AJ Burnett has the reputation for being able to throw a no-hitter or a complete bomb every time he takes the mound. His stuff is excellent but often unpredictable. If he locates his fastball and can command his “hook”, as he likes to call it, then batters beware. If he can’t, then he sometimes gets lit up in spectacular fashion. Blue Jays fans and Yankees fans have a name for this Jekyl and Hyde act: Good AJ/Bad AJ.
On Twitter two weeks ago, JMK the Overshare asked if any pitcher other than Burnett lived entirely by the “good/bad” moniker. Oliver Perez, Carlos Zambrano and Cole Hamels were tossed out, and I suggested Josh Beckett. Is this fair? Let us find out!
First, the methodology. I’ve evaluated every start by Burnett and Beckett in 2007, 2008, 2009 and 2010 and classifying them as either Good, Bad or Average. Over that time period, Burnett made 99 starts and Beckett made 95. As a side note, would you have guessed that Burnett’s made more regular season starts than Beckett since the start of the 2007 season? I would not have.
Using Baseball Reference’s fantastic Play Index, I searched between 2007 and 2010 and discovered that there were 15,456 total starts. Of those starts, 3,567 had a WPA of over 0.200, a percentage of 23.1%. Conversely, 3,878 had a WPA of under -0.200, a percentage of 25.1%. Using these cutoffs, I classified Burnett and Beckett’s starts. Any start with a WPA of over 0.200 received a “Good”. Any start with a WPA of below -0.200 would receive a “Bad”. Anything in between would receive an “Average”.
[image title="burnett vs beckett" size="full" id="17461" align="center" linkto="full" ]
The average WPA for Burnett’s 99 starts is 0.045, and the standard deviation on that range is 0.261. For Beckett, the average WPA for his 95 starts is 0.068, or .023 higher, and the standard deviation on that range is 0.25. Off the bat, this shows us that Beckett averages a higher WPA per start and has a less extreme distribution of range. By applying the .200/-.200 labels, we see that in 2007-2010 Burnett had 19 Bad starts, 47 Average starts, and 33 good Starts. In the same time period, Beckett had 18 Bad starts, 45 Average starts, and 32 good starts. They appear strikingly similar.
From there I ran the same calculations for Jon Lester and CC Sabathia, neither of whom have the same Jekyl and Hyde reputation. Here’s what I found:
[image title="all four" size="full" id="17463" align="center" linkto="full" ]
Here we see, and note that I’m excluding Lester’s 2007 campaign, that Sabathia has a roughly similar Average Start percentage, but less Bad starts and more Good starts. Lester, however has both a higher Good and Bad percentage. We might conclude here that Lester deserves the Jekyl and Hyde moniker more than any of the four, but I think that would be inaccurate. Here’s why:
[image title="all four " size="full" id="17552" align="center" linkto="full" ]
When you dive into the underlying information behind the bad starts, you see that Burnett has a higher average Bad Start score than anyone at -0.337. Behind him is Beckett with a -0.329. On the Good side, Sabathia has the highest Average score with 0.375 with Lester right behind him at 0.361. Burnett comes in third, with Beckett in last. To me this shows that Burnett does indeed have a high upside and downside risk, confirming what fans have long thought. Josh Beckett has a strong element of Jekyl/Hyde in him as well, something that was illustrated perfectly last night at Fenway when he dominated through the first few innings before collapsing in the sixth. CC Sabathia and Jon Lester have fantastic Good scores, but seem to avoid the huge explosions. In fact, Sabathia’s Bad score would be dramatically lower if you excluded his early season struggles in 2007 and 2008.
What do you think? Is there a better way to quantify the Jekyl and Hyde act for starting pitchers?
If you enjoyed this article, please consider sharing it!
6 Responses to Good Josh, Bad Josh?
Leave a Reply
TIQ IQ
Nice post, and it shows that a few big games in October make a huge difference with how these two are regarded. Beckett’s best has been more consistent, but overall they are very similar pitchers. I just did a post the other day about how maybe the Good AJ/Bad AJ is going away based on his early results. I doubt it will happen but it would be nice. Beckett the myth is much better than Beckett the pitcher.
Thanks. As an FYI, this data doesn’t include Beckett’s start from yesterday. I didn’t check the WPA, but I’m guessing it was below -0.200, which means he would have the exact same number of bad starts as Burnett.
[...] full post on TYU var AdBrite_Title_Color = 'FFFF66'; var AdBrite_Text_Color = 'FFFFFF'; var [...]
Beckett is still a stud. Glad to see us pound him.
I think its a stretch to put CC in that class because he’s a relatively slow starter (though not this year) and so his bad starts may be clustered. Likewise I’d wager a guess that a fair number of Lester’s bad starts are clustered in his first two years. A more reliable means of quantification might be to determine how guys like Beckett and Burnett string together starts. Do they alternate good and bad starts frequently? Do they have hot and cold streaks? Etc etc. It’s a nice post and an interesting conversation, I just think if you tweak your parameters you might get data that does a better job of supporting or refuting the premise.
Yeah, that’s all fair. I view this post as more of a conversation starter than the definitive word on the subject. FWIW, I wasn’t putting Sabathia in the Good/Bad class. I was using him as a cross-reference since he has more of a reputation as a consistent starter, as opposed to a jekyl/hyde kinda guy.