Miscommunication slows the adoption of baseball's advanced stats

Joe Sargent / Getty

I think there were some salient points buried beneath Bob Ryan's latest column decrying WAR and advanced statistics. I think there is often a failure to communicate what is beyond the metrics produced by independent baseball statisticians. And I think nothing makes that more clear than how poorly Bob Ryan -- a sportswriter who has never had a problem engaging with statistics -- understands the concept of replacement level.

Ryan cites two examples to express his displeasure with the replacement level idea. The first is Bob Hazle, a 26-year-old outfielder who came up from Triple-A to hit .403/.477/.649 in 155 plate appearances for the 1957 Milwaukee Braves. The second is Willie McCovey, who came up from Triple-A in 1959 to hit .354/.429/.656 for the Giants, won the Rookie of the Year, and of course went on to make the Hall of Fame.

To show the problems with these two examples, let's work backwards.    In 1957, as Hazle was earning the nickname "Hurricane" in Milwaukee, McCovey was finishing off an incredible debut season at the Double-A level. At 19 years old, McCovey was nearly eight years younger than the average Texas League player, but he hit .281/.372/.463 with 11 home runs in 115 games. The next year, he moved up to Triple-A Phoenix and hit .319/.387/.507 with 14 home runs in 146 games despite being 8.3 years younger than the league average player. And before his callup in 1959, McCovey hit .379/.459/.759 with 29 home runs in just 95 games.

McCovey was not in the major leagues, but he was by no means a replacement player. But there is no way the Giants would have traded McCovey to the Cubs for outfielder Walt Moryn (.234/.316/.386, 0.6 bWAR in 117 games) or to the Dodgers for catcher Joe Pignatano (.237/.346/.302, 0.3 bWAR in 52 games), both slightly better than Baseball-Reference's replacement level player.

If McCovey's talents were freely available, we can be sure they would have gone for far more than the league minimum $6,000 contract. It's hard to know for sure, as the owners still held complete control of the baseball labor market due to the reserve clause, but the case of Ken Harrelson a decade later tells us it would have surely been more than the league minimum.

As I wrote in March, Harrelson managed to get himself unconditionally released by Kansas City Athletics owner Charlie Finley, effectively for insubordination, on August 22nd, 1967. Entering the season, Harrelson owned a .232/.313/.384 (97 OPS+) career line as a first baseman and right fielder. As a 25-year-old, he hit .203/.261/.316 in his first major league stint with Washington, but after moving to Kansas City, Harrelson hit a robust .305/.361/.471 in 61 games. After Harrelson was cut, a fierce bidding war ensued between the teams still competing for the pennants, and he eventually earned $75,000 from the Boston Red Sox for the rest of the season -- $25,000 more than Carl Yastrzemski's 1967 salary.

The defining aspect of the replacement level player is that his talent is freely available, or available for a negligible price. What that exactly means has changed over the years -- in the pre-free agency and pre-draft years, when teams like the Cardinals owned entire minor leagues, there were likely a great many players with major league talent occupying the minor leagues. But there has always been a need to fill out a 25th roster spot or an open position on a Triple-A club, and it is specifically these players -- not just any minor leaguers -- who make up replacement level.

And so back to Hazle, who really does qualify as a replacement player. Hazle only played in two other seasons, 1955 and 1958, in which he combined to go 27-for-127 (.213) with four extra-base hits and combined for -0.2 WAR. Hazle apparently had a poor spring training according to this AP article. The Tigers acquired him for a player to be named later or cash on May 24, 1958, and according to Hazle's Baseball-Reference page, the Tigers chose the cash. Hazle was out of the majors just a year after earning the "Hurricane" moniker in Milwaukee.

But still, how do we reconcile the idea of Hazle being a replacement level player with his 1957 performance? Baseball-Reference had him at 1.9 WAR for his 155 plate appearances with the Braves. But consider the group of players who accrued 155 plate appearances or fewer in 1957. While there will be a number of prospects making a major league debut, this list will be most heavily populated with bench players (I excluded pitchers), those most likely to belong in the "replacement" family. And in 5,880 at-bats, these players combined to hit .230/.303/.332 according to the Baseball-Refernece Play-Index -- a line that fits much more neatly into the idea of the replacement player.

Ryan asks of Hazle, "In 41 games he hit .403 and compiled an OPS of 1.126. As a replacement player. Where’s the calculation on him?" The calculation is in those other 5,500 at-bats taken by players like Hazle. Even including Hazle's blustery performance, these players as a group were awful -- the kind of players who were typically replaced on next year's rosters.

Ryan is not a Luddite. He employs plenty of newer statistics, and is generally considered a student of the games he covers. So even though it is odd to see him expressing such confidence in his incomplete idea of the replacement player, when somebody like Ryan is so far off, it implicates those of us who profess the benefits of these stats as well.

Obviously there is still much work that needs to be done in concisely communicating what these concepts are. Perhaps it was unwittingly, but Bob Ryan's latest column proves that point.

Miscommunication slows the adoption of baseball's advanced stats
  Got something to say? Leave a comment below!

Daily Newsletter

Get the latest trending sports news daily in your inbox