I’m not going to re-hash analysis I’ve already done on Holliday in other places (one, because I can draw inferences from that work as the deal isn’t too different and two I don’t have my simulation in front of me right now). The assumptions in any analysis like this are key so I’ll quickly run through those that are the most important, where I stand on them, and how I handled those in the simulation referenced in the link above.
- How good is Holliday’s defense? He derives a decent amount of value from above average LF defense. My projections have him worth around 6-7 RAA next year. There are those in the scouting community that disagree with the metrics about him. Tied in here is the rate at which his defense ages. I must admit, I don’t have a good feel for this factor. When I did previous simulations I looked at two methods 1) a linear decrease based on Jeff Z’s work 2) a step function that had greater decreases as the player got older. I think 2 is more realistic.
- Health. Clearly a significant injury would greatly impact whether or not Holliday will be productive enough to justify the deal. Within the simulation I handled this using a random injury generator. The generator would generate half season’s missed due to injury (10% of the time) and adjust the playing time accordingly. Is this enough? Should I have had a chance of missing a full season or had the chances grow with age?
- Dollar value of a win. this assumption sets the “break even” point for a deal. It determines the amount of production you’d expect for a given contract. Going into this off season the fangraphs value was ~$4.5M per win, but up until this deal the number was hovering around $3.8M-$3.9M so far this off season. I’ll do the math for you, for a ~$17M average annual value (AAV) that works out to about a 0.7 win difference. I assumed 4.5 in the above referenced piece.
Given those assumptions and the parameters of the contract (7 yr ~$120M) how’d the Birds on the Bat do? I think that there’s definitely a chance that Holliday performs up to his contract (especially when accounting for inflation and the potential rising cost of wins). My final estimate (albeit the most pessimistic) had him somewhere between a 30-50% chance to be worth more than the formerly rumored 8yr 128M deal, and I think that estimate is still has value. So slightly below even money to break even on the deal? Not exactly what I would call a win for the team over the long haul. The “badness” is magnified if you believe (as most, including me do) that the Cardinals were basically bidding against themselves, but still gave in to Holliday/Boras. All that being said, I still think I’d rather have Holliday and his contract going forward than Bay and his contract (though I MUCH would’ve preferred 5/80).
OK, enough gloom and doom for now, we’ll worry about the coming years after we hang the banners for the next couple of years right
What does Holliday do for the Cards this year? I though the best way to answer that would be to a quick playoff probability added exercise. I’m basing it off of the work displayed here. If I use Erik’s projection of 87 wins pre Holliday I get a playoff probability of ~56%, while adding in Holliday yields a playoff probability of ~80% for an addition of 24%. Note that these values are based on historic numbers, and do not take into account division strength. To look at that I decided to simulate the division again using Erik’s projections as a starting point. Using the projected wins and a 9 win standard deviation (fairly wide, but Nick mentioned it in the comments of a BtB post, so I thought I’d run with it) to get to “actual wins”. I came up with the cards winning the division ~33% of the time before the signing and ~51% of the time after it.
All in all I think I’m much more indifferent towards the deal than most of the saber community. I don’t like the odds that he performs up to his deal being <50%, but I think I can grit my teeth and deal with it given the playoff probability payoff in the short term. Now, ask me the same question again in 3 years and I may have a very different answer