The 1922 Season as seen through the Prism of Player Won-Lost Records
(revised February 21, 2015)
In its Fall, 2013, update just after Thanksgiving of that year, Retrosheet released play-by-play data for three new seasons: 2013, 1939, and 1922. This article looks at the earliest of these three seasons: the 1922 season.
The 1922 season featured an All-New York World Series for the second year in a row, the most wins by any team in St. Louis Browns history, and three .400 hitters, the first 30-30 player in major-league history, and a Triple Crown winner.
The Best of 1922
I calculate Player won-lost records two ways: pWins, which tie to team wins and eWins, which control for context and the ability of one's teammates. For players with more pWins than eWins, their Player wins contributed to more team wins than one might expect; for players with more eWins than pWins, just the opposite is true: their Player wins translated into fewer team wins than expected. Or more briefly: a player with more pWins than eWins was better in context, a player with more eWins than pWins was worse in context.
The top 10 players in pWins above Positional Average and Replacement Level were as follows.
The top 10 players in eWins above Positional Average and Replacement Level were as follows.
So far, Retrosheet has released play-by-play data for just over 70% of all major-league games played in 1922. This is a remarkable total - we have play-by-play data for 870 games from the 1922 season, more than 90 years ago. Unfortunately, the missing games are not distributed evenly. We have complete play-by-play data for the Pittsburgh Pirates and have play-by-play data for more than 83% of all games for six other teams: Brooklyn, both Chicago teams, Cleveland, and both New York teams. The Boston Red Sox, Detroit Tigers and Washington Senators, on the other hand, are missing data for over half of their games (82, 88, and 87 missing games, respectively).
To attempt to control for this, the next two tables mirror the previous two tables except that Player won-lost records have been scaled up for players based on their actual number of games played. This scaling is done at the level of the individual player.
For example, Rogers Hornsby, who won the National League Triple Crown in 1922, played in 154 games for the St. Louis Cardinals that year. So far, Player won-lost records have been calculated for 122 of these games. Hence, Rogers Hornsby's Player won-lost records are multiplied by 1.26 (154/122).
The top 10 players in normalized pWOPA, pWORL, eWOPA, and eWORL in 1922 are shown in the next four tables.
The top 10 players in normalized eWins above Positional Average and Replacement Level were as follows.
Players' performance in missing games may not be exactly the same as their performance in games for which play-by-play data exist. So, the results in these last two tables should be taken with a grain of salt.
For example, before I started writing this article, I had never heard of George Mogridge. Mogridge did have a very nice career highlighted by a solid four-year prime from 1921 through 1924 over which he earned 65 traditional pitcher wins for the Washington Senators. But the above table almost certainly over-states precisely how valuable George Mogridge was to the 1922 Washington Senators.
In 1922, George Mogridge pitched in 34 games, starting 32 of them, completing 18 of those, pitching 251.2 innings with an ERA somewhat better than league-average of 3.76 and a traditional pitcher W-L record of 18-13. That's a good, solid season, but almost certainly not the second-best season in all of major-league baseball.
I have play-by-play data for 17 games in which Mogridge appeared. In these 17 games, Mogridge started 11 games, completing 9 of them, pitching 104.2 innings with an ERA of 2.49 and a traditional W-L record of 9-4. In the 17 games in which Mogridge pitched for which I do not have play-by-play data, however, Mogridge pitched 147 innings with an ERA of 4.35 and a traditional W-L record of 9-9. Still, I think it's fun to have players I've never heard of show up on these tables. So, hooray for George Mogridge!
As I noted above, three qualifying players batted over 0.400 in 1922, all of whom are now in the Hall of Fame: George Sisler Rogers Hornsby, and Ty Cobb. The next table compares the Player won-lost records for these three players for the games for which I have play-by-play data so far.
The next table blows up the Player won-lost records based on the actual number of games played by these three players.
We are missing a significant number of games for all three players, so it's hard to draw definitive conclusions. It is not too surprising, of course, that Rogers Hornsby's season was the best of the three. Cobb's record here seems a little weak for a 0.400 hitter. Of course, we only have data for about 40% of Cobb's 1922 season, so he might have been considerably better in the 74 games for which we don't have play-by-play data yet.
It's hard to be absolutely certain, because of missing games, but the best pitcher in the major-leagues in 1922 may have been Urban Shocker, who amassed a traditional pitching W-L record of 24-17 with a 2.97 ERA in 348 innings pitched.
In the first version of this article, the adjusted numbers (which were based on team games rather than player games, as they are now) showed Shocker with more than 20 pWins as well as 20 pLosses. As I noted at the time, the latter of these was "probably overstated", and, in fact, with more data and a more accurate player-specific adjustment method, Shocker now has considerably fewer than 20 estimated pLosses.
As I have noted elsewhere, pWins and pLosses are on the same scale as traditional pitcher wins and losses. This can be seen, for example, by comparing Urban Shocker's adjusted pWins and pLosses (21.9 - 17.6) to his traditional W-L record of 24-17.
Just as it is very unusual to see a pitcher who both wins and loses 20 games in the same season, so is it also unusual for a pitcher to amass 20 pWins and 20 pLosses in the same season.
The next table shows every season for which I have calculated Player won-lost records for which a pitcher amassed both 20 pWins and 20 pLosses. Keep in mind, of course, that I have complete play-by-play data only since 1937 and have not estimated Player won-lost records for any season before 1922.
The 1922 World Series saw the New York Giants sweep the New York Yankees, although the Series included two one-run games and a 10-inning tie.
The top performances of the 1922 World Series, as measured by Player won-lost records, are presented in the next table.
The top players of the 1922 World Series, as measured by Player won-lost records, were Giants' 3B Heinie Groh, who batted .474/.524/.579 for the Series; Hall-of-Fame 2B and 4-time World Series winner Frankie Frisch, who batted .471/.474/.529 for the Series; and Giants' pitcher Jack Scott, who pitched a complete-game 4-hit, 1-walk shutout in Game 3.
Best of 1922 by Factor and Position
Next, let's look at the top players in (context-neutral, teammate-adjusted) eWins over Positional Average in various aspects of the game. The numbers in this section have all been normalized to extrapolate player games for which I am missing play-by-play data. As above, this adjustment is based on individual player games for which I have play-by-play data.
Best by Factor: Batting, Baserunning, Pitching, Fielding
There are four basic factors for which players earn Player won-lost records: Batting, Baserunning, Pitching, and Fielding. The top players in 1922 in eWOPA by factor were as follows.
Positional Average excludes pitcher offense
It's a little surprising to not see Babe Ruth as the top hitter in the major leagues for a season in the 1920s. But Ruth only played 110 games in 1922. And it's not like Rogers Hornsby backed into his slot atop the leaderboard here. All he did was win a Triple Crown, batting .401/.459/.722 with 46 doubles, 42 home runs, 141 runs scored, 152 RBI, 250 hits, and 450 total bases - he led the National League in every one of those numbers.
Positional Average excludes pitcher offense
Pie Traynor is a guy who I'm looking forward to taking a closer look at as we get play-by-play data for more of his career.
In the 1950s and 1960s, he was widely regarded as the greatest third baseman in major-league history. This was almost certainly an overbid (certainly by the time Eddie Mathews hit his prime). But now, Traynor is so disregarded in sabermetric circles that, for example, he is not in the Hall of Merit.
So far, I have calculated Player won-lost records for five seasons of Pie Traynor's career: 1922, 1925, 1927, 1931, and 1934. He rates as the best defensive third baseman in the majors here in 1922 (at age 22), as the best overall third baseman in the majors in 1927 and 1931, and the best third baseman in the National League in 1925. It's a small sample size, of course, but I suspect Player won-lost records may be much kinder to Pie Traynor than other sabermetric measures.
Anyway, back to the the best fielders of 1922.
Best by Position
Next, we look at 1922 Major-League leaders in eWOPA by position. The figures shown here only include Player decisions earned while playing this particular position, and include no contextual adjustments (expected or actual).
For relief pitchers, context-neutral records may not be the best measure of how good they are, as context can matter a great deal, depending on how a pitcher is used. Here are the top relief pitchers of 1922 in context, in terms of pWins and pWOPA.
It's a little interesting, and indicative of the relative (un)importance of relief pitching in the 1920s, that context doesn't really improve the records of any of Barnes, Decatur, or Murray, who nevertheless still show up among the top 5 relief pitchers of 1922.
Finally, here are the best at three oft-forgotten positions that can nevertheless matter: pitcher offense, pinch hitting, and pinch running.
Major-League Baseball in 1922
Conventional wisdom has the last year of the deadball era being 1919 with baseball since 1920 looking more like the modern version. And, indeed, in 1922, four players hit 35 or more home runs, including Rogers Hornsby's major-league leading 42. Nobody in 1922 matched Chris Davis's 53 home runs (although Babe Ruth bested it in both 1920 and 1921), but only five players hit more than 35 home runs in 2013, only one more than in 1922 (with 14 more teams playing 8 more games).
On the other hand, though, neither the Boston Braves nor the Cleveland Indians hit 35 home runs as a team in 1922. The Philadelphia Phillies led the major leagues in home runs in 1922 with 116. Only three teams hit fewer than 116 home runs in 2013.
In 2013, there were a total of 4,661 home runs and 772 triples in major-league baseball. In 1922, there were only 1,055 home runs hit but 1,248 triples.
Aspects of the Deadball Era definitely still lingered in 1922. In 2013, major-league baseball teams collected 1,383 sacrifice bunts in 4,862 team games. In 1922, in 2,476 team games, there were 2,958 sacrifice bunts. In 2013, there were 2,693 stolen bases and 1,007 caught stealings. In 1922, there were 1,451 stolen bases and 1,165 caught stealings.
One statistic that was striking to me. In 2013, the major-league-wide strikeout-to-walk ratio was 2.51 (36,710-to-14,640) with 7.6 K/9. In 1922, the strikeout-to-walk ratio was 0.96 with a K/9 rate of 2.8. The major-league walk rate, however, was extremely similar in the two seasons: 3.0/9IP in 2013, 2.9/9IP in 1922.
Another number that was surprisingly (to me) similar was defensive efficiency: 0.692 in 2013, 0.682 in 1922, despite more errors in 1922 (3,215 to 2,745) in almost half as many games. And one more defensive shocker (to me). In 1922, there were more errors than double plays (3,215 to 2,242) while in 2013, there were more DPs than errors (4,340 to 2,747); no surprise there, but the difference is entirely because of errors: the number of double plays per game was actually slightly lower in 2013 (0.89) than in 1922 (0.90).
So, it was the same game, but it was different.
The next three tables compare the net win values of various offensive events in 1922 versus 2013. The 1922 win values here are calculated based only on the play-by-play data which has been released by Retrosheet.
Very different frequencies for a lot of events, but fairly similar values, although positive offensive events were somewhat more valuable in general in 2013, just because of the lower run-scoring environment.
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.
List of Articles