly indicate. Therefore, I pared the list down to people who pitched at least 200 innings in both of the seasons, and matched the innings totals so that everyone contributed to each sample equally, which will again help remove any possible selection bias (better pitchers getting more innings in year two, etc.)
So this group went from an ERA+ of 84 to an ERA+ of 100 (exactly league average), and cut their runs allowed by over 2/3 of a run per nine innings.
The point here is that basing your player evaluation on one season of ERA is not smart. Single season ERA can have a lot of fluctuation, and is a poor predictor of future performance.
Replacing Johnson with a combination of Jeff Karstens, Darrell Rasner, etc., is probably a 2-3 win downgrade. I re-ran my simulations from last week with them substituted for Johnson and the Yankees went from allowing 809 runs a season to 833. That’s based on a pretty optimistic projection for Johnson, but it’s still something that needs to considered.
Now, if the Yankees do end up trading Johnson and saving a decent amount of money, it may still make sense, particularly if they can use the money or extra depth to shore up another are of weakness. Apparently a Roger Clemens return is one such option. I guess we’ll just wait and see what happens, but it is imperative to me that the Yankees don’t just give RJ away.
Update: Darren asked me to look at players over the age of 37. Since I don’t want to give Darren the satisfaction of doing precisely what he asked, I pulled the same data above but restricted the pool of pitchers to those above the age of 35 in their season of an ERA+ of less than 90.
It seems that the bounceback effect is real, even for the older group.
Another update.
I would think that you would also have to adjust these sample sizes so they have the same innings pitched, because like you said the first time you did it, the truly bad players didn’t pitch (or saw their innings significantly drop in relief) the following year.
Jeteupthemiddle is of course correct, so here’s the same group of pitchers, restricted to those who threw at least 162 innings in each of the two seasons, with their innings totals matched up to have the same weight in both samples.
Once again, we see the bounceback in effect pretty clearly.
Page 1 of 1 pages: