March 25, 2007
Probabilistic Model of GDPs
A few days ago I introduced the idea of a probabilistic model of Ground into Double Plays (GDP). The probabilistic model of range just measures the ability to turn a ball into an out. For infielders, however, they're often asked to turn a ground ball into more than one out. The idea is to take a very specific situation; ground ball hit, man on first, less than two out and build a model that measures both plays made and GDP turned. With that model, we can ask which fielders perform well in that situation.
In building this model, I left parks out of the parameters. Basically, I thought the sample size would be too small if I left the parks in. This probably hurts the three teams that play of artifical turf.
Let's start by looking at the ability of shortstops to start a double play. The following table looks at three indexes for each fielder. The Plays Made (PM) index measures Plays Made / Predicted Plays Made. This measures the fielder's ability to turn a ball into an out. The GDP index does the same for ground double plays. Does the fielder start the expected number of double plays? And finally, an outs index that looks at the total number of outs accured to the fielder on these balls in play. It could be a fielder is making up for a lack of range by being really good at starting GDPs, or vice versa. Remember, this says nothing about the pivot man or the receiver at first base. In this context, we're only looking at the fielder who starts the play.
Probabilistic Model of GDPs, Ground Balls, Man on First, Less than Two Out, Shortstops Starting GDP (2006 Data Used to Build Model)
Player | Ground Balls In Play | Actual Plays Made | Predicted Plays Made | PM Index | Actual GDP | Predicted GDP | GDP Index | Actual Outs | Predicted Outs | Outs Index |
Craig Counsell | 183 | 40 | 32.14 | 124.44 | 32 | 21.23 | 150.72 | 72 | 53.38 | 134.89 |
Khalil Greene | 219 | 53 | 45.27 | 117.08 | 31 | 28.68 | 108.07 | 84 | 73.95 | 113.59 |
Stephen Drew | 121 | 26 | 23.68 | 109.82 | 17 | 14.29 | 118.92 | 43 | 37.97 | 113.25 |
Clint Barmes | 283 | 56 | 53.75 | 104.18 | 42 | 33.03 | 127.16 | 98 | 86.78 | 112.92 |
Juan Uribe | 274 | 58 | 50.75 | 114.29 | 34 | 31.85 | 106.75 | 92 | 82.60 | 111.38 |
Hanley Ramirez | 352 | 71 | 67.58 | 105.05 | 49 | 43.18 | 113.48 | 120 | 110.76 | 108.34 |
Miguel Tejada | 342 | 76 | 72.84 | 104.33 | 52 | 45.35 | 114.66 | 128 | 118.20 | 108.30 |
David Eckstein | 298 | 61 | 56.74 | 107.50 | 38 | 34.92 | 108.81 | 99 | 91.67 | 108.00 |
Jack Wilson | 290 | 64 | 62.47 | 102.45 | 46 | 39.39 | 116.77 | 110 | 101.86 | 107.99 |
Rafael Furcal | 396 | 92 | 85.54 | 107.55 | 58 | 55.24 | 104.99 | 150 | 140.78 | 106.55 |
Bill Hall | 228 | 55 | 50.83 | 108.21 | 33 | 32.90 | 100.30 | 88 | 83.73 | 105.10 |
Bobby Crosby | 237 | 44 | 42.12 | 104.47 | 29 | 27.37 | 105.94 | 73 | 69.49 | 105.05 |
Alex Gonzalez | 244 | 50 | 51.18 | 97.69 | 38 | 33.41 | 113.74 | 88 | 84.59 | 104.03 |
Jimmy Rollins | 333 | 70 | 69.50 | 100.72 | 48 | 44.05 | 108.97 | 118 | 113.54 | 103.92 |
Carlos Guillen | 303 | 65 | 64.24 | 101.18 | 45 | 41.63 | 108.10 | 110 | 105.87 | 103.90 |
Adam Everett | 309 | 66 | 63.51 | 103.91 | 41 | 40.15 | 102.11 | 107 | 103.67 | 103.22 |
Ronny Cedeno | 240 | 49 | 46.28 | 105.87 | 28 | 29.54 | 94.79 | 77 | 75.82 | 101.55 |
Michael Young | 410 | 88 | 86.89 | 101.28 | 54 | 55.41 | 97.46 | 142 | 142.30 | 99.79 |
Jason A Bartlett | 214 | 50 | 47.40 | 105.49 | 28 | 31.86 | 87.90 | 78 | 79.25 | 98.42 |
Jose Reyes | 304 | 67 | 66.70 | 100.45 | 40 | 43.66 | 91.62 | 107 | 110.36 | 96.96 |
Omar Vizquel | 297 | 64 | 65.48 | 97.73 | 42 | 44.00 | 95.45 | 106 | 109.49 | 96.82 |
John McDonald | 170 | 25 | 26.52 | 94.26 | 18 | 17.91 | 100.50 | 43 | 44.43 | 96.77 |
Orlando Cabrera | 319 | 58 | 60.40 | 96.03 | 38 | 39.02 | 97.38 | 96 | 99.42 | 96.56 |
Felipe Lopez | 319 | 67 | 64.47 | 103.93 | 34 | 40.58 | 83.78 | 101 | 105.05 | 96.15 |
Angel Berroa | 337 | 69 | 72.29 | 95.44 | 46 | 47.91 | 96.02 | 115 | 120.20 | 95.67 |
Jhonny Peralta | 357 | 82 | 84.45 | 97.10 | 50 | 55.63 | 89.88 | 132 | 140.08 | 94.23 |
Alex Cora | 127 | 30 | 32.94 | 91.07 | 21 | 22.16 | 94.76 | 51 | 55.10 | 92.56 |
Marco Scutaro | 146 | 34 | 37.58 | 90.47 | 24 | 25.22 | 95.15 | 58 | 62.81 | 92.35 |
Edgar Renteria | 347 | 63 | 67.67 | 93.10 | 39 | 43.19 | 90.30 | 102 | 110.86 | 92.01 |
Yuniesky Betancourt | 350 | 60 | 66.40 | 90.36 | 43 | 46.02 | 93.44 | 103 | 112.42 | 91.62 |
Julio Lugo | 182 | 32 | 35.29 | 90.68 | 21 | 23.66 | 88.74 | 53 | 58.95 | 89.90 |
Ben T Zobrist | 131 | 25 | 28.09 | 89.00 | 17 | 18.93 | 89.82 | 42 | 47.02 | 89.33 |
Juan Castro | 146 | 23 | 25.18 | 91.33 | 13 | 15.27 | 85.11 | 36 | 40.46 | 88.98 |
Derek Jeter | 336 | 63 | 70.96 | 88.79 | 40 | 45.81 | 87.32 | 103 | 116.77 | 88.21 |
Royce Clayton | 234 | 43 | 47.02 | 91.44 | 20 | 29.67 | 67.40 | 63 | 76.70 | 82.14 |
Aaron W Hill | 108 | 16 | 20.33 | 78.70 | 7 | 12.82 | 54.62 | 23 | 33.14 | 69.39 |
Notice how few chances fielders get to turn GDPs. On the best teams, they get a little over two chances a game. Secondly, Arizona does a good job of picking out shortstops, as Counsell and Drew are near the top of the list. And if you don't like Derek Jeter, here's another area where you can pick on him.
The other thing that strikes me about the list is that shortstops who are good at making plays are also the ones good at starting double plays. Ronny Cedeno is unusual in that he's good at getting an out, but didn't do well starting DPs. Could it be that Todd Walker was just a poor pivot man? I hope further research using these models will help answer that question.
It's virtually useless without looking at the DP partner. Are Drew and Counsell really great, or is it that Orlando Hudson is so good on the receiving end. Interesting start to an idea, but that's all.
David, indeed an interesting start. It looks like your using BIS's vector data, so you've got a good data set from which to start. The problem I see here is that you're giving the SS credit for things that are out of his control. On a ground ball, the SS's job is to get to the ball (range), pick up the ball (hands), and throw it to the 2B covering the bag (arm). The rest is up to the 2B covering (to catch the throw, pivot, and throw to first) and 1B (catch the ball from 2B). You might counter that it's the SS's job to do all of his assignments *quickly*, so that the 2B and 1B can do their jobs. (That's going to be a function of how far away he has to range and how far of a throw he has to make, in addition to his range and arm skills.) You'd be right...
Then, there's the issue of runner speed and handedness. Try to make a good measure and all anyone tells you is what's wrong with it!
I didn't need another reason to hope Derek Jeter decides to join an international humanitarian aid organization (bad idea to wish harm on anyone) but thanks.