30 November 2011

A second look at Yu Darvish

A few weeks back we looked at how to translate Yu Darvish's performance in the Japanese Player's League to Major League Baseball.  That projection system put Darvish as being capable of throwing an FIP of 2.82 in a league average environment.  That is rather exceptional and seemed unrealistic.  Instead, after looking at the limited data I had at hand...I suggested that he looked more like a 3.50 FIP pitcher.  A couple issue with the translation is that it was based on three data points and those three data points were awfully successful data points with greater weight being given to the pitchers who threw more innings.  In other words, there was a bit of a survivor bias.

To make a different and potential better system, I compared that model's prediction to the actual performance of six recent transitions from JPL to MLB: Hiroki Kuroda. Dice-K, Kenshin Kawakami, Colby Lewis, Koji Uehara, and Ryota Igarashi.  These six pitchers were given equal weight in the translation.  Last time, I present the coefficients as values to divide.  This time to make things less confusing in the future, these values are to be multiplied by the JPL numbers.
Original System Coefficients
K - 0.93
BB - 1.81
HR - 1.24

New System Coefficients
K - 0.98
BB - 2.6
HR - 1.93

15th Percentile Coefficients
K - 0.83
BB - 3.34
HR - 2.53

85th Percentile Coefficients
K - 1.13
BB - 1.83
HR - 1.33

One of the improvements here is that we now have a range that covers 70% of the possible outcomes.  Here is what Darvish's projections look like now over 200 IP:
15th percentile - 177 K, 137 BB, 15 HR, 4.49 FIP
50th percentile - 209 K, 106 BB, 12 HR, 3.47 FIP
85th percentile - 240 K, 75 BB, 8 HR, 2.43 FIP

If Darvish maintains this performance over five years, these are the following WARs and associated values he would be worth:
15th percentile - 5.9 WAR, 29.5 MM
50th percentile - 13.9 WAR, 69.5 MM
85th percentile - 24 WAR, 120 MM

It appears that my first approximation of Darvish's value may have been a bit bullish.  It was based on an aggressive projection and an aggressive assumption on contract inflation (about 10% as opposed to my normal 5% assumption).  That said, it was a pretty decent approximation of value given that I did not think long about it.  Now, I would say the a more reasonable approximation would be some a mix of 70 MM between the posting fee and a five year deal.

What will make it worth Yu Darvish's time?

Darvish made 6.6 MM last year for the Nippon Ham Fighters.  If he does not agree to a contract after this year or next, Darvish would be a free agent after the 2013 season.  If he was to stay in Japan, he could see 13 MM over the next two years and then come across the Pacific as a free agent.  If he maintains his play, he should be worth about 15 MM in the open market.  Over five years, he could earn 58 MM.  If I was his agent, I would be floating 12 MM per year as what to expect in order to sign a contract, but be willing to accept something as low as 45 MM.  There has to be some concern about getting injured in the next couple years.

Suggested move to get Darvish:
30 MM posting fee / 5 years, 40 MM

Likely move to get Darvish:
50 MM posting fee / 5 years, 50 MM

3 comments:

John said...

From a purely abstract standpoint, would you be recommend any team commit $100MM to Darvish?

From the O's standpoint, with their clear need for impact pitching, doesn't an aggressive bid for Darvish make more sense than just about anyone else on the free agent market?

Anonymous said...

I think Darvish makes a lot of sense for 100 MM. It likely is not a great value move, but it should be close to even value. With clamping down of money spent on amateurs...it might be a good way to get young talent in 4 years or so. Question is how much talent is worth the front end investment of about 80 MM? That is a risk.

To be clear...I cannot see the Os being competitive for another 3-5 years.

-JS

Anonymous said...

I wouldn't go above $80M total. The Orioles need to be finding WAR on the discount, not paying fair market value.