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:

0.288 — Dee Gordon batting against Jered Weaver
0.285 — J.T. Realmuto batting against Jered Weaver
0.277 — Nolan Arenado batting against Matt Moore
0.275 — DJ LeMahieu batting against Matt Moore
0.273 — Brandon Phillips batting against Jerad Eickhoff
0.273 — Cesar Hernandez batting against Jaime Garcia
0.271 — Christian Yelich batting against Jered Weaver
0.271 — Marcell Ozuna batting against Jered Weaver
0.271 — A.J. Pollock batting against Kenta Maeda
0.271 — Freddie Freeman batting against Jerad Eickhoff
0.271 — Martin Prado batting against Jered Weaver
0.271 — Jonathan Schoop batting against Steven Wright

With Jered Weaver on the mound, it’s a good day to be playing for the Marlins. Weaver’s problem this season has not been the quantity or frequency of hits allowed, but the quality of hits allowed. Opponents are hitting .231 against, him, but with a .508 slugging percentage. His three year profile is pretty poor, however.

The NN with Park produces this list:

0.288, 0.715 — Dee Gordon batting against Jered Weaver.
0.263, 0.708 — Daniel Murphy batting against Jacob deGrom.
0.275, 0.707 — DJ LeMahieu batting against Matt Moore.
0.285, 0.707 — J.T. Realmuto batting against Jered Weaver.
0.273, 0.700 — Brandon Phillips batting against Jerad Eickhoff.
0.271, 0.699 — A.J. Pollock batting against Kenta Maeda.
0.256, 0.699 — Jose Altuve batting against Blake Snell.
0.257, 0.696 — Corey Seager batting against Robbie Ray.
0.277, 0.696 — Nolan Arenado batting against Matt Moore.
0.261, 0.695 — Mookie Betts batting against Jayson Aquino.

It appears to be a Dee Gordon day. Gordon tends to produce a high batting average with a relatively low OBP, exactly the kind of player who is likely to get a hit. A big component of his three-year hit average, however, was his steroid induced 2015 season. The NN doesn’t know or care about that.

As always, even your best picks will fail to get a around 25 to 30 percent of the time.

I would like to readers opinion on the Beat the Streak strategy of doubling down. Beat the Streak allows you to pick two hitters on a given day. If both get a hit, your streak goes up by two. If only one of them gets fails to get a hit (or a push), your streak ends.

The probabilities among the best players are not that different, so the probability starting from scratch and picking 57 days in a row is the same as starting from scratch and picking two players a day for 28 days, and then the last on day 29. Both are approximately .72 to the 57th power. In that sense, it doesn’t matter which strategy you choose.

On the other hand, you could look at the game as a survival exercise. Your job today is to survive until tomorrow. Once you’ve survived 57 chances, you win. With a .72 probability of surviving on a single player, doubling down reduces that to about .52. However, you only need to survive for half the days.

I could be convinced that doubling down early in the streak is the right strategy, becoming more conservative as time goes on. I see this as similar to someone in the twenties starting to save for retirement. You have a 40 year investment period in front of you, so you go for higher risk, as that is more likely to do well over a long time period. As you approach retirement, you move the money into something less likely to crash and wipe out your savings.

Can some probability expert comment on this? My thought is that there is so little chance of either method working, it doesn’t really matter.