"Man Versus Machine" and My Annual MLB "Wins" Predictions (April 3, 2013)

Time for my annual MLB forecast for 2013!  Every year I trot out my trusty regression model (built in 1992) and attempt again to predict wins for each team, and also challenge a group of informed baseball fan friends to try to beat me in what I call the “Man Versus Machine” competition.

The model itself has its roots in Bill James (and Yankees GM in the early 90’s, Gene “Stick” Michael), seriously predating Billy Beane and the Moneyball guys.  Those guys knew all about on base percentage and the like, but they lacked the modeling and wealth of data available in this century.

My model essentially has two variables, one for hitting, one for pitching.  The hitting variable is fairly fancy: “OPS”, which is the sum of on base percentage plus slugging percentage.  This is a variable that, 21 years ago, only the real die-hards calculated (with slide rules and abacuses).  Now it is the hitters’ “one true measure” that you can find everywhere.  The pitching variable is more straightforward: ERA.  Basically, I come up with a prediction for each team’s overall OPS and ERA and plug those numbers into the regression equation I developed (using 20 years of historical data) and voila, a forecast for Team Wins. 

The difficult part is to actually come up with the forecast for OPS and ERA for each team.  Here it gets a bit “granular”:  I make a prediction for OPS (or ERA) for each player on the team roster, and then also predict their number of plate appearances (or innings pitched).  Then I multiply the OPS (or ERA) by that player’s percentage of the team’s total plate appearances (or innings pitched), and then add up all the players to get to the total team.  Ah, the wonders of weighted averages!

So let’s say Robinson Cano had an OPS of .929, and he has been between ..871 and .929 for the last 4 years.  It is reasonable to conclude he will do about the same this year.  And I expect him to have about 700 plate appearances this year (he’s averaged 687 of late), which is about 10.4% out of the Yankees total (of about 6,250 expected team plate appearances).  I multiply the .929 times 10.4% to get .097, and then do the same thing for the other Yankee players, and add them all up to get the team OPS.  That process typically yields a team OPS number between .700 (say, for the Twins) and .800 (say, for the Red Sox).  Brute force, but it works pretty well!

And I do the same thing to predict team ERA….CC will have a 3.40 ERA in 200 innings, for example, and I do the same math for all the pitchers.  I end up with a Team OPS and a Team ERA which I plug into my equation and out pops Team Predicted Wins.

So, I did this for every team and the results are below.  You can see where my friends and I part company….for example, I think, the Tigers will be superb, not just very good, and the Pirates will be abysmal, not mediocre.  You can also see, in comparing to the 2012 stats, how offseason moves will change some team’s fortunes, for instance how much the Yankees hitting takes a beating, and how much improvement the Blue Jays will show by importing Dickey, Cabrera and half of the Marlins.

Despite their weaknesses, I still have the Yanks hanging on to win a wild and wooly AL East (see my Yankee prediction here:  http://www.borntorunthenumbers.com/2013/03/yankees-2013-prediction-is-it-1965-i_31.html).

Comments welcome!  And if you want the detailed models for any team, contact me at tom@borntorunthenumbers.com.

Play ball!

	2012	2012	2012	2013	2013		Average
	Actual	Actual	Actual	Proj	Proj.		of Informed
	OPS	ERA	Wins	OPS	ERA	Tom	Fans
AL EAST
New York	0.780	3.85	95	0.757	3.86	90	87
Boston	0.730	4.70	69	0.778	4.18	89	80
Tampa Bay	0.711	3.19	90	0.733	3.66	89	89
Toronto	0.716	4.64	73	0.751	4.05	86	88
Baltimore	0.728	3.90	93	0.744	4.09	83	86
AL CENTRAL
Detroit	0.757	3.75	88	0.800	3.78	101	93
Chicago	0.740	4.02	85	0.734	4.00	83	84
Kansas City	0.716	4.30	72	0.733	4.23	79	78
Cleveland	0.705	4.78	68	0.735	4.56	73	78
Minnesota	0.715	4.77	66	0.726	4.43	73	71
AL WEST
Los Angeles	0.764	4.02	89	0.778	4.04	92	92
Texas	0.790	3.99	93	0.771	4.15	88	89
Oakland	0.714	3.48	94	0.747	4.23	82	87
Seattle	0.665	3.76	75	0.693	4.07	73	77
Houston	0.673	4.56	55	0.708	4.93	60	62
NL EAST
Washington	0.750	3.33	98	0.741	3.46	95	95
Atlanta	0.709	3.42	94	0.726	3.87	84	91
Philadelphia	0.716	3.83	81	0.703	3.74	81	84
New York	0.701	4.09	74	0.703	4.32	71	74
Miami	0.690	4.09	69	0.700	4.70	63	67
NL CENTRAL
Cincinnati	0.726	3.34	97	0.743	3.61	92	91
Milwaukee	0.762	4.22	83	0.743	4.03	84	83
St. Louis	0.759	3.71	88	0.749	4.17	83	87
Chicago Cubs	0.680	4.51	61	0.696	4.51	66	70
Pittsburgh	0.699	3.86	79	0.707	4.63	66	81
NL WEST
San Francisco	0.724	3.68	94	0.731	3.56	90	92
Los Angeles	0.690	3.34	86	0.729	3.68	88	90
Arizona	0.746	3.93	81	0.730	3.91	84	81
Colorado	0.766	5.22	64	0.754	4.67	75	70
San Diego	0.699	4.01	76	0.687	4.31	67	74