MLB Aging Curves

types

Header aging curve courtesy of beyondtheboxscore.com

On November 22, 2013, GSABR members Preston Barclay, Kyle Franco, Camden Hu, Nik Oza, and Xavier Weisenreider competed in SABR President Vince Gennaro’s Diamond Dollars Case Competition against twelve other schools at NYU in New York City. The case involved the forecasting of the free agent contracts of catcher Brian McCann and pitcher Ubaldo Jimenez by forecasting the players’ future production and other factors inherent to the free agent market, particularly the attachment of a qualifying offer on the two players, which would result in the loss of a draft pick for the signing team. The article below is part 3 of a series on the team’s research over a five day span from Sunday, November 17 until the competition.

CLICK HERE FOR PART 1

CLICK HERE FOR PART 2

The biggest mystery when you sign a baseball player, especially one at the edge of his prime, is the actual value that he will add to your team versus how much you pay him as he ages. It is essential to have a quality aging curve and prediction model to judge a free agent’s value. Although there are plenty of very good WAR projections publicly available, the actual specifics of the WAR projection and aging curve methodology isn’t something that is by and large available to the public.

For the Diamond Dollars Case Competition, we decided to make our own WAR projections for both Brian McCann and Ubaldo Jimenez. From those projections, we were able to come up with contract estimates for them (of course, the contract estimates here aren’t adjusted for the lost value of the compensatory pick, which we detail more in part two).

In determining the WAR projection for McCann, we predicted his future WAR by finding an aging curve, not just for total WAR, but for the individual components that make up WAR: batting, base-running, defensive, double play, and positional. By looking at these individual components and their aging trends, we were able to come up with a more precise projection. In making these aging curves, we computed the average year to year change of every player that recorded at least 300 at-bats in both years for each age year. We then graphed them additively.

First, looking at Batting WAR, we found the aging curves for both all batters, and for catchers specifically. Catchers tend to have less batting regression as they get older, which is probably because they are, on average, less valuable as hitters and therefore have less regression to the mean, as we can see in the chart below. However, we made the determination that Brian McCann is not like “most catchers” and has consistently been a very good hitter, so we used the all-hitter curve as our aging curve for Batting WAR.

aging

Double Play WAR has very little significant variance and change, as a player gets older. Therefore, for McCann, we set his Runs Above Average (RAA) for a double play as a constant value equal to his career average thus far.

Base-Running WAR has a very small, linear decrease, so we found the regression equation for the line and used the coefficient (.2007) as a constant decrease for every year older.

aging2

Positional WAR was a little more tricky in that running a regression wouldn’t really accomplish much, since playing catcher versus any other position is a binary variable that varies greatly from catcher to catcher. Although McCann has shown little trend of increasing or decreasing his games started at catcher to this point, it isn’t much of a stretch to assume that the percentage of games he starts at catcher will decrease as he ages.

aging3

Failing to find a purely quantitative method to project the percentage of games that McCann would start at catcher, we instead decided to use a subjective adjustment of decreasing 5 percentage points per year, starting at a 90% approximation for 2014.

Combining all the components and going through the aging curves, we found the total runs above average and then converted this into runs above replacement and then into his total WAR. We then used this WAR value to project McCann’s production over the course of his contract and its dollar value based on our $/WAR projections from part one.

aging4

aging5

3 Year Contract – $56 Million, $18.67MM AAV

4 Year Contract – $66 Million, $16.5MM AAV

5 Year Contract – $75 Million, $15MM AAV

6 Year Contract – $82 Million, $13.67MM AAV

If these numbers seem low for McCann (which they obviously are given the actual 5 year, $85M contract he received from the Yankees), you would be right. First of all, it’s the Yankees, so paying above the market value isn’t a foreign concept. But an additional factor that has affected baseball is the rise of Pitch f/x, which is very useful in determining the runs a catcher can add through framing pitches. McCann is very good at this, and factoring the pitch framing runs above added, we can see that McCann’s WAR value increases substantially.

aging7

Of course, this assumes that all teams have started using pitch framing in their valuations of players, and framing is a much newer consideration and something less well-known currently.

The WAR aging curve for Ubaldo Jimenez was slightly similar in that the only component that is considered is pitching WAR (for time purposes, accounting for defensive WAR or the potential of batting WAR really wasn’t practical). Again, we used the past ten years of consecutive pitching seasons in order to determine the average change as the player ages. Here is the additive visualization of the part of the aging curve from 30 to 37 years old.

aging8

Similar to our Batting WAR projections for McCann, before we applied the aging curve, we had to find a projection for the current Pitching Runs Above Average by calculating the expected values of the components of Pitching WAR. To do this, we ran a regression for each component on the previous values of that component using past data from consecutive pitcher seasons to get projections for Jimenez’s 2014 season.

Through this regression, we were able to predict the following values for WAR components for Jimenez’s 2014 season.

  • FB% – 35.70%
  • HR/FB – 10.47%
  • K% – 21.98%
  • BB% – 9.80%
  • BABIP – 0.30089219
  • LOB% – 71.9%

Plugging these values into the RAA formula, we get a predicted RAA value of -.08 for Jimenez in 2014 (after also predicting his innings pitched). Of course, in addition to his value and innings itched, we needed to project both his future as a starter, in terms of how many innings he would pitch each season over the course of his contract. From year to year, we used the significant variables of previous year’s ERA, games started, and innings pitched as the predictive values for games started and innings pitched of each current year.

Note that the games started doesn’t mean necessarily that Ubaldo will only start 14 games in a season down the road (something that would be pretty weird for a team to do), just that the average value of games that he would be projected to be starting goes down. As Ubaldo ages, he projects more as a reliever than as a starter.

Screen Shot 2014-03-27 at 9.44.43 PM

1 Year Contract – $15.50 Million, $15.50MM AAV

2 Year Contract – $25.00 Million, $12.50MM AAV

3 Year Contract – $29.00 Million, $9.67MM AAV

We see very quickly that Jimenez’s WAR rapidly decreases, even projecting at producing negative WAR by the 4th year of the contract (and thus not worth projecting a 4-year contract). Our model looks at Jimenez much more skeptically than the general consensus, especially when you consider the 4-year, $50 million dollar contract that he actually received. Jimenez projects to many teams as a higher-variance player, which probably led to one team (the Orioles) falling in love with him and his upside and giving him a contract that hugely overvalues his actual projected worth.

Right now, our projections are not making definitive claims. There are many aspects that our projections leave out, such as pitch framing, which is still a developing tool used to varying degrees by different teams in player evaluation. Jimenez also has gone through a significant change in his delivery (as we will talk about later in this series), and could very well outperform our projections. However, developing an aging curve and WAR projection is a very effective tool in developing an idea for what a player will contribute as he ages and should be a process that is undertaken when evaluating any player seeking a new contract.

Data taken from baseball-reference.com

Xavier Weisenreder
Georgetown University Class of 2016

Follow Xavier on Twitter: @BeMoreChillNext
Follow GSABR on Twitter: @GtownSports
Like GSABR on Facebook

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s