Occasionally, you see something or get involved In a discussion that changes your thinking. I started this site because it allows me to react to these things in real time. So, today we are taking a step back from the index briefly so we can address a debate that has been raging on the internet. Essentially, there has been a backlash against WAR. It has been going on for a lot longer than I’ve been making out, and I won’t be avoiding WAR for too long, but since I have the platform I’ll try to put my own spin on the subject.
Last time, we introduced total runs as an alternative number to use. It has its advantages. First, it’s easy to understand because the components are right there in front of you. With the components being there you can also manipulate them. We certainly will do that. Second, it accounts for longevity. That’s one of the hallmarks of the Hall of Fame. The longer you play the more likely you are to get in. Of course, there are some that debate the merits of longevity as it compares with quality. One compares players that compile numbers and others compare players with the average.
WAR and win shares attempts to marry the two by comparing players with a replacement level player. So, one can accrue value over time by being average or even below average. That becomes important when looking at a total runs metric. We will look at the findings for those in the Hall of Fame at second base along with four players that some people think definitely should be in the Hall of Fame. As promised, there will be no WAR today, but there will be two iterations of total runs. The first combines runs created, Rfield, and Rbaser. That’s a direct reenactment of the formula seen on billjamesonline.com
Depending on your familiarity with these numbers, some of you are either soaking it in or yelling at the computer screen. First, I should mention that I don’t think Bill ever intended for his total runs metric to be used this way. His are always shown through the eyes of a single season. There is an important reason for that and it has to be mentioned before we move forward. Runs created are not normalized throughout history. Simply put, 100 runs created in 1968 is far different comparatively to 100 runs created in 1998. So, comparing a player like say Ryne Sandberg to Jeff Kent or Rogers Hornsby is nearly impossible using runs created.
This becomes problematic when trying to reach any conclusions from this particular formula. Looking at runs created ignores the quality in which these runs were created. In other words, while a Craig Biggio was second all-time at the position in runs created, we cannot assume he was the second-best hitter at the position. He produced the second most runs partially because he spent most of his career in a great hitter’s era and also because he played for a really long time. So, some would question his placement in the fourth spot in history and rightfully so.
When one considers WAR they would consider that at a certain point, Biggio was no longer providing value. Yes, he was creating runs, but if the Astros put someone like Chris Burke in his spot they might have gotten a better rate of runs created, baserunning runs, and fielding runs. So, in a table like above, Biggio was continuing to add value. In reality, he wasn’t. That’s where statistics like win shares or WAR are more descriptive than numbers like above.
Of course, we can make adjustments above by replacing runs created with baseball-reference’s Rbat statistic. It is the number of runs created above average. This does a number of things for us. First, it brings more fidelity because Rbat, Rfield, and Rbaser are all compared to the average player. Secondly, it helps solve the problem of different eras because each player is compared with the average player from that era. Finally, it doesn’t reward longevity nearly as much. You only get extra credit if you are actually good. It is interesting to see how the rankings with the same players differs when switching methodologies.
One of the joys of living with a scientist is that I get to see this kind of work poked fun at because of all of the rules we break. Most baseball statisticians are guilty of something called confirmation bias. It is defined as the tendency to search for, interpret, favor, and recall information in a way that confirms one’s preexisting beliefs. James himself demonstrated this when he asserted that any metric that did not have Ruth on top was inherently inaccurate. That may or may not be true, but it obviously doesn’t rigorously follow the scientific method.
I say that to point out that the above list probably matches most people’s beliefs about second basemen closer than the first one. That would tend to get most to say that the above method is more accurate. The problem is that average performers have value. According to the above chart they don’t. So, there is an inherent problem here. The first chart asserts that everyone has value. The second says you have to be above average to have value. What we need is something that marries the two.
This is where metrics like WAR and win shares come in. Craig Biggio is not the worst player in the group and he isn’t the fourth best player in the group. He is likely somewhere in between. The same is true of a player like Bobby Grich in reverse. Some credit has to be given for longevity, but we have to be careful about how much. If we learn anything it is that one set of numbers (no matter how carefully compiled) can tell us the whole story. That includes WAR.