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 updated the models, and the results of those tests are here.

For 2018, 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:

0.295 — Jose Abreu batting against Jordan Zimmermann
0.295 — Avisail Garcia batting against Jordan Zimmermann
0.279 — Tim Anderson batting against Jordan Zimmermann
0.276 — Starlin Castro batting against Nick Pivetta
0.275 — Eric Hosmer batting against Tyler Anderson
0.272 — Dee Gordon batting against Kyle Gibson
0.271 — Welington Castillo batting against Jordan Zimmermann
0.268 — Charlie Blackmon batting against Joey Lucchesi
0.267 — Jean Segura batting against Kyle Gibson
0.267 — Elvis Andrus batting against Daniel Mengden

Since 2016, the Zimmermann allowed a .301 BA with few walks, so it’s not surprising to see the better hitters on the White Sox showing up on the list. Abreu is off to an 8 for 20 start this season.

Here is how the NN with Park ranks the players:

0.295, 0.713 — Avisail Garcia batting against Jordan Zimmermann.
0.295, 0.711 — Jose Abreu batting against Jordan Zimmermann.
0.272, 0.706 — Dee Gordon batting against Kyle Gibson.
0.267, 0.699 — Jean Segura batting against Kyle Gibson.
0.268, 0.697 — Charlie Blackmon batting against Joey Lucchesi.
0.276, 0.691 — Starlin Castro batting against Nick Pivetta.
0.264, 0.688 — Robinson Cano batting against Kyle Gibson.
0.267, 0.686 — Elvis Andrus batting against Daniel Mengden.
0.275, 0.686 — Eric Hosmer batting against Tyler Anderson.
0.264, 0.683 — Scooter Gennett batting against Steven Brault.
0.279, 0.683 — Tim Anderson batting against Jordan Zimmermann.

The NN rates Garcia higher than Abreu. We can get an idea of why by looking at the parameters.

Avisail Garcia Parameters: [‘0.217’, ‘0.281’, ‘0.223’, ‘0.282’, ‘0.213’, ‘0.223’]
Jose Abreu Parameters: [‘0.217’, ‘0.281’, ‘0.230’, ‘0.277’, ‘0.213’, ‘0.223’]

Since they are teammates, four of the parameters will be the same. The first two are the hit average for the pitcher this year and over the last three seasons, including this year. The last two are the league hit average for this season, and the park hit average over the last three seasons. The middle two parameters are hit average for the batter this season (regressed to the league hit average over 200 plate appearances) and the three-year hit average for the batter (regressed to the league hit average over 600 PA). In the Log5 method, the one year and three year are simply averaged together. The NN assigns weights, and it assigns the most weight to the three-year batter hit average. So in this case, Garcia wins. Since they are tied in the Log5 method, Garcia becomes the consensus pick.

By the way, Anthony Rendon owns about the weakest six-game hit streak you may ever see.

Remember, your best pick will fail about 25% of the time. Good luck!

from baseballmusings.com https://ift.tt/2GCrZE3