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 updated 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. I added a graph that gives a visual representation of the probability and success each day.
First, the Log5 Method picks:
0.333 — Jean Segura batting against Tyler Anderson
0.328 — Andrelton Simmons batting against Bartolo Colon
0.321 — Travis Shaw batting against Tom Milone
0.318 — Corey Dickerson batting against Nick Martinez
0.314 — Wilmer Flores batting against Zach Davies
0.313 — Daniel Murphy batting against Jeff Samardzija
0.313 — David Peralta batting against Ivan Nova
0.311 — Hernan Perez batting against Tom Milone
0.310 — Ryan Zimmerman batting against Jeff Samardzija
0.310 — Eric Sogard batting against Tom Milone
0.310 — Manuel Pina batting against Tom Milone
0.310 — Chris Owings batting against Ivan Nova
For the record, the highest predicted hit average for a batter against a starting pitcher today belonged to Mike Trout, now on the disabled list, .337. Segura is likely a very good pick as the game is being played in Colorado. Note the Log5 doesn’t know anything other than hit averages.
Here is how the NN with Park ranks the players:
0.333, 0.759 — Jean Segura batting against Tyler Anderson.
0.313, 0.750 — Daniel Murphy batting against Jeff Samardzija.
0.318, 0.737 — Corey Dickerson batting against Nick Martinez.
0.313, 0.732 — David Peralta batting against Ivan Nova.
0.280, 0.730 — Charlie Blackmon batting against Ariel Miranda.
0.302, 0.724 — Robinson Cano batting against Tyler Anderson.
0.284, 0.723 — Xander Bogaerts batting against Jose Quintana.
0.310, 0.721 — Chris Owings batting against Ivan Nova.
0.310, 0.720 — Ryan Zimmerman batting against Jeff Samardzija.
0.328, 0.720 — Andrelton Simmons batting against Bartolo Colon.
0.294, 0.716 — Starlin Castro batting against Chris Tillman.
0.300, 0.716 — Brandon Drury batting against Ivan Nova.
Trout came down lower on this list, .729. Three other injured players would have made the list; A.J. Pollock, Yunel Escobar, and Justin Turner. With the park added in (the three-year hit average at Coors is a whopping .270), Segura comes out on top again. One last note on Segura, his numbers against a starting pitcher is likely a floor, as he kills relief pitchers this year.
As always, your best pick will fail to get a hit about 25% of the time. In fact, if you look at the spread sheet linked above, the NN is hitting that target well. So far, the average daily probability is .738, while the actual fraction of best picks getting a hit is .75. So the NN predicted 41.34 games with a hit, and delivered 42 out of 56.
Here is the daily list of active streaks of plate appearances without a hit:
Good luck!
from baseballmusings.com http://ift.tt/2qBzI8w
No comments:
Post a Comment