Baseball Player Won-Lost Records
Games and Seasons for which Player Won-Lost Records are Calculated
Last updated July 29, 2020
At present, Player won-lost records are available for the following seasons. Seasons in italics are missing some games.
1916 - 1927
1928 - 2019
Most Recent Updates to Player Won-Lost records
Player won-lost records are calculated using play-by-play data. The source of this play-by-play data is Retrosheet. Retrosheet released its most recent set of play-by-play data on July 11, 2020.
The primary addition to Retrosheet's data with this release was the release of two new seasons, 1916 and 1917. In addition, as part of Retrosheet's latest release, a combination of play-by-play and deduced accounts have also been released to complete the 1928, 1929, 1930, and 1931 seasons. In addition to these seasons, Retrosheet added play-by-play data for 44 games from 1927. Overall, play-by-play data for 2,435 new games were released by Retrosheet in its latest update.
In addition to play-by-play data for games for which no data previously existed, Retrosheet has made fairly significant upgrades to its play-by-play data for a number of additional games. These include 73 games which have been promoted from "deduced games" to "event files" (see below for a discussion of this distinction).
All of these new and improved data have been incorporated into my Player won-lost records as of July 29, 2020.
For seasons with missing data, teams are missing the following numbers of games. Team records in missing games are shown in parentheses (W-L-T).
1927
American League
National League
1926
American League
National League
1925
American League
National League
1924
American League
National League
1923
American League
National League
1922
American League
National League
1921
American League
National League
1920
American League
National League
1919
American League
National League
1918
American League
National League
1917
American League
National League
1916
American League
National League
Deduced Games
For games for which Retrosheet is unable to find play-by-play data, they construct what they call "deduced" games. These are play-by-play accounts which are constructed based on box score data and newspaper stories. In many cases, the resulting games are "plausible" but not necessarily unique. That is, there may be some uncertainty, for example, as to whether a particular batter singled in the first or third inning of a game.
Retrosheet distinguishes between what it calls "event files" - which consist exclusively (or, at least, almost exclusively) of known play-by-play events - and "deduced accounts" - which may contain some uncertainties. Retrosheet also only releases deduced games for full seasons. Working backwards in time, Retrosheet has released deduced games which cover every game for which event files are not available from 1928 through the present. I have helped to deduce a number of these games, starting with the 1949 season, and am continuing this deduction work at present for the 1926 season.
The total number of deduced games per season which are currently used in constructing Player won-lost records are shown in the next table. As of its latest release, Retrosheet has complete play-by-play data for all seasons since 1973 as well as 1960, 1961, 1969, and 1970.
For my purposes, I have calculated Player won-lost records from both sources: event files and deduced files. I have chosen to do so largely under the basic premise that some data is better than no data.
For (context-neutral, teammate-adjusted) eWins, the timing of events doesn't really matter, only that events happened - i.e., all that matters is that a player went 2-for-4 with a single, double, and strikeout in his 4 at bats, not when the hits happened or how many runners were on base when he struck out. Strictly speaking, this knowledge does not require play-by-play data, merely box-score level data, which are available and known with relative certainty for all deduced games (and, in fact, for all games dating back to 1901).
For pWins and pLosses, which are context-dependent, the exact timing of these events will matter. Even for deduced games, however, the timing of most events actually are known, and this is more true the more important the event (e.g., events associated with run-scoring innings are typically reasonably well covered in newspaper stories). Because pWins and pLosses are a perfect accounting structure with a fixed number of pWins and pLosses assigned within any single game (2 pWins and 1 pLoss to the winning team, 1 pWin and 2 pLosses to the losing team), any errors in credit based on the uncertainty of events will be limited to a single game and a single team: that is, if I have inadvertently credited too many pWins to a specific player in a specific game, I must have similarly credited too few pWins to this player's teammate(s). This could certainly still lead to errors in the calculation of Player won-lost records, but, I believe, it minimizes the likely significance of such errors.
Nevertheless, Player won-lost records based on deduced game accounts should, perhaps, be viewed with more caution than Player won-lost records based on event files. More generally, the quality of Player won-lost calculations is a direct function of the quality of the underlying play-by-play data and that quality generally declines the farther back in time one goes. This is perhaps most significant with respect to fielding records. It is common, for example, for the exact nature of batting outs to be uncertain (i.e., we know that a batter made an out, but don't know if he grounded out to the second baseman or flew out to left field, etc.) for most deduced games, but also for many event files as one moves farther back in time. In general, these uncertainties are probably somewhat greater within deduced games than within event files, but the reality is that these uncertainties move along a continuum from highly uncertain deduced games to very good deductions to somewhat weak event files to highly detailed event files.
Overall, while recognizing that there are uncertainties surrounding certain plays in certain games over time, I believe that the Player won-lost records presented on this website are calculated as accurately as possible given the data available to me as of today. I am also reasonably certain that the quality of these data will only continue to improve over time as Retrosheet continues its ongoing and impressive efforts to constantly add more and better data to its already impressive historical dataset.
A Comment About Deduced Games
I would also like to point out a couple of things about deduced games.
First, note that the number of deduced games is not uniformly increasing as Retrosheet moves further back in time. The number of deduced games in the earliest season for which Retrosheet has released deduced games, 1928 (107), is fewer games than in any season between 1940 and 1950, and the number of deduced games in 1929 (44) is fewer than in any other season prior to 1955.
Second, as of Retrosheet's previous update (in December 2019), Retrosheet was missing play-by-play data for 909 games in the four seasons from 1928 through 1931. Those numbers have been reduced to zero. But of the 909 games added for these two seasons, only 285 of these games ended up being classified as "deduced games". The other 624 games (68.6% of the total games added) are straight event files. In some cases, Retrosheet had scorecards which had not been entered just yet last year. In some cases, in working through deducing the missing games, play-by-play data was found in some newspapers. And in some cases, in working through newspaper stories to deduce a game, it was discovered that filling in everything explained in newspaper stories left no uncertainties: some newspaper stories are detailed enough to place every baserunner in a game, sometimes multiple newspaper stories complement each other in a way that leads to all baserunners being covered, sometimes a few baserunner uncertainties are left that turn out to only be able to fit together in one possible way. It is impossible to say with certainty exactly how subsequent seasons will go, but it is likely that additional historical seasons will end up with fewer than 100 deduced games (perhaps, in some cases, far fewer).
I hope you enjoy the Player won-lost records that I have available here. Thank you for visiting and please come back often!
Article last updated: July 29, 2020
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