Compiling Game Logs

Having determined the basic formula for offensive player fantasy WAR, I now needed a way to determine weekly variation by teams. One of my league’s members has been compiling and publishing team weekly means and standard deviations for each scoring category, which was useful. However, since our league is changing teams, I didn’t have a way to reliably adjust that for our 16-team league in 2017.

I decided that in the interest of being able to see the effect that team count might have on weekly stat variation to design a system to simulate a league of any size. In order to do that, I first needed daily statistics for every viable fantasy player.

MLB has a JSON API that can return daily game logs for a given player. You can request it from this endpoint:

http://m.mlb.com/lookup/json/named.sport_hitting_game_log_composed.bam
?game_type=%27R%27&league_list_id=%27mlb%27&player_id=#{player.id}&season=#{year}"

Where &player_id=#{player_id} should be the player’s ID (for Mike Trout that would be &player_id=545361), #{year} would be the year. You can even request non-regular season games by changing the game_type= parameter, or request minor league stats using the league_list_id= parameter.

(The string interpolations here are Ruby-style, so you omit the #{} surrounding the variable.)

With a list of player IDs, you can use this endpoint to build your own database of daily hitting statistics.

In the next post, we’ll simulate a draft to create the correct number of teams for each hypothetical league size we might want to analyze.