We are monitoring computer simulation picks from fivethirtyeight.com. In their own words on how they pick the games…
This forecast is based on 100,000 simulations of the season and accounts for starting pitchers, travel distance and rest. It updates after each game.
What they do is they provide probability of winning for each team each day according to their computer simulations. In our opinion, there’s two ways to interpret these picks:
- Compare their probability to the actual odds of the game and find the perceived value.
- Take the pick “as is” and select the team with the higher probability as the selection.
MLB Computer Simulation Predictions Record
Value Pick Record (since 5/27/17)
“As Is” Pick Record (since 5/22/17)
|-250 or Greater||33||10||5.60||77%|
|-200 to -250||83||35||7.60||70%|
|-150 to -200||217||146||-26.86||60%|
|+100 to -150||400||357||-41.25||53%|
MLB Computer Simulation Predictions
Derived Value Picks
Here we take the computer simulation probability and compared to the actual odds, the further away from the actual odds, the more value to the pick (i.e. if computer simulation suggests 60% chance of winning and the actual odds suggests 55% change of winning, then there’s a perceived 5% value in taking the favorite in this scenario). We give higher units to higher value plays, and the units are capped at 5 units.
|538 Value Pick||Odds||Risk Units||To Win Units||Value|
|Chi. White Sox||230||0.87||2||2%|
“As Is” Picks
These picks are more straight forward, basically which ever team has the higher probability of winning according to the computer simulation would be the selection. We make all picks as 1 unit plays.
|538 As Is Pick||Odds|