Big Free Agent Pitching Contracts: A Bad Idea

About a week ago, I took a look at how much value teams realized on several large ($70M+) contracts given out free agent position players, as well as compared that to the value realized on similarly large contract extensions covering free agent years. The next logical step was to do the same thing for pitchers, although I made some changes to the way the list was put together to make the inclusion criteria more consistent and to avoid missing any contracts.

With the hitters, I established a cutoff of $70M of guaranteed money for inclusion. The problem is, this didn't account for significant inflation in baseball salaries and contracts. For example, a $50M contract in 2000 would easily be over $100M today, but one would be included and the other not. Also, pitchers generally receive smaller contracts. In order to account for this, I established a cutoff of $50M in 2010 real dollars. I estimated an annual salary inflation increase of 6.3% to come up with the following table of guaranteed contract dollars needed for inclusion by year:

The other problem was trying to get a complete dataset, rather than trying to build a list off from names of the top of my head and using Cot's list of the largest contracts (which are not adjusted for salary inflation). In order to capture all possible contracts, I used the Fangraphs leaderboard to generate a list of all pitchers who recorded a season with at least 2.0 fWAR from 1995 to 2011 in which they had qualifying IP. I suppose it could be possible a large contract could be awarded without reaching this criteria, but I don't think it's a large possibility and in any event would not change the conclusions.

This gave me a list of 345 pitchers, and for all those who had reached free agent eligibility I checked their contract details. I cross-referenced this against Cot's list of the largest contracts and annual salaries to pick up a few relievers who had large enough contracts (but missed the innings criteria). While I had some contracts from the early to mid-90s, I started at 1997 for two reasons. First, it's harder to find contract details, but more importantly, I'm already extropolating value measures from before 2002, and I'm uncomfortable extropolating much further back for fear that the market was quite different. This left me with 34 free agent contracts, and 22 contract extensions to analyze.

Methodology

A quick review of how I evaluated the contracts to determine the value, and some other important details:

Free Agent Pitcher Contract Performance

Unsurprisingly, given the conventional wisdom on long-term pitching contracts, this is truly a bloodbath from a team perspective, as on average they recoup roughly half of the salary that pay out in salary. Only 5 pitchers delivered positive value, with one contract breaking even, meaning that 85% of the contracts delivered negative value. The list is littered with pitchers who delivered almost no value above replacement value, in many cases due to injuries. In addition, the volitility is very high, with the % Gain/Loss sample having a standard deviation of 0.61. This compares with the position players in the other study with an average return of -22% and a standard deviation of about 0.24. Normally, if you accept extra risk, you want a higher expected return, but we observe the opposite here. Not only do you expect to have a large loss, but you really have no idea whether you'll get no value whatsoever, or receive positive value.

Again, we have an issue in this data set. The conventional wisdom says that the backend of a big contract is the worst part, but some of these back-ends are cutoff because the contracts are not complete, with may overestimate the return on the contract. If we project the contracts through to the end, we get the following:

Surprisingly, we actually see that the return gets slightly better. This is most likely just some statistical noise, but it's also possible that it's indicative of better contracts being given out recently. I'll return to that idea below.

Trends in the Data

First, to see whether there's any relationship betwee the size of contract and the value received, I've plotted the real value of the contract (in 2010 dollars) against the ratio of surplus to salary:

I've omitted Cliff Lee from this graph, as well as the others that follow, since it's only one year of data. Basically, the linear relationship is non-existant, and there's no discernible relationship. This is particularly evident among contracts between $50M and $100M, which is the majority of players.

Next, to return to the question of whether or not the contract are getting better over time, I've plotted the surplus to salary ratio against the year the contract was given out:

Basically, it's a fairly random plot. I've omitted the linear line of best fit, since it's basically a flat line with no explanatory power. Instead, I've added a second order polynomial regression, which also has a fairly weak explanatory power. That said, it appears that towards the end, there is an uptick in value received, with more contracts around the breakeven point. While there appeared to be a downtrend prior to 2003, I would discount this somewhat considered it's pulled up by contracts given to Clemens and Johnson, who delivered historically excpetionally performance. If we take those out, the downtrend completely disappears.

Finally, the age of the pitcher could figure into the value equation as well. Younger pitchers are at a better spot on the aging curve and have less mileage on their arms. I've plotted the surplus to salary ratio against the age of the pitcher at the beginning of the first year of the contract:

We actually see that older pitchers deliver better value. While this is somewhat counter-intuitive, I think there are good reasons for it. Older pitchers will get fewer years, and presumably will only get a larger contract if they have a clean injury history. If they have an injury history, their services will get discounted accordingly. Younger pitchers, on the other hand, are still on the wrong side of the pitcher aging curve, and even if they had a clean injury history are susceptible to injuries with as the wear and tear of years of pitching accumulate.

What About Contract Extensions?

With the position players, we saw that contract extensions of similar size delivered much better value. There were 22 pitchers who signed extensions of similar size to the free agent contracts over the same time period:

I've included Dice-K here, because I think his contract most reflects an extension situation - he could only negotiate with one team. I've included the posting fee, since we're looking from the team perspective (and because inflation is accounted for through the rising $/WAR, it doesn't matter how it's allocated year by year) Again, I've projected the performance of contracts which are not done:

In both cases, we see that players who receive contract extensions deliver extraordinarily superior returns relative to gree agent. Essentially, the teams come close to receiving full value at the marginal market rate, similar to what is observed with position player contract extensions. In the case of pitchers, the contrast is even starker.

Once again, I think a good portion of this can be explained by the Winner's Curse. However, in this case, I think there's also a selection bias at play. In most cases, these players will have been with the team that extends them for many years, and possibly will have come up through their system. The biggest risk with a pitcher is a catastrophic injury, and the team will have substantially more information about the player's historical usage, pitch mix, medical/injury history, etc. In other words, if they have doubts, they will not risk extending the player, and the player will hit free agency. If they are comfortable with the risk, they extend the player at market value. This could explain why younger free agent pitchers deliver less value relative to older players - they didn't have many, if any, free agent years bought out due to a perceived injury risk. They manage to stay healthy despite this, and then the injury risk catches up with them. In other words, it's the market for lemons, baseball edition; and caveat emptor to a team signing a free agent pitcher.

Conclusions

Free agent pitchers deliver very poor value to the team signing them, on average producing only 55-60% of their contract value. In addition to the large negative expected return, the volitility of the return is quite high, with a substantial risk of a player delivering no value compared to a replacement level pitcher. There is little, if any relationship between the value received and time, or the size of the contract.

Pitchers of similar quality who sign contract extensions covering free agent years deliver fair value for their teams, and drastically outperform their free agent peers. This may be due to information asymmetries, and should teams to be very wary of committing significant dollars to free agent pitchers. In other words, it's no wonder Edwin Jackson hasn't found the 5 year, $60M contract he's apparently looking for.

In terms of the Blue Jays, they've gone very pitching heavy in the draft over the past couple years, and I think the data validates this. We know premium pitching is needed to compete in this division. If you have premium homegrown pitching, you don't need to go fishing in the free agent market where the returns are terrible. There is an opportunity cost to drafting pitching (not drafting position players), but having to find premium position players is far less perilious in free agency.

In terms of further research, I've compiled a list of pitchers who signed smaller contracts (less than these contracts, but more than half of the contract value). I intend to perform a similar exercise and see the bang for the buck tams get. Also, it occurred to me that it would be interesting to see whether the pitchers who performed porly had red flags in terms of pitch type usage. I also intend to do the same thing for position players, in terms of going back and updating the dataset in a more comprehensive manner.