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. The sheet also includes a table that summarizes the length of positive and negative streaks.
First, the Log5 Method picks:
0.338 — Jean Segura batting against Daniel Gossett
0.329 — Scooter Gennett batting against Patrick Corbin
0.323 — David Peralta batting against Homer Bailey
0.323 — Ender Inciarte batting against Joe Ross
0.317 — Daniel Murphy batting against Sean Newcomb
0.317 — Freddie Freeman batting against Joe Ross
0.316 — Trey Mancini batting against Kyle Gibson
0.314 — A.J. Pollock batting against Homer Bailey
0.313 — Zack Cozart batting against Patrick Corbin
0.311 — Matt Kemp batting against Joe Ross
Segura is starting to approach Daniel Murphy territory, with a very high batting average and even fewer walks. Segura walked just 14 times this season in 251 at bats. He also struck out just 40 times, so he puts the ball in play, and he puts the ball in play well. He’s working on a .405 BABIP right now. He is an excellent choice for advancing a hit streak. Gossett, pitching against Segura, has a low strike out rate and a low BABIP. Lots of balls are put in play against him, but usually turned into outs, despite Gossett allowing a high number of line drives. So Log5 appears to have made a good pick here.
Here is how the NN with Park ranks the players:
0.338, 0.768 — Jean Segura batting against Daniel Gossett.
0.317, 0.763 — Daniel Murphy batting against Sean Newcomb.
0.294, 0.745 — Jose Altuve batting against J.A. Happ.
0.323, 0.744 — David Peralta batting against Homer Bailey.
0.323, 0.740 — Ender Inciarte batting against Joe Ross.
0.329, 0.738 — Scooter Gennett batting against Patrick Corbin.
0.314, 0.737 — A.J. Pollock batting against Homer Bailey.
0.316, 0.730 — Trey Mancini batting against Kyle Gibson.
0.310, 0.727 — Brandon Phillips batting against Joe Ross.
0.268, 0.727 — DJ LeMahieu batting against Carlos Rodon.
0.317, 0.727 — Freddie Freeman batting against Joe Ross.
There’s good agreement between the two systems today, the biggest difference in the person of Jose Altuve. He is not in the top ten of the Log5 method, but ranks third by the NN. Both systems agree on Jean Segura as the top pick.
Altuve was the top pick yesterday, and I ended up with a push because I was driving and didn’t know Altuve didn’t start until the pick locked. That’s always dangerous, because a day off can turn into a pinch-hit plate appearance, and you’re stuck with the luck of the single PA. The NN model assumes the player starts, so Daniel Murphy became the top ranked player on Saturday.
If you look at the spreadsheet linked above, you’ll see the NN is on a roll. After missing three out of four from 6/11 to 6/14, it has picked 22 of 24 correctly. The current streak is nine, so if Segura starts and gets a hit, the NN will go into the All-Star break with it’s longest streak of the season.
As always, your best pick will fail to get a hit about 25% of the time.
Here is the daily list of active streaks of plate appearances without a hit, with pitchers eliminated:
Good luck!
from baseballmusings.com http://ift.tt/2uEYZ4P
No comments:
Post a Comment