Dollar per WAR model broken? Um, no

Cool Papa directed us to Tango's response to complaints that the $/run (WAR) model might be broken.  I did scan Tango's initial response, though not the comments yet.

...............

Re: Tango's having to address protests that the model is "broken" -- it seems a lot of people don't get what the model is.  A person who asks whether the model might be broken, definitely does not understand the model, because it can't be broken.

That's okay.  Here are a few things we need understand and $-per-run and inflation:

.................

1) What $4.5m per win means.  It  just means, here was the average cost ACTUALLY PAID.

It is crucial to understand that this is a description of what businesses did -- nothing more, and nothing less.

MLB, in past years, paid $450-500 million total for 1,000 runs* total.  That's all $/WAR means.   It doesn't mean a thing more than that.

.

That $4.5 - $5.0 average can, and will, change next year.  That doesn't break the model.  It updates the model's data return.

.................

2) Or that the simple dollars-available-vs-runs-available calculation is very simple, and is 20 years old.  John Benson CPA laid the process out in the 1980's.

In a considerably more sophisticated form, we might add.  Benson's $/WAR model adjusts many times each winter -- to describe reality more dynamically.

$/WAR might be $4.5m for the 2008 offseason.  But it also might be $4.0m in November, $5.0m in January, and $6.0m in February.

One figure is as "correct" as the other.  They were never anything more than descriptions of what businesses are doing.

.

This model can't be broken.  You might as well say that trading sticks for carrots is broken.  It's as fundamental as economics gets.

How do you "break" a model that says, on eBay, $200,000 were paid for 1,000 cuisinarts?  And that on average, cuisinarts sold for $200 last year?

The only way you break the model, is to misuse that information.  Notably by assuming that $200 is the "correct" price for the cuisinart you are looking at today.

For one thing, the average cost of cuisinarts has probably changed since last published.  Right now, they're probably averaging $100 or $300 or something else.  You want to use current information, right?

But there are many reasons that the average-cost-paid $200 cuisinart might be worth $175, or $225, to you.    If cuisinart supply is down, and your wife really wants one, then the $200 average, that other people paid in the past, is meaningless to you.

...................

True, it would be a slight refinement on the model if you could FORECAST dollars available.  And say, in 2009 teams will spend $1,200 million on free agents.  Then you could just calculate and say, there are 2,000 runs available.   That would tell you, in advance, that $6m per win would be the average paid.

That's what Benson does in a roto keeper league -- he knows how many $ are left in each team's budget.  But I don't know how to guess what MLB teams will pay in the future, do you?

Even then -- $450k or $600k per FA run is only an average.  It's information.  It's not a traffic law.

.

...................

3) And many of the columnists definitely do not seem to get that AVERAGE cost paid does not equal CORRECT cost TO pay.

We're only too happy to scale that $4.5m down -- Chone Figgins should not be paid his $5m per win -- but for some reason v-e-r-y averse to scaling it up -- Figgins is worth more than $5m per win in this specific city under these specific circumstances.

If $5 is an average, and sometimes $4 is appropriate, then obviously sometimes $6 must be appropriate under proscribed circumstances.

Just not in my city, right?  :- )

Comments

1

The only point I don't see you making that is important about paying free agents is that they generally tank more than they provide a pleasant surprise (that's my intuition at least).  Let me state it this way.  If you made a bar graph of the dollars spent per win for all free agents, you'd see an extremely assymetric distribution extending to positive infinity on one side (actually it is worse than that because you can pay money for worse than replacement performance) and zero on the other.  What is really desirable is to know the median $/WAR value and that is what you want to shoot for, not the mean $/WAR.
Regarding your general point, surely you can pay more than $4.5M/WAR and have it be a good idea even from a finances point of view.  Those playoff games make money and have a significant impact on attendance the following season.  Making the playoffs, versus missing them, can make a team tens of millions of dollars, so the player acquisition that converts you from an also ran to a World Series winner is worth a lot.

2

is a factor that teams really would be stupid not to analyze, and to the Nth degree.
...............
I definitely appreciate your point that in theory, it's a lot easier for a Jason Bay (Richie Sexson) to lose 70% of his value than to add 70% to his value.
...............
That said, a basic tenet of Stars & Scrubs is that it's much easier for a Civic like Russ Branyan to collapse and be worth nothing, than for Ichiro to do so :- )

3

I've been resisting the urge to launch a rant about the whole "$/WARPed view of the world" paradigm for awhile.
I tired long ago of most of these arguments, because too many of the people wielding the info like cyber-Rambos with the newest BFG-5000 don't get just limited in scope this toy happens to be.  It doesn't apply to Frosh-to-Junior years, or arb players - or even internal salary extensions, (based on my understanding).  It's just the market value of FAs in a given year. 
Attempting to discuss monetary matters in MLB in a "logical" fashion is a fool's errand.  It cannot be done, because the MLB system is so bizarrely contrived with so many disperate moving parts that it's long ago passed beyond comical, past farce, into just plain insane. 
And, of course, it's a moving target.  Today, USA Today had a major article on the whining from Boras about collusion.  Why?  Because the clubs are suddenly unwilling to spend 8 figures and give out multi-year contracts to guys over 36.  Apparently, Boras thinks it's a criminal act for people to learn from mistakes and actually make more reasonable decisions.
Ultimately, the real sad part of all this is that most of those bowing at the temple of $/Warp don't understand that $ are only loosely tied to production.  The truth is, dollars are MUCH more tied to BUTTS IN SEATS.  Between two guys with identical RC/27 scores --- the guy who has a more marketable image is going to get the better deal.  Hero -- anti-hero -- doesn't matter.  If you're a NAME, you can get a deal.  If you're just "productive", but nobody knows who you are ... forget it.  Abreu was a SABR demigod, but the Yanks saw him as expendable, because NOBODY in NY was buying a ticket to see Abreu.  Abreu was old and busted -- Granderson is the new hotness.
Many star-nerds were horrified at Ichiro's deal, based on his sabr-lacking production chart, coupled with his age.  But Seattle's market is MASSIVELY influenced by the Japanese connection, (which is the real reason Ichiro makes the ASG each year).  The teams know that 75% of them won't make the playoffs.  Yeah, they may WANT to win, but the more immediate concern for most is -- break even, or make a profit.  So, you sign Griffey, because you KNOW he's going to put butts in seats, (even if his production tanks).  You sign Branyan for pennies on the production dollar because nobody on earth pays to see Branyan, (at least not before 2010).
Nobody wants to pay Washburn, because nobody really likes watching Washburn --- even if he's pitching well, and winning.  He's a 2400 baud modem in the modern world.  Sure, he may get the job done, but I'd rather be watching the Flintstones than downloading another 4-K, 2-walk Ambien from Washrag.  Who cares that he was the ace of a staff that won a World Series.  Nobody followed that team because of him, even then.
For all the love of glove that Beltre gets, let us not pretend for one second that the major contract he got from Seattle wasn't solely and completely a result of his Mt. St. Helens season. 
You think Dunn was signed by the Nats because they expected they'd WIN more?!?  IIRC, their win total went down one game, (not that the diff between 58 and 59 really matters).  They signed him because, if you know you're going to be losing, some people WILL pay to see massive home runs.  It's a bone to the mongrel hordes of baseball fans desperate to find an excuse to go out to the park. 
But, hey, if someone wants to spend their times computing what the FAs on the market might be getting from year to year ... no skin off my nose.  But, given the Hudson River clarity of the MLB monetary system, I'd think even M.C. Escher would get a headache if he stared at it too long.

4

Suffice it to say, that every age-31 Richie Sexson comparable is not the next age-31 Richie Sexson.
Sexson was the worst case scenario for a player of his age and ability, and a rare one at that.  Cyber-Seattle would do itself a favor not to call every Bay type "the next Richie Sexson."  Jim Edmonds would also have been "the next Richie Sexson" at 31.
Bay's b-ref.com comps, beginning at age 31, averaged 6 more years at 125 OPS+ ... including J.D. Drew, Geoff Jenkins and Jim Edmonds, who are still pushing that 6-year figure out.

5

Attempting to discuss monetary matters in MLB in a "logical" fashion is a fool's errand.  It cannot be done, because the MLB system is so bizarrely contrived with so many disperate moving parts that it's long ago passed beyond comical, past farce, into just plain insane.
Moving parts is another good way to put it.
I guess my #1 pet peeve is the current tendency for smart guys to believe that they've captured all the variables.  And then to condemn those who dissent from their (naive) pronouncements.
Don't mean to get sour about it, but the dogma is pervasive these days.  We're suffocated by the "correct/incorrect" pronouncements on a day-in, day-out basis.

6

Learn from people what they have to offer and be patient with their weaknesses.
 
 

7
Taro's picture

There were also a fair number of guys (in his comps) who were done as impact hitters at age 31,32, 33.
I think it comes down to whether or not Bay is on performance inhancers (so he can dodge normal aging processes). All three of those guys I've been suspicious of in the past.

8
Taro's picture

It comes down to how much how much you think a guy is worth currently, and how you think he'll decline in correspondence with how you think the market is going to develop in the future.
$15mil over 5 for Bay basically gives him '07 market FA value in the short term, and expects him to hold his value or decline minimimally unless the market recovers strongly. Its just a bad gamble.
I expect average GMs to pay $3.5-4.5mil per W this offseason, and I except skilled GMs to find the bargains (Z, Epstein), and unskilled GMs to overpay (Bavasi, KC types).

9

ALL GMs get bargains.  All GMs overpay.
Bavasi signed Ibanez, who was a 5-year production explosion for pennies on the dollar.
And Atlanta, for all of the brilliant FA pickups they made throughout their 15 year run, traded for Hampton.  MLB is set up so players hit FA just as they begin to DECLINE.  So, the aggregate win/loss on FA contracts is probably a loss ... as few sub-27 players are in the FA bin. 

10
Taro's picture

Well I mean overall Bavasi had an awful record in FA/trade. Just awful. I'm not convinced Ibanez/Johjima/Carlos Guillen were  Bavasi decisions either, but thats all in the past.
I generally don't even like seeing GMs pay market value for a FA, unless there are truly no other options in trades/farm/FA available. Market value  FA $ per W is an overpay in itself.
If you're going to shop for a guy in FA or trade, it just makes more sense to target value guys like Figgins or Granderson, or look for undervalued unproven assets in FA/trade like Seth Smith or Gabe Gross. 4/60 for Jason Bay is just sloppy.

12

Taro, you have the unfortunate mental gift to want to view the monetary portion of MLB through a prism of logic.  IMO, you'd have better luck catching flies with chopsticks.  The system isn't logical -- largely because the players and owners both wanted it this way.
The typical FA is age 29 and up.  As a group, they *ALL* are in decline.  So, spending money on Figgins or Bay or Teixeira really isn't different.  Only in the extremely rare case will a 25-year-old actually make it to free agency, (Beltre) and actually have "some" upside still remaining.
So, why do the GMs overpay for pieces that they know with 100% certainty are going to be declining?  Because they wanted it that way.  The guys hitting FA are the guys who have just established their levels.  So, whether you're signing Vlad for 6/80 or Sexson for 4/50, you're making the same choice -- to grossly overpay someone several years in the future.
And 95% of the fans could care less about $/WARP or age-decline rates.  They love seeing big name stars ... guys who hit HRs ... or steall 100 bases.  Signing Vlad paid immediate monetary dividends to Anaheim - and signing Sexson and Beltre did the same for Seattle.  In BOTH cases, the inevitable decline arrived.  For Seattle, it came earlier than expected.  For LAA, they had managed to stock their team with a dozen productive pre-FA players. 
The ultimate winner in those two cases had almost nothing to do with the actual decisions to sign the toxic contracts.  EVERY FA contract over 1-year could be described as Toxic.  The Angels won because they managed to develop a bunch of cheap talent that curtailed the critical need to revisit the FA market so frequently.  Seattle developed Lopez and Yuni.
But, the side effect of the toxic tails to these FA contracts is that the club can dump these players with less financial backlash as they fade.  When Oakland was dumping their stars IN THEIR PRIME, the fans didn't like it.  When the Braves started dumping their FADING stars in their sunset years, the fans grumbled, but understood -- you can't keep winning if you're paying 8 figures to a bunch of guys who are only average today.  The toxic-tail contracts are a get-out-of-jail-free card for the GMs, (for most fans).
The players agreed to a system designed to make anyone who could survive into their 7th year filthy rich.  Is the system stupid?  Absolutely.  Are the decisions on salary insane?  Obviously.  Is that going to change?  Not a chance.  In truth, a large chunk of the FA dollars can be viewed as longevity pay -- JUST LIKE THE REAL WORLD.  Most Americans don't get raises on merit -- they get raises on time served.  And years of experience is one of the primary factors in the arbitration process, too. 
The SABR-savvy crowd "want" their to be a connection between production and dollars.  So, we get $/WARP -- the ultimate dollars for production number.  But, that doesn't change the reality that half (or more) of the choices about payroll have little to do with production.  The focus is on PERCEPTION.  If a club "appears" to be trying to win, the fans will reward them.  If they aren't, the fans will stay away.  Last year, Seattle could get away with only buying from the close-out rack.  But, due to the success of 2009, the club HAS to give the APPEARANCE of trying to win in 2010. 
The comments throughout blogdom confirm that if the club were to actually have gone into 2010 with Tui, Saunders, Carp and Moore all starting, the fans would've revolted.  I personally still believe that would in fact be the best, most efficient route to building a sustainable winning franchise.  Instead, we have Figgins signed, and the club continues to look for additional IMMEDIATE help.  The fans want "win now!"   And in doing so, they make the same error that every GM who has signed any FA over the age of 27 to a multi-year contract makes.  That's MLB today. 

13

I think that much of the snarky behavior comes from people feeling the obligation to convince others that they have a better way of looking at things.  Once your point has been stated clearly, there isn't much else to say or do.
Examples:  I think quantitative analysis is fun, but I think it is of dubious value when it comes to differentiating a good GM from a bad GM.  The job is just too difficult and the tools too blunt.  For this reason, I don't understand why people get bent out of shape over a transaction not following sabr best practices.  I also hate the notion that you can only judge the quality of a deal based on the information widely known to the public at the moment of the transaction and attributing an unexpected success to luck, rather than acknowledging somebody may have known something I didn't and it helped make a better decision than I would have.  These are my pet peeves, but I'm not a missionary by spirit, so I just move on to the next comment when I see things that make no sense to me if they have already been beaten into the ground. 
Now I acknowledge that it is different when ideas I don't like are associated with me because I am associated with the product, but if that makes me behave anti-socially I should just turn off the comments.  You don't seem to have a problem with different ideas, nor do Lonnie and the mods at Mariner Central, which is why these are the only places I bother to comment.  I do think that other websites do have a really hard time having ideas posted on their sites that they find wrong.  I still read the sites, but I don't comment.

14
CA's picture

So, in layman's terms:  Calculations are made looking backwards at the average cost per win on the free agent market.  It is then totaled and applied to individual players speculated upon? This despite the wildly disparate inputs that went into the calculations?  That information is then homogenized and it is determined that x player at y salary is a no go?  If that's true I'm not sure why anyone would give the theory a second glance.  

15

People do give the theory second glances :- )
..............
Granted, it is helpful to know that, in the past, teams have been averaging $475,000 per run on the FA market. 
Then you are AWARE OF how much you are deviating from your little baseline, so that you can make a good judgment as to whether it's advisable right here.
Again CA, I think you and I are on the same page here... using $/WAR as one more compass by which to steer, great .... using it as a traffic law makes very little sense...
I don't believe I've ever read, a single time anywhere, "Granted, Shlabotnik's production was paid an average price of $11m last year, but to this team right now maybe he IS worth $14m" ... not one time...  and that is what makes the whole sabr-industry paradigm more harmful than helpful...
..............
On somebody like Bay, Taro's concerns about the fit for the park, possible bad defense, possible aging, etc .... $450,000 per run above "replacement" would be good to be aware of, so that you don't (say) DOUBLE that in a gray-area situation :- )
Problemo here is, we're already talking about being below the average $/WAR and THEN the argument is made, offsite, that it's brain-dead...

16
Taro's picture

I'm with you there Sandy. In general FA isn't a good idea. I do think it becomes neccesary at times though, and there are some guys that slip through the cracks.
If market value is $4mil per W in FA and "True value" is around $2.5mil per W, then the closer you can get to true value in FA, the better chances the deal will work out for you. If Figgins and Harden are getting paid like arbitration players and Bay like a normal market value FA, then odds are that the Figgins/Harden deals are probably more likely to work out.

Add comment

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd><p><br>
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.

shout_filter

  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.
  • Web page addresses and e-mail addresses turn into links automatically.