Beat the Streak Picks

Here are the top picks my programs produced for use in Beat the Streak. This post mostly explains the ideas behind the calculations. In addition, this post shows tests on the Neural Network (NN). This post discusses an NN that includes the ballpark. I recently update the models, and the results of those tests are here.

For 2017, I am just going to publish the Log5 hit averages and the NN probabilities with parks factored in. I am keeping track of the results here.

First, the Log5 Method picks:

0.289 — Starlin Castro batting against Adam Wainwright
0.287 — Daniel Murphy batting against Jerad Eickhoff
0.280 — Zack Cozart batting against Wily Peralta
0.276 — Jacoby Ellsbury batting against Adam Wainwright
0.274 — Chase Headley batting against Adam Wainwright
0.273 — Wil Myers batting against Bartolo Colon
0.272 — Cesar Hernandez batting against Gio Gonzalez
0.271 — Dee Gordon batting against Matt Harvey
0.271 — Corey Seager batting against Taijuan Walker
0.270 — Eugenio Suarez batting against Wily Peralta
0.270 — J.T. Realmuto batting against Matt Harvey

I’m surprised to see so many batters against Adam Wainwright today. Those Yankees hitters have done extremely well this season, and Wainwright got knocked around a bit this year. Daniel Murphy, whose hit streak ended yesterday, still ranks very high.

The NN with Park produces this top ten:

0.287, 0.734 — Daniel Murphy batting against Jerad Eickhoff.
0.271, 0.701 — Dee Gordon batting against Matt Harvey.
0.268, 0.700 — DJ LeMahieu batting against Jeff Samardzija.
0.260, 0.699 — Jose Altuve batting against Jharel Cotton.
0.263, 0.696 — Xander Bogaerts batting against Alexander Cobb.
0.271, 0.696 — Corey Seager batting against Taijuan Walker.
0.270, 0.695 — J.T. Realmuto batting against Matt Harvey.
0.289, 0.694 — Starlin Castro batting against Adam Wainwright.
0.258, 0.691 — Dustin Pedroia batting against Alexander Cobb.
0.257, 0.690 — Yunel Escobar batting against Ian Kennedy.

Of the Yankees, only Starlin Castro survives the cut. Early in the season, the batter three-year hit average tends to dominate the calculation, so Headley and Ellsbury drop out. There really is a huge difference today between the two systems. It will be interesting to see which one does better on the day.

Once again, Daniel Murphy is the consensus pick.

The NN did manage a hit streak of nine, which is pretty good. So far, the top pick of the NN with Park generated a hit in a game 10 out of 12 times, an .833 success rate. The predicted success rate is .729, so at this early stage, the NN is doing very well. The probability of getting at least 10 of 12 correct would be .328. Nothing outstanding, simply good luck so far. As yesterday demonstrates, even the best pick is going to fail about 25% of the time. Good luck!

from baseballmusings.com http://ift.tt/2oAcDng