AdamsMLBPicks.com
"Bringing Machine Learning to Baseball"

Top Picks

I will post my top picks every day for you to compare our picks against the actual outcomes. Usually my picks will be for the home team to win because these are easier to pick consistently (see: the home team advantage), however in the rare case of a significantly disperate matchup, you may see an away team pick.


Current Accuracy of Top Picks:     63.22% (452/715)


Today (2010-09-07):

Day Visitor Home Home Team to Win?
Tue KCR MIN YES
Tue BAL NYY YES
Tue SEA OAK YES
Tue FLA PHI YES



Previous 10 Picks:

Click on the Top of a Column to Sort the Table--IE Only
Date Day Visitor Home Home Team Won?
20100906 Mon KCR MIN TRUE
20100906 Mon FLA PHI FALSE
20100906 Mon SEA OAK TRUE
20100906 Mon BAL NYY FALSE
20100905 Sun MIL PHI FALSE
20100905 Sun TEX MIN TRUE
20100905 Sun TOR NYY FALSE
20100904 Sat MIL PHI TRUE
20100904 Sat LAA OAK TRUE
20100904 Sat CHW BOS FALSE



To download all past predictions click here.



Details:

When analyzing the potential for a team to win in baseball, several noticable factors are immediately discernable. 
The Home Team Advantage: 
    On average, the Home Team will win 54% of the time in any given season. Scholars debate the exact nature of this advantage, however it holds constant across teams, ballparks, and seasons. The net mean score for baseball games played between 2000 and 2009 is around +0.16.

Difficulty of Score Prediction: One of the main problems with forecasting outcomes in baseball games is analyzing the net score. Many of the variables which one can obtain before the game occurs, while good indicators of past performance, serve as poor indicators of future performance. A large part of this stems from the spread of net baseball scores being very wide. Around 50% of all games have a net score between -3 and +3, however the remaining 50% on the tails of the distribution can reach as high as +/- 20 points. Because of this, it is easier to predict whether the net score will be greater than zero.


More about Me



How It Works:

  • While the details of our methods must remain prevent people from stealing our advancements, below are some details to give you a general idea of the improvements we've made upon existing methods, and why our models is so successful.
  • Our methods involve a combination of data mining, machine learning, and association rule learning techniques.
  • Our secret formula uses a series of multiple moving averages taking into account different periods of time to better discern trends in team playing, these results are then combined with stochastic measures of team performance in order to account for streaks and hot periods in playing.

Contact Us


Phone: 951-833-8033

E-Mail:
Adam@adamscorner.com