Where Crowdsourcing Can Go Wrong: Likely Contracts for Tanaka, Garza, Jimenez and Santana

The on-again, off-again news of the posting of Masahiro Tanaka has thrown the offseason market for “front line” starters into disarray.  Not only have we not seen any of the offseason’s top three (or four) starters sign, but to my knowledge, we also have not heard any credible information from front offices regarding offers or likely offers.  All we have is information about what agents are seeking, and our own speculation.

On the topic of speculation, FanGraphs has done some great features lately on crowdsourcing contracts.  Yesterday, Dave Cameron posted a crowdsourcing survey for Tanaka, posting the results that afternoon.  In terms of years, the mean (6.3) and median (6) were about what we expected, and in terms of average annual value, the mean ($19.7M) and median ($20M) should only surprise you if you haven’t been following Tanaka news.  But what really caught my eye was the distribution, with about a fifth of respondents going at or above $22M a year.  And I was also stirred by part of Cameron’s conclusion:

As a general rule, the crowd has been consistently too low on large contracts; missing on Cano by two years, missing on Ellsbury by one year and $3M in AAV, missing on Choo by two years and $3 million in AAV, and missing on McCann by one year and $2M in AAV. If this estimate follows the trend of previous forecasts for big contracts, maybe a more realistic projection would be 7/$154, or almost exactly equal to the Ellsbury contract.

There may be no one better versed on the application of “winner’s curse” to baseball than Cameron, so I suspect what I’m about to say is not news to him.

I first heard the term “winner’s curse” in relation to the initial public offering for Google stock in 2004.  It was a bidding process, but what caused alarm among some investors, if memory serves, was that the number of bids was expected to be about four times greater than the number of shares being offered.  In other words, only about 25% of the hoped-for shares would be distributed.  But the bidders were “the market,” and if there were quite a few more bids than shares, one would expect the winning bids to outpace that market.  In other words, the winners of shares would immediately find themselves with shares that were valued by the market at quite a bit lower than the bid price.

We’re fond in the offseason of mentioning that it only takes one GM to inflate the market for a player.  It only takes one Arte Moreno to blow the bidding for an Albert Pujols or Josh Hamilton to the stratosphere, and when there’s a significant difference between supply and demand, the possible “winner’s curse” applications can also go through the roof.  Again, I’m not saying anything new here.

Perhaps it’s time, then, to stop using the mean or median in crowdsourcing exercises as the benchmark for what a likely contract is.  With larger contracts, there are few players to sign, but a lot of teams interested in signing them.  It’s the perfect recipe for “winner’s curse,” and although it means that we’re just guessing, perhaps we ought to use a different number from the crowdsourcing results, such as a number one standard deviation above the median.

Before getting to the four players in question, a few caveats:

1.  For this exercise, we’re assuming that a player will always take the “best” offer given to him — but of course we know that this is not always true, and that the exceptions are not always confined to examples of hometown discounts.

2.  Because we have two numbers here (dollars and years), “best” can sometimes be in the eye of the beholder.  We can’t use total value as “best” because of course, it’s much better for a player to sign a 3 year, $24M contract than a 4 year, $24M contract.  We also can’t use highest average annual value as “best” because I suspect most players would prefer a 4 year, $31M contract ($7.75M AAV) to a 3 year, $24M contract ($8M AAV).

3.  In terms of changing numbers around, even though dollars can be altered to whatever degree we like, we can only deal with (relatively small) integers for years.

4.  Even if we knew for sure that adjusting upward from crowdsourcing results was a reliable methodology for “large contracts” (and we don’t), we don’t know that all or any of the contracts for the below pitchers would be “large contracts.”  We might find later that the idea worked a lot better for the Tanaka contract than it did for the other three contracts.

5.  Option years and vesting, no-trade, and opt-out clauses can have significant impacts on contract length or value.  We’ll just have to pretend they don’t exist for this exercise.

6.  Qualifying offer, shmalifying offer.

MASAHIRO TANAKA

Yesterday, I floated the idea that a Tanaka contract could exceed $150M.  The median numbers from the FanGraphs crowdsource was 6 years, $20M.  As Cameron noted in the excerpt above, though, he thinks 7/$154M might make sense based on the discrepancy of recent crowdsourcing for large contracts and the actual contracts signed.  I think that’s reasonable, and it puts the AAV for the contract at $22M.  An eight-year contract probably would not sport the same AAV, but adjusting to $20M would leave us with 8/$160M, which does not seem like much of an upgrade from 7/$154M.  After signing a large contract, I’m not sure there’s much incentive to “play it safe” with an extra year (he’ll be pretty “safe” no matter what), and as long as Tanaka is a vaguely serviceable starter at that point in his career (he’d be age 32 for all of the eighth year of a contract), he can probably get an AAV higher than $6M for that season.  In addition, the time value of money comes into play — by stretching out payments an additional year, he’d essentially be deferring payment of some of his salary to the eighth year.  So let’s settle on these two possibilities, based on adjusting the crowdsourcing numbers up:

7 years at $22M AAV ($154M total) OR 8 years at $21M AAV ($168M total)

MATT GARZA

FanGraphs did its crowdsourcing for most free agents earlier in the offseason (October), so it’s possible that they’re not directly comparable to the results for Tanaka.  Not much we can do about that.  For purposes of this “adjusted crowdsourcing” exercise, I think we can use FG’s unrounded results, applying an adjustment of about 20% to both years and AAV (taking Cameron’s lead from his Tanaka analysis) and rounding where appropriate.

The mean length of years for Matt Garza was 4.4, with a mean AAV of $14.7M.  Steep!  We’ll make that five years for Garza, and $17.6M AAV.  Those numbers shock me more than a little, but that’s probably par for the course in this exercise; we expect these numbers to be somewhat higher than most of us would guess.

4 years at $18M AAV ($72M total) OR 5 years at $17M AAV ($85M total)

UBALDO JIMENEZ

I’m probably not the only person who would prefer Jimenez to Garza; in general, I do like injury-prone players more than most, but when we’re talking about contracts this big, you might struggle to get equal value (in terms of performance) even if the player doesn’t miss any time.  Still, despite some sustained dominance in Colorado and a great second half in 2013, Jimenez did not get as much love from the FanGraphs crowd as Garza.

The mean number of years for Jimenez was 3.6, and the mean AAV was $12.2M.  A relative bargain!  Our adjustment stumbles a bit between 4 years and 5 years here (20% above 3.6 is 4.3, but 30% is 4.7), but our AAV would be around $14.6M or higher.

4 years at $15M AAV ($60M total) OR 5 years at $14M AAV ($70M)

ERVIN SANTANA

The FanGraphs respondents were even more leery of Santana than they were of Jimenez — this time last year, the market thought that Santana was only worth maybe $10M AAV for a single year, but an outstanding 2013 is likely to have opened many doors for him.  The rumor mill does not seem to indicate that Santana is anyone’s top choice (personally, I’d prefer Randall Delgado, even at the same salary), but he has been durable for quite some time (average of 194 innings pitched in the last eight seasons) and a 3.69 xFIP suggests that 2013 was not a mirage.

In the FanGraphs results, the mean number of years for Santana was 3.5, and the mean AAV was $13.3M.  We have the same problem with years that we had with Jimenez, with 5 years being even more of a longshot.  The AAV should be around $16M.  Cripes, that sounds terrible.  I’d hazard that Santana’s public demands for $100M may have inflated these numbers a bit, although ESPN’s Jim Bowden predicted a 5-year, $75M contract for him.

4 years at $16M AAV ($64M total) OR 5 years at $14.5M AAV ($72.5M)

Although we don’t have the full distribution of survey responses for the last three of the pitchers above, we can be relatively sure that some of the respondents filled in guesses like the ones above.  I have a very hard time believing the above numbers for Garza and Santana, in particular, but I would not be surprised at all to see contracts like the ones floated above for Tanaka or Jimenez.  And remember — this exercise is about numbers with which relatively few of us are comfortable, by explicit design.  It only takes one GM to inflate the market for a player.  We’ll see what we see as the offseason progresses.