Sunday, October 12, 2014

Bill James Stability Index Graphed

Bill James has a fascinating article on his website (subscription required) tracking the stability of five different areas of baseball history (see key in the above graph) plus the overall stability measured as the average of the five factors. The data was calling out to me to be turned into a line graph, so here it is. I've used a rolling four year average trend line for the numbers James provided. (He didn't include the numbers for every year, just pointed out certain years in a narrative summary.)

Wednesday, October 8, 2014

New Top 100 & Thoughts

Last February, March, and April, I painstakingly rated my top 100 players of all time and made note of any connections each one had with KC. Click here to see all my top 100 related posts. With another season in the books and some new thoughts about how to adjust my rankings, I've rejiggered them a bit, and you can find my new list below.

In my original formula that greatly informed my rankings, I did not include a "timeline adjustment" to give more recent players a bonus. I considered it, thanks to Bill James arguing for and using one in his Historical Baseball Abstract top 100. Last off-season, my thinking was that all that really matters is how much a player dominates in his own era. If Babe Ruth was worth 10 WAR one season and Barry Bonds worth 10 WAR in another, then they were equally valuable. But after finalizing last off-season's ratings, I did a check to see how well my list was distributed throughout baseball history. Using a reverse numbering system where my #1 player was rated as worth 100 points in each year of his career, #2 worth 99 points in each year of his career, etc., I came up with a year-by-year index of how my top 100 were distributed. It looked like this:

The replacement and average level player keeps getting better and better, making it harder and harder to dominate the later in baseball history you get. I did not doubt that before, but now I understand how important it is to take into consideration when comparing one era to another. While thinking about this recently, I came across a fantastic 1977 study by Richard Cramer that actually quantified this ever-increasing skill level in baseball. (I would love to see someone smarter than me update that study.)

So I've added this timeline adjustment, stolen from James, to my formula: (year of birth - 1852)/6. On the extremes of players that made my list, Albert Pujols has an 18 point edge on Cy Young, worth roughly 9 WAR in the formula.

Here's how the year-by-year greatness breaks down in my new list:

It's especially satisfying how much this evens things out between 1953-2007. There's still that conspicuous bump centered around 1927, which I'm still not sure what to make of. It could be that there was just a random huge influx of greatness at that time. It's also possible that I'm overrating the 17 Negro leaguers included in my list. But maybe the most likely thing is that another adjustment is needed to penalize players from the segregated era. Keeping out a large population of the best players surely drove down the overall replacement/average skill levels, making it easier for very good players to dominate at all-time great levels. I'm going to keep thinking about that, but may add in a segregated era penalty in next year's revision.

Only four players fell off of my original list, victims mostly of the timeline adjustment: Joe Jackson, Ernie Banks, Home Run Baker, and Yogi Berra, replaced by newcomers Adrian Beltre, Chase Utley, Alan Trammell, and Carlos Beltran. Here is the new list, with a note for whether the player moved up, down, stayed in the same spot, or is new relative to my first rankings (the new top five by position is at the end):
  1. Babe Ruth -
  2. Willie Mays -
  3. Ted Williams -
  4. Oscar Charleston -
  5. Walter Johnson -
  6. Barry Bonds -
  7. Ty Cobb -
  8. Rogers Hornsby -
  9. Hank Aaron -
  10. Roger Clemens 

Tuesday, October 7, 2014