1 of 36

Progress Report

​

MLB: Automatic Strike Zone

Devon Goetz, Zack Kopstein, Simran Pabla, Claire Wichman

​

​

April 7, 2020

​

2 of 36

3 of 36

4 of 36

Project Specifications

5 of 36

Objective

Quantify the impact of an automatic strike zone.

6 of 36

Objective

Quantify the impact of an automatic strike zone.

  • Comparing “corrected” innings
  • Confidence intervals
  • Significance testing

7 of 36

Objective

Quantify the impact of an automatic strike zone.

  • Walks
  • Strikeouts
  • Runs scored
  • Pitches thrown

8 of 36

Objective

Quantify the impact of an automatic strike zone.

  • Maintain certain human elements
  • Game manipulation

9 of 36

Approach and Methodology

  • Classify mistakes
  • Compare existing innings
  • Create predictive algorithm
  • Analyze differences

10 of 36

Approach and Methodology

  • Classify mistakes
  • Compare existing innings
  • Create predictive algorithm
  • Analyze differences

Completed Work

Future Work

11 of 36

Approach and Methodology

  • Classify mistakes
  • Compare existing innings
  • Create predictive algorithm
  • Analyze differences

12 of 36

Approach and Methodology:

Classify Mistakes

  • Four categories:
    • Correctly called balls
    • Correctly called strikes
    • Balls that were called strikes (defensive)
    • Strikes that were called balls (offensive)
  • Inning based counting

13 of 36

Approach and Methodology:

Classify Mistakes

  • Four categories:
    • Correctly called balls
    • Correctly called strikes
    • Balls that were called strikes (defensive)
    • Strikes that were called balls (offensive)
  • Inning based counting

Called Ball

14 of 36

Approach and Methodology:

Classify Mistakes

  • Four categories:
    • Correctly called balls
    • Correctly called strikes
    • Balls that were called strikes (defensive)
    • Strikes that were called balls (offensive)
  • Inning based counting

Called Strike

Called Ball

15 of 36

Approach and Methodology:

Classify Mistakes

  • Four categories:
    • Correctly called balls
    • Correctly called strikes
    • Balls that were called strikes (defensive)
    • Strikes that were called balls (offensive)
  • Inning based counting

Called Strike

Called Strike

Called Ball

16 of 36

Approach and Methodology:

Classify Mistakes

  • Four categories:
    • Correctly called balls
    • Correctly called strikes
    • Balls that were called strikes (defensive)
    • Strikes that were called balls (offensive)
  • Inning based counting

Called Strike

Called Strike

Called Ball

Called Ball

17 of 36

Approach and Methodology

  • Classify mistakes
  • Compare existing innings
  • Create predictive algorithm
  • Analyze differences

18 of 36

Approach and Methodology:

Compare Existing Innings

  • Main metrics:
    • Number of walks and strikeouts
    • Number of runs scored
    • Number of pitches thrown
  • Influence of mistakes on metrics

19 of 36

Preliminary Results

20 of 36

21 of 36

22 of 36

23 of 36

24 of 36

25 of 36

26 of 36

27 of 36

28 of 36

29 of 36

30 of 36

Moving Forward

31 of 36

Approach and Methodology

  • Classify mistakes
  • Compare existing innings
  • Create predictive algorithm
  • Analyze differences

32 of 36

Approach and Methodology:

Create Predictive Algorithm

  • Calculate probabilities of outcomes
  • Change mistakes in the game
    • Identify the mistake
    • Correct the count and state
    • Predict the inning outcome based on the new state

33 of 36

Approach and Methodology

  • Classify mistakes
  • Compare existing innings
  • Create predictive algorithm
  • Analyze differences

34 of 36

Approach and Methodology

Analyze Differences

  • Quantify actual outcomes
  • Calculate final outcomes of “corrected” games
  • Determine the impact of an automatic strike zone

35 of 36

Next Steps

  • Finalize feature set
  • Train Model
  • Predict new outcomes
  • Quantify difference

04.02

Team Strategy Two Due

​

Finalize statistical analysis and visualization of Corrected Inning Comparison.

​

Begin feature set definition.

04.09

Progress Report Presentations

​

Build and train HMM/CI model.

​

Iterate on key visualizations.

04.16

Visualization Due

​

Finalize Predictive Algorithm.

​

Draft Executive Report and Engineering Summary..

04.23

Team Technical Two Due

​

Analyze changes in corrected games.

​

Finalize ER and ES.

04.30

Team Communication Two Due

​

Wrap up coding deliverables.

​

Finalize visualizations.

​

Prepare final presentation and handoff.

05.07

Final Presentation and Handoff

​

Ensure all code and presentations are finalized and ready for handoff.

36 of 36