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:

0.319 — Daniel Murphy batting against Tyler Anderson
0.297 — Josh Harrison batting against Brett Anderson
0.293 — Elvis Andrus batting against Phil Hughes
0.292 — Kevin Pillar batting against Ricky Nolasco
0.286 — Corey Seager batting against Matt Cain
0.285 — Trea Turner batting against Tyler Anderson
0.285 — Adam Frazier batting against Brett Anderson
0.284 — David Freese batting against Brett Anderson
0.283 — Ryan Zimmerman batting against Tyler Anderson
0.283 — Nomar Mazara batting against Phil Hughes
0.283 — Bryce Harper batting against Tyler Anderson

The predicted Murphy against Tyler Anderson produces a huge .319 predicted hit average. Anderson’s career is short, but in that time he allowed a fairly high batting average with few walks, exactly the kind of pitcher that gives a batter the chance to collect a hit. With Murphy off the charts on the batting side, they are a match made in heaven. The Log5 method doesn’t even take into account that the game is played in Coors Field.

The NN, however, does take park into account and produces this list:

0.319, 0.750 — Daniel Murphy batting against Tyler Anderson.
0.286, 0.714 — Corey Seager batting against Matt Cain.
0.278, 0.710 — A.J. Pollock batting against Jhoulys Chacin.
0.297, 0.709 — Josh Harrison batting against Brett Anderson.
0.270, 0.709 — DJ LeMahieu batting against Joe Ross.
0.285, 0.707 — Trea Turner batting against Tyler Anderson.
0.293, 0.701 — Elvis Andrus batting against Phil Hughes.
0.292, 0.698 — Kevin Pillar batting against Ricky Nolasco.
0.264, 0.695 — Charles Blackmon batting against Joe Ross.
0.268, 0.694 — Nolan Arenado batting against Joe Ross.

The NN is kinder to the Andersons than the Log5 method, but Daniel Murphy still comes out way ahead of anyone else playing this evening in terms of probability of getting a hit in the game, .750. Notice that the NN produces a high probability in this situation, but no where near a certainty. A great hitter against a hittable pitcher in a park that helps offense, and Murphy still has a 25% chance of not getting a hit.

Two new items. On a reader request, I’ve added “%Games with Hit” to the Batting Log. It appears in averages summary line at the bottom.

A reader also asked for a list of players with the longest hitless streak in terms of plate appearances (they might have reached base by something other than a hit):

I suppose the players at the top of the list are due. Good luck!