Baseball Calculus

So after yet another stellar performance from Stephen Strasburg, he has climbed to 22nd on our pitcher rankings. Here's the top 25:

		E OPSA
Jonathan	Broxton	0.566
Carlos	Marmol	0.602
Luke	Gregerson	0.609
Mike	Adams	0.619
Joe	Nathan	0.623
Mariano	Rivera	0.624
Matt	Thornton	0.628
Tim	Lincecum	0.629
Billy	Wagner	0.631
Joakim	Soria	0.632
Brian	Wilson	0.641
Rafael	Soriano	0.645
Jose	Valverde	0.647
Andrew	Bailey	0.649
Adam	Wainwright	0.650
Josh	Johnson	0.651
Takashi	Saito	0.654
Francisco	Rodriguez	0.654
Ubaldo	Jimenez	0.654
Hong-Chih	Kuo	0.655
Neftali	Feliz	0.656
Stephen	Strasburg	0.658
Huston	Street	0.663
Roy	Halladay	0.664
Arthur	Rhodes	0.665

Since this metric values pitchers based upon expected perfromance per plate appearance, it is obviously biased towards relievers (who have well understood statistical advantages over starters). Thus, Strasburg's position is all the more impressive as he is the 5th ranked starter (ahead of Roy Halladay!).

In parallel to the Strasburg Watch I have been running, my Dad asked me to follow a less heralded rookie, Chad Huffman of the Yankees, who happens to hail from my hometown. In any case, Chad debuted last Sunday for the Yankees against our hometown Astros (surely a special thrill for him and his family – and my Dad who happened to be at the game). He went 1 for 4 with a walk and a strikeout. Coming into the game I had him projected with the universal rookie projection line (.319 OBP / .396 SLG / .715 OPS). After the game, his projection is little changed at (.322/.390/.712); not quite the impact of Strasburg’s debut. He hasn’t played since and I don’t know how long he’ll stay up with the club, but Dad I’ll keep you posted when/if he does.

Yesterday I looked at the top pitchers in the league and those whose stock had risen the most since the start of the season, so today I figured I would look at the hitters. So first off, here are the top hitters according to our models in terms of expected OPS right now against an 'average' pitcher.

Albert	Pujols	0.9784
Miguel	Cabrera	0.9339
Kevin	Youkilis	0.9069
Barry	Bonds	0.9036
Lance	Berkman (L)	0.8921
Prince	Fielder	0.8878
Alex	Rodriguez	0.8875
Ryan	Braun	0.8844
Justin	Morneau	0.8834
Adrian	Gonzalez	0.8828
Joey	Votto	0.8822
Joe	Mauer	0.8768
Adam	Dunn	0.8756

A couple of notes. First of all, notice that the #4 hitter on this list is the 45 year old Barry Bonds, who has been out of baseball for 3 years. I've been saying for years that it was absolutely crazy that nobody signed him 3 years ago, and the numbers are saying that it is still crazy. Of course, Bonds is such an outlier, that any projection of how good he might be right now is going to be extremely high variance, but this analysis is at least saying he is worth a shot.

I should also note, that this set of projections is actually split by hand for switch hitters, (given how the models are built it was just easier that way), so the Lance Berkman in this list is just the left-handed incarnation of him. Right-handed Berkman is a mere mortal .733 hitter.

As for the hitters whose stock has risen the most.

		as of
		1-Apr	15-Jun	delta
Andres	Torres (L)	0.636	0.732	0.095
Brennan	Boesch	0.715	0.797	0.082
John	Jaso	0.714	0.787	0.073
Jason	Heyward	0.715	0.775	0.061
Chris	Heisey	0.715	0.775	0.060

 Again, the Andres Torres here is the left-handed incarnation, which this year has an OPS 299 points higher than his career numbers, as opposed to right-handed Torres who is down 135 points

Last week I commented on the huge 80+ point swing Stephen Strasburg’s sparkling debut had on his expected ability according to the baseballcalculus player ability model. After his 2^nd start (5 1/3 innings, 2 hits, 5 walks, 8 K’s), Strasburg’s current ability estimate (expected OPS against) is .691, moving him up to 51^st from 64^th in the overall rankings. So after two starts, Strasburg has now made the 6^th largest  improvement in estimated ability among all pitchers from opening day. The list of players ahead of him is:

		E OPSA as of
		4/1/2010	6/14/2010	change
Jonny	Venters	0.775	0.677	-0.098
Colby	Lewis	0.824	0.731	-0.093
Mat	Latos	0.780	0.692	-0.088
Francisco	Rodriguez	0.775	0.689	-0.086
Matt	Belisle	0.828	0.742	-0.086
Stephen	Strasburg	0.775	0.691	-0.084

Not too surprising for those familiar with this sort of analysis, hot rookies and a couple relatively young players doing a lot better this year than they had previously.  And by the way, that Francisco Rodriguez is not the K-Rod you are familiar with, but the new K-Rod in Anaheim, whose given up 3 hits and recorded 11 K's in 8 MLB innings so far. Now for the flip side, let's see the players who have dropped the most

		E OPSA as of
		4/1/2010	6/14/2010	change
Trevor	Hoffman	0.665	0.747	0.082
Ryan	Rowland-Smith	0.745	0.818	0.074
James	Russell	0.775	0.845	0.071
Rich	Harden	0.695	0.758	0.063
Randy	Williams	0.753	0.805	0.052

For the most part a collection of pitchers that have had just dreadful years (at least compared to their historical performance). The most interesting of these might be Russell, whose stats look reasonably respectable, except for the 7 HR allowed in 21 career innings. For reference, here are the top 5 as they stand right now:

First	Last	E OPSA
Jonathan	Broxton	0.566
Carlos	Marmol	0.604
Luke	Gregerson	0.609
Joe	Nathan	0.623
Tim	Lincecum	0.623

Gregeorson is the obvious surprise on this list, but this really shouldn't be that shocking since he was ranked in the top 20 to start the season.

So after yet another blogging hiatus, here I am. I know I’ve said it before, but we’ll try to keep some semi-regular posts coming here, breaking down some of the insights that our models generate. At least now we’ve got the models regularly updating player, game and season projections (although we’re still ironing out some kinks trying to get the player projections updating on the site), so it should be much easier to slice and dice these and get at the meat of it all. As always, I hope you enjoy what is here, and feel free to write me at brad at baseballcalculus dot com at any time with questions or comments.

A lot has been written about Stephen Strasburg's impressive MLB debut last Tuesday, but for what it is worth, here's my two cents:

As some of you know, the player ability estimates and subsequently all of the predictions created here at baseballcalculus.com, are based upon historical Major League data. Thus, if a player has no Major League experience, we assume he is just an average pitcher. It can get tricky to summarize the prior and posterior distributions that represent this understanding of a player’s ability (see the research link for more on that), but to simplify, the expected OPSA (OPS against – or expected OPS by opposing batters) of any new pitcher is .775 according to these models. Thus, that was Stephen Strasburg’s expected ability according to our models as of Tuesday afternoon.

After Tuesday’s game though, most notably the 14 strikeouts with no walks, Strasburg’s expected OPSA immediately fell to .699, moving him from the 446^th ranked pitcher in (or out of – since we maintain estimates for retired pitchers) baseball, to the 64^th, just ahead of Tim Hudson. A large component of this was his expected K rate jumping from 16.8% to 26.3%. That’s pretty impressive for about two hours work. By comparison, the next best single day change in expected OPSA for a pitcher this season was .039 points by Dodgers rookie John Ely (moving his expected OPSA from .779 to .740) on May 6^th after a 7 inning, 4 hit, 1 run, 0 walk, 7 strikeout performance.

We’ll be sure to keep track of Strasburg’s projections as he logs more Major league starts. If he can reduced his expected OPSA another .076 points he'll be the top rated starter in the league, and number 3 pitcher overall after Jonathan Broxton and Carlos Marmol.

All- The 2010 projections are up and better than ever. We now provide game predictions as well as batter, pitcher, and team projections. We have also added sortable functionality and will be updating the projections regularly, so keep checking in. As always, please email me at brad at baseballcalculus dot com with any questions or comments, and especially if any of the numbers seem out of line!

Welcome to baseballcalculus.com! Forecasts - Click on the following links to see forecasted Batter and Pitcher statistics for 2009.

Batters                                                          Pitchers

Coming soon

• In-season updates of these player stat predictions along with estimated variance in player performance.
• Predictions of the results of every game including the probability of each team winning, the expected number of runs scored, predicted player stats by game, and analysis of these results versus live gambling lines.
• Forecasted standings for the entire league updated regularly along with an assessment of the likelihood of each team to make the playoffs, win the pennant, and win the World Series.
• Optimization tools
  • To evaluate the value of different players to different teams and examine which possible trades and acquisitions make the most sense for specific teams
  • To determine how a team's performance would change with different lineups or different strategies
• And much more, so check back often!

How we do it - These predictions are based on a series of stochastic models developed in my doctoral research at Stanford University. (For more details visit my academic website at www.stanford.edu/~null.) Feedback - As with any model of complex real-world events, these models are a perpetual work-in-progress, so your feedback is welcome and integral in improving them.