Recent Updates
Recent Updates to Player Won-Lost Records
Last updated January 12, 2018
Player won-lost records have been updated most recently as of January 12, 2018. The most recent update is due to the discovery of a fairly significant error in the calculation of (context-neutral, teammate-adjusted) eWins and eLosses.
Briefly, eWins and eLosses were being calculated without controlling for ballpark effects. Rather than averaging across every play within a particular ballpark, as I intended to do (and as I described in my book, I was mistakenly averaging across all plays within a particular league-season.
- pWins and pLosses were - and are - being correctly calculated using ballpark-specific win probability matrices.
- eWins and eLosses were not ballpark-adjusted. This has been corrected in the data which is now available on my website.
The impact of this error was strongest on players who played in extreme ballparks (e.g., Coors Field). Failing to correctly adjust for ballpark effects led to an over-valuing of hitters in hitter-friendly parks (e.g., Coors Field, Fenway Park) and pitchers in pitcher-friendly parks (e.g., Dodger Stadium, the Houston Astrodome) and an under-valuing of hitters in pitcher-friendly parks and pitchers in hitter-friendly parks.
Some significant examples of this are discussed in more detail in a separate article.
- The extent to which Player won-lost records are shared between batters and baserunners and between pitchers and fielders on certain plays is calculated using eWins and eLosses (which control for context as well as the performance of a player's teammates).
Because of this, pWins and pLosses are also affected by this change. This effect is fairly modest and non-systematic (as opposed to the more significant, systematic impact of this change on eWins and eLosses).
- Positional averages and replacement levels are also calculated after controlling for context. As with shared credit, then, positional averages and replacement levels - and, by extension, pWOPA and pWORL - are affected by this correction.
The last update to Player won-lost records prior to this change was on December 13, 2017. That update reflected two changes.
- Retrosheet released new data on November 26, 2017. Since the publication of my book, complete play-by-play data have been released for 1941 - 1944 and 2016 - 2017 and partial play-by-play data have been released for the 1926, 1928, and 1929 seasons. Some additional data have also been revised and updated for some additional games, primarily for seasons before 1941.
- I corrected an error which was inadvertently treating some non-ground ball double plays as ground-ball double plays if they occurred in a "double play situation" (runner on first base, less than two outs). This affected the split between Components 7 (ground-ball double plays) and 8 (baserunner outs) including the split of Player won-lost records between batters and baserunners and between pitchers and fielders.
Details of these updates can be found here.
Retrosheet releases new and revised play-by-play data twice each year, usually in late June and late November. It is my intention to continue to update Player won-lost records to incorporate these new data as they become available. Because of the way in which Player won-lost records are calculated, new data may lead to updates to data which are not revised, due to possible changes in the calculation of shared credit and - for seasons for which additional data may be added - the empirical calculation of positional averages and replacement levels.
It is my sincere hope that I have not made any additional errors in the calculation of Player won-lost records so that no further adjustments of this type will be necessary. If I do find such errors, however, I will do my best to update them as quickly and completely as possible and will be sure to report any such modificiations on my website.
My apologies for having to make these corrections at this time. I hope that you will continue to enjoy these new and improved Player won-lost records and come back to my website often!
All articles are written so that they pull data directly from the most recent version of the Player won-lost database. Hence, any numbers cited within these articles should automatically incorporate the most recent update to Player won-lost records. In some cases, however, the accompanying text may have been written based on previous versions of Player won-lost records. I apologize if this results in non-sensical text in any cases.
Home
List of Articles