A basic model for free agent pricing

[villain612]

The Reds did something with Homer that seems to be a dramatic departure from the market. Much like the contracts of Pence and Granderson, it's likely that the Reds significantly overvalued Homer relative to the market using projected WAR as a proxy for the market.

It provides insight into how the Reds estimated Homers true value/skill. Are they smarter than the projection systems on this one? We'll see but again, even if they are right, the structure of his contract doesn't seem to offer tremendous surplus value or manage risk particularly well. In other words, as a bellwether, it's not strong support for the notion that the Reds outsmarted the market.

I'm not a WAR expert other than having a general familiarity with it, but is the Reds deal that much of a dramatic departure from the market when you compare it to what the Yankees did with Tanaka? I'm not trying to denegrate the Japanese leagues (I admittedly know nothing about them), but isn't the Yankees contract with him a much larger departure from the normal market considering the guy has never pitched in MLB and got a much richer deal than Bailey?

[kaldaniels]

As said, using this system Bailey was an overpay.

So let's discuss.

Assuming a club with decent payroll spending (all NL Central teams would qualify).

If after the 2013 season, Bailey was a free agent, would most teams interested in Bailey offer 4/55MM or 6/105MM?

I think things would skew towards the latter, which is why I think Bailey himself is the outlier, not the Reds offer in specific.

[757690]

You just jumped the shark.

The reality is that projected WAR predicts player salary very, very well. Its not a shocker that a model based upon that relationship predicts contract duration and the contract's total value very, very well. You can't explain that away with references to ice cream, the weather channel or lunar phase.

The Reds did something with Homer that seems to be a dramatic departure from the market. Much like the contracts of Pence and Granderson, it's likely that the Reds significantly overvalued Homer relative to the market using projected WAR as a proxy for the market.

It provides insight into how the Reds estimated Homers true value/skill. Are they smarter than the projection systems on this one? We'll see but again, even if they are right, the structure of his contract doesn't seem to offer tremendous surplus value or manage risk particularly well. In other words, as a bellwether, it's not strong support for the notion that the Reds outsmarted the market.

I can guarantee you that the Reds did not give Bailey 6 years instead of 4 years because they projected his WAR in 2014 to be 3.5 instead of 2.5.

The Reds gave Bailey 6 years instead of 4 because he is only 27 and has 5+ seasons of MLB experience. Period. I am sure they considered his projected WAR when determining the amount of salary per year to give. But they considered his age and experience when determining how many years to offer.

This is such an obvious, simple fact that I can't believe it is even being discussed.

And btw, I do think the Reds overpaid for Bailey, and I would have rather they traded him and signed a short term replacement like Cameron suggested. However, that is irrelevant to this discussion.

[villain612]

I can guarantee you that the Reds did not give Bailey 6 years instead of 4 years because they projected his WAR in 2014 to be 3.5 instead of 2.5.

The Reds gave Bailey 6 years instead of 4 because he is only 27 and has 5+ seasons of MLB experience. Period. I am sure they considered his projected WAR when determining the amount of salary per year to give. But they considered his age and experience when determining how many years to offer.

This is such an obvious, simple fact that I can't believe it is even being discussed.

And btw, I do think the Reds overpaid for Bailey, and I would have rather they traded him and signed a short term replacement like Cameron suggested. However, that is irrelevant to this discussion.

I'm no expert but just to add, in addition to considering age and experience and the WAR projections, wouldn't the condition of the current market for starting pitching be a factor too? I just think the Tanaka deal was much a bigger deal than people realize currently.

Also, and I might be way off on this, but a lot of teams have new TV deal money kicking in over the next few years, thus allowing much more flexibility on their payroll. Could the Reds have done the deal thinking its better to buy him now than try to compete for a #2 starter in the FA market when the new TV money kicks in??

[dougdirt]

Another thought: Is it possible that baseball teams don't really think pitcher WAR is as accurate as some fans do?

[jojo]

I'm not a WAR expert other than having a general familiarity with it, but is the Reds deal that much of a dramatic departure from the market when you compare it to what the Yankees did with Tanaka? I'm not trying to denegrate the Japanese leagues (I admittedly know nothing about them), but isn't the Yankees contract with him a much larger departure from the normal market considering the guy has never pitched in MLB and got a much richer deal than Bailey?

He really doesn't have a good comp s the posting system just changed etc. but he kinda smoked Dervish and daisaka.

[jojo]

Another thought: Is it possible that baseball teams don't really think pitcher WAR is as accurate as some fans do?

You sill haven't explained why the model works but it shouldnt be treated like it does.

[kaldaniels]

I think the system is decent.

But if I'm not mistaken jojo, I haven't seen you address the point that I believe in your 5% accuracy claim, you are allowing over and underpaid contracts to cancel each other out in the calculation. Am I seeing that correctly?

As a whole, the system comes out accurate when you sum every thing up. But when you go player by player, I'm seeing a lot more than 5% error when you go one contract by one, and find the mean or median of the percent error of each one. (Granted, it nails some guys)

[jojo]

I can guarantee you that the Reds did not give Bailey 6 years instead of 4 years because they projected his WAR in 2014 to be 3.5 instead of 2.5.

The Reds gave Bailey 6 years instead of 4 because he is only 27 and has 5+ seasons of MLB experience. Period. I am sure they considered his projected WAR when determining the amount of salary per year to give. But they considered his age and experience when determining how many years to offer.

This is such an obvious, simple fact that I can't believe it is even being discussed.

And btw, I do think the Reds overpaid for Bailey, and I would have rather they traded him and signed a short term replacement like Cameron suggested. However, that is irrelevant to this discussion.

Youre arguing that the Reds basically pulled a number out of the backside because Homer is 27 and has 5 years of service time. If what you say we're true about the Reds not projecting some level of production and trying to attach a monetary value to it, then, that paints them in an even worse light than if they had overvalued his projected production.

You're so hung up on trying to discredit WAR and caught up in some flaccid snark, that you're really missing the whole point.

Whether the Reds or any GM specifically uses WAR is besides the point. Projected WAR serves as a great proxy for whatever valuation system a GM uses because it predicts salary well. It offers a prism to evaluate contracts relative to the market, I.e. how well a GM estimated a player's true talent and valued it relative to his competitors. This prism highlights Homer's contract as one that is significantly different than would be predicted and gives a framework to try and understand why.

[dougdirt]

You sill haven't explained why the model works but it shouldnt be treated like it does.

Because it doesn't work on individual contracts with anywhere near the accuracy that grouping them altogether tries to say. Again, the error bars are rather large on some of these contracts. It is kind of like BABIP. .300 is average, but the error bars are Joey Votto (.370) and some crappy hitter (.220). Sitting around claiming that BABIP is 95% right at .300 and saying Votto and that other guy are going to regress to the mean doesn't really work simply because the overall package says .300 is right. There are variances and in some cases, big ones.

[jojo]

I think the system is decent.

But if I'm not mistaken jojo, I haven't seen you address the point that I believe in your 5% accuracy claim, you are allowing over and underpaid contracts to cancel each other out in the calculation. Am I seeing that correctly?

As a whole, the system comes out accurate when you sum every thing up. But when you go player by player, I'm seeing a lot more than 5% error when you go one contract by one, and find the mean or median of the percent error of each one. (Granted, it nails some guys)

I've addressed it. It's like any other model in the regard you're describing. Concerning whether it works or not, it sure explains a ton of variation in duration and total amount of free agent contracts.

It's just a tool/toy but if Homer was given his contract by the Cards, i wonder if it would be treated with more reverence.....

[jojo]

Because it doesn't work on individual contracts with anywhere near the accuracy that grouping them altogether tries to say. Again, the error bars are rather large on some of these contracts. It is kind of like BABIP. .300 is average, but the error bars are Joey Votto (.370) and some crappy hitter (.220). Sitting around claiming that BABIP is 95% right at .300 and saying Votto and that other guy are going to regress to the mean doesn't really work simply because the overall package says .300 is right. There are variances and in some cases, big ones.

Except it does work pretty darn good on individual contracts Doug. Jeeps, does a correlation get much better than .97 as models go? It's a strange argument IMHO.

[dougdirt]

Except it does work pretty darn good on individual contracts Doug. Jeeps, does a correlation get much better than .97 as models go? It's a strange argument IMHO.

The correlation shouldn't be run on the group as a whole. The model isn't trying to predict the group as a whole. It is trying to predict the individual and it isn't 97% accurate on the individuals.

[jojo]

The correlation shouldn't be run on the group as a whole. The model isn't trying to predict the group as a whole. It is trying to predict the individual and it isn't 97% accurate on the individuals.

I'm not sure how to respond to this other than to say, "That's not how this works. That's not how any of this works."

We'll have to agree to disagree.

[757690]

Youre arguing that the Reds basically pulled a number out of the backside because Homer is 27 and has 5 years of service time. If what you say we're true about the Reds not projecting some level of production and trying to attach a monetary value to it, then, that paints them in an even worse light than if they had overvalued his projected production.

You're so hung up on trying to discredit WAR and caught up in some flaccid snark, that you're really missing the whole point.

Whether the Reds or any GM specifically uses WAR is besides the point. Projected WAR serves as a great proxy for whatever valuation system a GM uses because it predicts salary well. It offers a prism to evaluate contracts relative to the market, I.e. how well a GM estimated a player's true talent and valued it relative to his competitors. This prism highlights Homer's contract as one that is significantly different than would be predicted and gives a framework to try and understand why.

I actually like the model, and its use of WAR. Just not for predicting the length of the contract.