THE BOOK--Playing The Percentages In Baseball

Darn it.  I wish I would have read Dave’s post first, so I wouldn’t have bothered to do the whole write-up and simply linked directly to him:

http://www.fangraphs.com/blogs/index.php/marlins-get-a-potential-bargain-in-jose-reyes/

He went a step or two further by giving more background to what he’d need to do to justify his deal.

Bottom-line: whether with my illustration or Dave’s, he needs to be +19 wins over the next 6 years.

I agree with the overall point, but I’ll repeat my criticism from the Cameron article because you’re using the same $/WAR model.

“I think those $/WAR estimates are extremely generous.

The cost of a Win by year:

2006 — $3.7 M
2007 — $4.2 M
2008 — $4.5 M
2009 — $4.5 M
2010 — $4.0 M
2011 — $4.5 M

Rates have stagnated for 5 years between $4-4.5 M per year.”

What is the data/reasoning behind $5 M per win in 2012 and 5% annual increase between now and 2017?

I don’t know where you got those numbers specifically, but if you want to use 4.5MM$ per win, then you can start Reyes at 4.8 wins in 2012.  Whether you think that is too high or just right is your decision.

As for the future increases: c’mon dude, now you are just arguing.  From the late 1980s through to a few years ago, the increase in payroll, revenue and team valuations was 10% annual.

Then, the world economy has stagnated for the last few years.

What does the future hold?  How the heck do I know.  Is it fair to presume a 5% increase each year for the next 6 years?  How the heck do I, or anyone, know.  What would a fair market of the reading suggest to that effect?

Do you want to make it 2% or something?  Then, fine.  The implicit assumptions the Marlins are making are:
5.0 wins in 2012
4.5MM$ per win in 2012
2% increase in $ per win each year after that

That gives us this:

Instead of arguing how something can’t be right, then present the implicit valuation based on your reasonable assumptions.

So, is it unreasonable therefore to have a valuation of 5 wins for Reyes in 2012?  That’s your decision to make.

The framework is sound.  If you don’t like one particular implementation, then make your own!

Under this Tango/3 scenario by the way, they are paying for 22.5 wins.

I did that, actually I got 21-23 wins too.

I never said you “couldn’t be right.” I asked for your data/reasoning.

I think there is more to this than “just arguing.” It’s a matter of evaluating the relevance of a ~5-year trend in the most recent data.

I got the numbers from the Fangraphs “Value” tab.

I think $5M per win going forward is conservative. 
This new CBA is going to send free-agent values higher.  All the controls on non-MLBPA spending - draft caps, international money, no ML contracts for draft picks, etc. - that money is going to get funneled into FA’s.  The players know it, they’ve been promised it, and that’s why they agreed to the CBA that they did.

I think wins are going to cost $6-7M by the end of this next CBA on the FA market.  Every team except the Rays will have a modern, revenue-generating stadium, more teams will be in playoff contention than ever before, and the game will remain very healthy financially.

This Reyes deal will end up looking very good, I think.  So too will Tulo and Braun’s extensions, which were panned by quite a few people at the time.

Just a question on the math:

Why the 2 significant digits on WAR (e.g., 4.44) and $/win when, as far as I understand it, the -.5 WAR decrement per year is a very very rough approximation. Is the -.5 WAR year more exact than I’d assumed?

It absolutely doesn’t change anything about the results and some might think this is the smallest nit you can pick, but that level of precision on your numbers might confuse people for whom the number of significant digits is a thought-out choice.

Thank you, and good question.  I never, and I mean, never, show more than 1 decimal place.

But in this particular case, I kind of forced myself to hold fixed the 5MM$ a win and the 5% baseball-inflation rate.  So, when calculating the starting point in WAR, I got 4.44.  I was going to just show 4.4, but then it wouldn’t come out to 106MM$, but rather 105MM$.  Then, adding up the gain/loss each year would have come out to a net negative, rather than cancelling out.

Normally, I play around with the numbers so I wouldn’t show anything to more precision than I wanted.  In this case, it was a bit tougher, because my point here was to show that it cancels out to zero.

I got a little more than 19 wins from my estimations/guesstimations as well, right about 21 WAR for 6 seasons.

Regardless, he has a very good chance of fulfulling the contract as we might say by providing 106M worth of performance (give or take).

The only thing that I said is that he likely will not provide such a surplus to make it a “bargain” (the terminology used at FG).

Count me as another one that hates the “he won;t be worth it in the last 2 years of the contract” stuff. That ignores the bulk of the data ... when making an overall analysis.

What primarily matters is that the overall value per cost, and even then from the team’s point of view they may be primarily just looking at how much performance occurs at the front end of the deal when the team is likely “making a run for a title”. I doubt there are any teams that look at the last 2 years of a contract and think “Man, he’s going to be really special during those last two years”. No, they’re looking at it as “In order to get him to play for us for the 1st four years, we have to pay him for those last 2 years”. Every team knows there’s going to be a decline as the years go on in a contract. We should really stop acting like they don’t know.

We should also stop acting like teams should just offer the player a 4-year deal, as if the player would sign such a thing.

This is what I have for 3 to 5 year deals that he would be worth, presuming he’s worth 6/106:

3/62 (almost 21MM$ per)
4/79 (almost 20MM$ per)
5/94 (almost 19MM$ per)

He signed for almost 18MM$ per.

If you remember my little rule of thumb trick: give a player 1 point for each year he signs and 1 point for each million$ per year.

So, at 6/106, that’s 24 “points”.  In the case of the above 3, that’s ALSO 24 “points”.

Basically, for every year less, they’d have to give him 1MM$ a year more.

That’s why you have to look at these deals as if he’s deferring money to future years.

My “back of the envelope” calculation while watching TV last night showed Reyes as a 4.5 win player and that he would need to garner 19.5 wins over the life of the deal for the Marlins to break even.

How should we account for the buyout/option in the 7th year of the deal? Should we tack on $5MM to the deal for the buyout?

Usually, the best thing to do is just count guaranteed money and guaranteed years.  So, if a player has a 5/50MM deal, with a team option for 10MM in the 6th year or a 2MM buyout, you count that as a 5/52 deal.

It’s not a hard-and-fast rule though.  For a contract like Longoria, where you likely expect the option to be exercised, you’d presume so.

Tango/10, I think the disconnect is because people claiming “he won’t be worth it in the last two years” are not considering the potential contracts that you describe.  I think they are implicitly saying instead of 6/106 it should be 5/88 or 4/71, (roughly the AAV multiplied by 5 and 4 respectively).  I think fans act as though the average annual value offered in a contract is independent of the number of years.  This clearly isn’t so, but I think it helps to explain the sentiment a bit.

A quick thought on the idea of structuring contracts so that players are paid for expected performance at each year. So, for example, Reyes would get $106m, but instead of $17.7m/year, he gets $21, $20, $19, $17, $15 and $12. Then the deal technically would be “worth it in the last two years”.

On the one hand, the team is out the time-value of the money paid earlier with that method.

On the other hand, Reyes is paid what he’s worth at any given time, so that in the 4th year of the contract, he is far more tradeable at a $15/$12 salary.
I guess with teams able to send along cash with players, the Marlins can just send along the $9m they earned early on in the contract, negating this last point.

So, it seems that it’s always in a team’s benefit to average it out over the whole contract and structure it as they did with Reyes as opposed to something more transparent in terms of expected value each year.

Does everyone else have the same sense?

Reyes is still fairly young, this deal will be for his age 29 to 34 seasons.

THT Forecasts has him at 5.5, 2.8, 2.9, 5.9 WAR the past 4 years, if you prorate his 2009 cut short by injury to a full season.

I project 3.8 WAR next year for 650 PA, gently sliding down to 2.4 WAR in 600 PA in 2017. 3.8, 3.6, 3.2, 3.0, 2.7, 2.4 for a total of 18.8

Year  Wins $perWin $Value  $Cost $Gain(Loss)
2012   3.8   $4.50  $17.3  $17.7  $(0.4) 
2013   3.6   $4.59  $16.4  $17.7  $(1.3)
2014   3.2   $4.68  $15.1  $17.7  $(2.6) 
2015   3.0   $4.78  $14.3  $17.7  $(3.4)
2016   2.7   $4.87  $13.1  $17.7  $(4.6)
2017   2.4   $4.97  $12.1  $17.7  $(5.6) 
total 18.8          $88.3 $106.0 $(17.9)

I see him as a solid 3-4 win player who is capable of a 5.9 season, but that’s been his career year.

A personal rule of thumb is not to give multi-year contracts into season where the player is projected to be below 2 WAR, especially 1.5 (average). This contract does not do that. Even for six years, this contract only goes to age 34, and I project Reyes at 2.4 WAR in the final season, still solidly above average.

Note Brian is giving 2% inflation rate.

I used the same dollar assumptions as Tango’s chart, changing only the WAR projections. With my WAR values, it would 10% a year inflation for the value to equal the payout in this contract.

How do these forecast methods deal specifically with Reyes’ injury history over the last few seasons?

Brian: my chart is in the main blog post.  The chart you are quoting is an example for someone else who thought I was too optimistic.

Oops, let me try that again

Year  Wins $perWin $Value  $Cost $Gain(Loss)
2012   3.8   $5.00  $19.2  $17.7   $1.5 
2013   3.6   $5.25  $18.8  $17.7   $1.1
2014   3.2   $5.51  $17.8  $17.7   $0.1 
2015   3.0   $5.79  $17.3  $17.7  $(0.4)
2016   2.7   $6.08  $16.4  $17.7  $(1.3)
2017   2.4   $6.38  $15.6  $17.7  $(2.1) 
total 18.8         $105.3 $106.0  $(0.7)

Look like a damn fine contract!

Beautiful!

My projection for Reyes averaged about 0.3 WAR a year decline, instead of 0.5, for ages 29-34, but started from 3.8 - based on his past stats and current age.