For those who don't know, OPSBI was created by Jim Bowden, the former Reds/Nats GM. The basic idea is this: you take the players on-base percentage, and add his slugging percentage (thus, the OPS). Now, eliminate the decimal point (or multiply by 1000, however you prefer to think about it). Then, add the player's RBI. That's it.
So, first, I want to point out why this is stupid. First of all, it doesn't measure anything. At all. It arbitrarily adds a rate stat (two rate stats, actually) to a counting stat, without any consideration for why. It doesn't correlate to anything. As we know, OPS has a reasonable correlation with run-scoring. RBI represent actual runs batted in (though not runs scored). So, you're adding something which correlates to run-scoring with something that actually is one-half of run-scoring. Now, someone clever could probably make the argument that if you added Runs to this, you'd solve some of that problem. But here's the question - why add anything to OPS at all? I don't get it.
Anyway, Bill James says, "For a statistic to have value, it has to be meaningful with reference to something other than its own formula" (The New Bill James Historical Abstract, in the player comment on Craig Biggio). OPSBI fails that test.
Of course, there are still defenders out there. I don't feel like looking for articles right now, but I remember reading at least two of them last offseason. Here's the thing they'll say, more or less: "Who cares if it doesn't measure anything - it gives the right answer!" Okay, well, in my mind, it gives the "right" answer - that is, it affirms (much of the time) the conclusion sportswriters have already made. But I personally believe that I can throw out a lot of other BS stats that will do the same thing (more or less). Anyway, I'll be looking at the last five years of data (not including 2012, of course, since we're still underway), and looking at the top MVP-finishers (non-pitchers only) for each season, and comparing them by different metrics. Those metrics are:
MVP Finish - where did the player finish in subjective MVP voting?
OPSBI - of course.
RunAvg - (R+RBI)/AB
ButTheKitchenSink*Games - Games*(RBI+R+TB+BB+HBP+SB)/(3*AB) ; I already posted about this before.
ButTheKitchenSink(rate) - (RBI+R+TB+BB+HBP+SB)/(3*AB) ; same thing, but as a rate stat (scaled to batting average); really, what this is, is RunAvg+BattingAvg+SecondaryAvg, and divided by three so that it looks like the players batting average.
StolenHomes - SB+HR
TripleCrowns - (3*HR)+((1000*Avg-100)/2)+(RBI)
HitByBallSacs - because I couldn't resist: HBP+BB+SF+SH
rWAR - because, wouldn't it be fun if I used a metric that actually seemed to represent real value?
One last note before presenting: in 2008, Manny Ramirez only played 53 games in the NL, so only his NL stats are included. Also, I hope this formats okay. Here come the charts:
| 2011 NL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Braun | 1 | 1105 | .391 | 57.9 | .386 | 66 | 343.0 | 66 | 7.7 |
| Kemp | 2 | 1112 | .400 | 63.7 | .395 | 79 | 366.0 | 87 | 7.8 |
| Fielder | 3 | 1101 | .378 | 62.2 | .384 | 39 | 345.5 | 123 | 4.3 |
| Upton | 4 | 986 | .326 | 54.2 | .341 | 52 | 294.5 | 82 | 5.7 |
| Pujols | 5 | 1005 | .352 | 50.0 | .340 | 46 | 322.5 | 72 | 5.1 |
| 2011 AL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Ellsbury | 1 | 1033 | .339 | 54.9 | .347 | 71 | 329.5 | 69 | 8.0 |
| Bautista | 2 | 1159 | .405 | 64.6 | .433 | 52 | 340.0 | 142 | 7.7 |
| Granderson | 3 | 1035 | .437 | 62.3 | .400 | 66 | 332.0 | 108 | 5.3 |
| Cabrera | 4 | 1138 | .378 | 62.3 | .387 | 32 | 337.0 | 116 | 7.3 |
| Cano | 5 | 1000 | .356 | 52.1 | .327 | 36 | 325.0 | 58 | 5.2 |
| 2010 NL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Votto | 1 | 1137 | .400 | 60.4 | .403 | 53 | 349.0 | 101 | 6.7 |
| Pujols | 2 | 1129 | .397 | 63.6 | .400 | 56 | 358.0 | 113 | 7.3 |
| C. Gonzalez | 3 | 1091 | .388 | 53.3 | .367 | 60 | 353.0 | 49 | 5.8 |
| A. Gonzalez | 4 | 1005 | .318 | 52.8 | .330 | 31 | 362.0 | 101 | 4.1 |
| Tulowitzki | 5 | 1044 | .391 | 44.6 | .365 | 38 | 306.5 | 59 | 6.5 |
| 2010 AL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Hamilton | 1 | 1144 | .376 | 49.6 | .373 | 40 | 343.5 | 53 | 8.4 |
| Cabrera | 2 | 1168 | .432 | 61.4 | .409 | 41 | 366.0 | 100 | 6.1 |
| Cano | 3 | 1023 | .339 | 52.3 | .327 | 32 | 326.5 | 70 | 7.8 |
| Bautista | 4 | 1119 | .409 | 66.3 | .412 | 63 | 362.0 | 114 | 6.6 |
| Konerko | 5 | 1088 | .365 | 54.1 | .363 | 39 | 345.0 | 83 | 4.3 |
| 2009 NL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Pujols | 1 | 1236 | .456 | 72.6 | .454 | 63 | 392.5 | 132 | 9.4 |
| H. Ramirez | 2 | 1060 | .359 | 53.9 | .357 | 51 | 325.0 | 76 | 7.1 |
| Howard | 3 | 1072 | .399 | 59.5 | .372 | 53 | 370.5 | 87 | 3.5 |
| Fielder | 4 | 1155 | .413 | 65.9 | .407 | 48 | 382.5 | 128 | 6.0 |
| Tulowitzki | 5 | 1022 | .355 | 54.6 | .362 | 52 | 304.5 | 85 | 6.3 |
| 2009 AL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Mauer | 1 | 1107 | .363 | 50.9 | .369 | 32 | 3028.5 | 83 | 7.6 |
| Teixeira | 2 | 1070 | .369 | 56.7 | .363 | 41 | 346.0 | 98 | 5.1 |
| Jeter | 3 | 937 | .273 | 46.3 | .302 | 48 | 269.0 | 82 | 6.4 |
| Cabrera | 4 | 1045 | .326 | 53.4 | .334 | 40 | 333.0 | 74 | 4.7 |
| Morales | 5 | 1032 | .343 | 50.8 | .334 | 37 | 329.0 | 56 | 4.0 |
| 2008 NL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Pujols | 1 | 1230 | .412 | 63.5 | .429 | 44 | 368.5 | 117 | 9.0 |
| Howard | 2 | 1027 | .411 | 59.0 | .364 | 49 | 367.5 | 90 | 1.5 |
| Braun | 3 | 994 | .324 | 49.3 | .326 | 51 | 322.5 | 52 | 4.3 |
| M. Ramirez | 4 | 1285 | .476 | 25.3 | .478 | 19 | 285.0 | 42 | 3.4 |
| Berkman | 5 | 1092 | .397 | 62.9 | .396 | 47 | 320.0 | 111 | 6.6 |
| 2008 AL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Pedroia | 1 | 952 | .308 | 48.1 | .306 | 37 | 280.0 | 73 | 6.8 |
| Morneau | 2 | 1002 | .363 | 53.7 | .330 | 23 | 325.0 | 89 | 3.9 |
| Youkilis | 3 | 1073 | .383 | 52.9 | .365 | 32 | 329.0 | 83 | 6.0 |
| Mauer | 4 | 949 | .341 | 46.4 | .318 | 10 | 267.0 | 97 | 5.3 |
| Quentin | 5 | 1065 | .408 | 50.8 | .391 | 43 | 316.0 | 89 | 5.1 |
| 2007 NL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Rollins | 1 | 969 | .325 | 53.5 | .331 | 71 | 314.0 | 62 | 6.0 |
| Holliday | 2 | 1149 | .404 | 60.2 | .381 | 47 | 379.0 | 77 | 5.8 |
| Fielder | 3 | 1132 | .398 | 63.2 | .400 | 52 | 363.0 | 108 | 3.4 |
| Wright | 4 | 1070 | .364 | 60.4 | .377 | 64 | 329.5 | 107 | 8.1 |
| Howard | 5 | 1112 | .435 | 59.2 | .411 | 48 | 364.0 | 119 | 2.8 |
| 2007 AL | MVP | OPSBI | RunAvg | BTKS | BTKSr | StoHos | 3Crowns | BallSacs | rWAR |
| Rodriguez | 1 | 1223 | .513 | 73.6 | .466 | 78 | 421.0 | 125 | 9.2 |
| Ordonez | 2 | 1168 | .430 | 60.9 | .388 | 32 | 376.5 | 83 | 6.9 |
| Guerrero | 3 | 1075 | .373 | 53.1 | .354 | 29 | 341.0 | 86 | 4.3 |
| Ortiz | 4 | 1183 | .424 | 62.6 | .420 | 38 | 353.0 | 118 | 6.1 |
| Lowell | 5 | 999 | .338 | 48.2 | .313 | 24 | 324.0 | 64 | 4.6 |
So, what often happens with a chart like this is, people either skip it and wait for the conclusion, or they read it and go, "so what?" If you're in the former group, that's annoying, because charts take the most time to make. So authors are upset at you for not reading the chart. If that's you, go back and look at them. We'll wait.
Okay, now that everyone's caught up, what the heck does any of this mean?
Well, if the goal of OPSBI is to correlate to a player's true value, we have to ask, "how do we best measure a player's true value?" Thus, the first and last columns of the chart. The last, WAR, is a statistical measure. The first, MVP voting, is a completely subjective measure. If OPSBI is so good at perceiving value, we'd really like it to have some correlation to one of these two or the other.
The problem is, it doesn't. Not at all. So either both the MVP voters and WAR are wrong, while OPSBI is the true best measure... or OPSBI is crap. Now, I'll admit freely that both WAR and voters take defense into account, which OPSBI doesn't... still though, that's not (for the most part) how MVPs are won, or how WAR is decided, since offense bears so much more weight.
For instance, if OPSBI is such a good measure, why would Jacoby Ellsbury have finished above Jose Bautista last year in MVP voting? Bautista was the better offensive player... even by these made up metrics. I just don't get what OPSBI is doing that different from... ANY of these other things I made up. Seriously. Well, maybe not HitByBallSacs, but that's just too hilarious to not include. Anyway, the others do just as good of a job predicting WAR and predicting MVPs... maybe better. So why OPSBI? I see no reason, since it doesn't even reflect voter tendencies.
No comments:
Post a Comment