As indicated in part two, I have created a set of pitcher rankings that are also derived from betting market information. Starting pitchers require several days of rest between games, so each major league baseball team is actually more like 4 or 5 distinct teams, and my rankings effectively treat them as such. Although that complicates the ranking methodology somewhat, it allows me to determine how the betting market evaluates the strength of each pitcher individually.
See the methodology page for a simple explanation of how my ranking system works. See part one of last week's post for a not-so-simple explanation (it involves algebra and exponents and such).
A CaveatI will open with the same caveat I called out in my part one post: These starting pitcher rankings will also reflect fielding strength and bullpen strength of the given team, and thus should not be considered a "pure" ranking based solely on the specific pitcher's attributes and performance.
Then again, this is not a unique drawback when it comes to ranking pitchers. The contribution of team fielding strength to a pitcher's performance is always going to be difficult to tease out (it's effectively a sub-discipline in the field of baseball analytics, first pioneered by the awesomely-named Voros McCracken).
Bullpen strength is much more straightforward to separate out for traditional stat-based approaches (fundamental analysis), but not really possible using my betting market approach (technical analysis). A team's bullpen is surely factored into the betting market moneylines and totals, but there's no way to tell what is attributable to the starting pitcher versus an ace reliever (or lack thereof).
To keep the table size manageable, I am only publishing rankings for the top 50 pitchers. In order to qualify for the rankings, the pitcher had to have recorded at least three starts over the past 28 days. These rankings are based on results and betting information known as of the morning of June 30, 2012.
The key metric here is RAR, which stands for "Runs Above Replacement". It's how many fewer runs a team is expected to score when facing that starting pitcher, compared to a league average starting pitcher. I realize that my use of "Above Replacement" here is a bit non-standard. Replacement in my rankings means "league average", whereas it often means an entry-level Triple A player in standard baseball analysis. I may revamp my terminology in the future to avoid confusion.
The last column, labelled Season, is a sparkline intended to show how each pitcher has progressed throughout the season in the eyes of the betting market (the specific metric being charted is rank). Think of it as a compact, bite-sized version of the Ticker feature I published during the NBA season (and which I hope to relaunch soon for MLB).
Scaling can be a problem with sparklines since, by design, they lack axes labels and markers. As a result, small movements can get misleadingly magnified if the sparkline is rescaled to the max and min of the particular data series. To address this, I have formatted the sparklines such that they are all on an identical scale, with the bottom being the 50th ranked pitcher and the top being the 1st ranked pitcher. If you see a "break" in the sparkline, that's when the pitcher dropped out of the top 50.
The ranking table is below. Here are some observations:
- Having not followed baseball that closely this year, I'm somewhat at a loss to determine how out of line these rankings are with the general consensus. Anything look glaringly off?
- I'm still learning the ins and outs of Fangraphs, but does anybody know if there is a particular stat on their site that would be a good benchmark for comparison purposes? I'm looking for a predictive stat, not necessarily an explanatory one. Should I use WAR? FIP?
- The Season sparklines are still a work in progress, but I'm encouraged by their ability to tell a story in a small amount of space. Take Cliff Lee, currently ranked 15th. You can see he started out with high expectations. The break in the sparkline corresponds to his absence due to injury, and the subsequent decline reflects his struggles since returning from that injury.
The plan is to get these up and running on the blog on a daily basis, once I work out all the kinks. I may also try my hand at season and playoff simulations, similar to the playoff version of The Ticker I created for the NBA.