Wonderful Analysis By Acer

Acer, one of the posters over at www.nyfs.com, did a wonderful job of analyzing potential outfield candidates for the Mets, in general, and Adam Dunn, in particular. I reprint this, assuming I have his permission, because he reads and participates here (would love him as one of my writers...)

Mack



A question: how do they establish a value for the player in monetary terms



Well first they measure everything in runs. Then, they look at actual contracts signed in the FA market, and use that to figure the value of a run above replacement player. A replacement player is basically a guy you'd get for MLB minimum. The trouble is in estimating how much of anything, offense or defense, is true talent, and how much is context dependent, due to things like park effects.

Park adjustments are usually done based on the impact an environment appears to have on average for all players, but not all players will be impacted equally by different environments. Another issue is that some methods for measuring defense are relatively new. I think that defensive park effects are one area that still needs work. In general though, defensive stats today are much better than they are given credit for, and not really any more unreliable than offensive stats. UZR measures the run value of every event from play by play data. There is still a human element here, as the zone to which a ball is hit and the type of hit, line drive, foul ball, or ground ball, is estimated by human spotters.

In the future the human element will be removed here as well, as hit tracker data is now becoming available in which the angle and speed of the ball off the bat is being measured. On the fangraphs player page where they have the value estimate, you will see it is broken down into columns. They start with estimates of Batting and Fielding measured in runs above or below an average player. The third column is "replacement" used to convert the base from an average player to a replacement player. They have estimated a replacement player at 20 runs below average per 600 PA, so this number will generally come to around 23 for a full season.

Next they add a positional adjustment. This is because defense is initially measured compared to a player at the same position, but some positions are more difficult than others.

For Holliday, we have: 2005 2006 2007 2008 2009 15.1 36.6 49.9 39.0 08.2 Batting 05.6 -5.6 14.2 09.1 05.9 Fielding 17.5 22.2 23.8 20.8 10.8 replacement -5.8 -7.2 -7.4 -06.4 -3.5 positional 32.4 46.1 80.5 62.5 21.3 Runs above replacement (the sum of the above) 11.1 16.4 32.2 28.0 09.6 Dollar Value (in millions)

The big concern with Holliday is he may have benefited more from Coors than the park adjustments show. So he may not have really been worth the $28M or $32.2M shown for 2007 & 2008. However, at $9.6M for this year so far, he's still on a pace to be worth nearly $20M. Some may be surprised at that, given his OPS is only around .800, but he is also a well above average defender. His UZR per 150 games in LF for the last three seasons: 2007: +14.7 2008: +10.9 2009: +13.1 His defensive value at least has survived the move from Coors intact. His UZR/150 for his career in LF is +7.8. Also, note that he's gone from an extreme hitters park to an extreme pitchers park. His lower Oakland numbers aren't a fair estimate of his true offensive talent either. Per Baseball-Reference, neutralized stats:

http://www.baseball-reference.com/players/h/hollima01-bat.shtml

BA OBP SLG OPS .320 .408 .537 .946 - 2008 .293 .394 .453 .847 - 2009 .302 .371 .517 .888 - career

Looking at his monthly splits this year, he had only a .648 OPS in April, then had an .872 OPS in May, and now .814 so far in June. My guess is he's still an .850-.875 OPS guy in a neutral park, but that the .888 "neutralized" number above for his career gives him a bit too much credit. One quick observation from the Baseball-Reference site, is you can get a second opinion here on his Batting Run value, with the stat called BtRuns (Batting Runs). It won't look much different from the fangraphs batting column above, but there are differences in how it's done, including different ways of doing park effects. 2005 2006 2007 2008 2009 15.1 36.6 49.9 39.0 08.2 Batting per fangraphs 08.9 33.3 48.1 34.2 08.4 Batting Runs per BR Now, since these batting numbers are calculated based on the average values of the different events, I also like to look at measures of how clutch a player is. One good one to check is WPA, win probability added, which measures the value of each event in the context it actually occurred. Some of this will be luck, but it is a more accurate measure of what a player has really contributed in wins: +0.45 2005 +2.37 2006 +4.64 2007 +4.87 2008 +0.32 2009 There are no park adjustments here, so he is obviously getting the benefit of Coors, but those are still impressive numbers. since this is less reliable as a measure of talent, I'm really just eyeballing these to see that they measure up to what the other stats suggest he should produce, and they do.

Finally, I like the Baseball-Reference situational stats, for base running and situational hitting: XBT% (extra bases taken) 52% Holliday 39% MLB average 44% Nick Johnson 46% A. Rios 49% Orlando Hudson Productive out % 26% Holliday 32% MLB average 33% Johnson 28% Rios 37.2% O. Hudson BRS% - percentage of baserunners scored 18.4% Holliday 14.0% MLB average 15.0% Johnson 14.9% Rios 15.3% O. Hudson (while this career mark may also be inflated by Coors, Holliday is still at 33/197 = 16.75% so far in 2009, so he still has been well above average in scoring runners) 0,2BAdv - no out, runner on second advanced 47% Holliday 43% MLB average 40% Johnson 45% Rios 49% O. Hudson In terms of run values for these, a 1% difference in ability to score base runners would be worth about 2 runs per 600 PA. But for advancing runners with no out and a man on 2B, a 1% difference is only worth about .14 runs directly. You worry here more that if a player is very poor here he will also be costing runs in other situations spots less easily measured.

For Holliday, I'd roughly estimate him overall as worth, per 600 PA, +25 runs batting, +8 runs defense, +20 replacement, -7.5 positional (right field), or 45.5 runs. I'd pay about .35M per run (note: fangraphs values are more like .45M per run), or $16M/yr For Rios, his values per fangraphs for the last few years are impressive, $12.1M in 2006, $18.9M in 2007, and #24.6M in 2008, but only $3.9M this year.

Here, I worry that the last two years are both due to statistical outliers. In 2008, they had his defense at +23.9 runs. Otherwise, in the last 4 seasons it hasn't been over +10. Likewise, his 2007 24 HR 114 RBI season looks like an offensive fluke. They have him +20.9 runs that year batting, but his next best year is +13.5. Baseball-Reference only has him +17.1 runs for his career, including a +20.2 for 2007. He's just OK in WPA, but still a bit above average in 2006 and 2008 as well as 2007. I'd estimate his bat going forward at only +5 runs, but his defense at +11, +20 replacement, -7.5 positional, for 28.5 runs. I'd pay about $10M/yr.

For Johnson, I have no doubts about the legitimacy of his numbers. Here the question is whether he will stay on the field. The problem with projecting value over 600 PA, is that he has only averaged 337 PA over the last 6 seasons (and I included zero PA for 2007 in that average). He has 325 PA already for 2009 however, and appears to be healthy. What I'll do here is project him at 600 PA, but only be willing to pay him for 450 PA. For his career, FG has him at 109.4 batting runs, and BR at 107.9. That comes to about 22.5 per 600 PA. His best season, 2006, both sites had him worth over 40 batting runs. This season, they have him at 10.1 and 9.8, so he's on a pace for about +20. He also does very well in WPA, already +2.11 this season, well above average 6 of his last 7 seasons, and averaging +2.6 over the last 4. There are also some signs he could be slowing down defensively, as UZR has him as a below average fielder this year for the first time in his career. And, he will turn 31 in September. His career average is +3.6 runs per 150 games at 1B. I'd value him at about +23 runs batting, 0 fielding, +20 replacement, -12.5 positional, or 30.5 runs total. That would be $10.7M over 600 PA, but I'd only pay about $8M due to injury risk.

Hudson isn't a trade target, but I threw him in as a top offseason FA target. Here, my doubts in the offseason were about whether his defense had really declined as much as the stats say. He was above average on the turf in Toronto, but below average in the tougher environment in Arizona, and I suspected park effects. However, the below average defense has continued in LA this year, while his offense surge has also continued. I now suspect he has likely lost a step defensively as he hit his 30s, but also become a better hitter. Here, there is some discrepancy between the two sites on his batting value: _BR _FG -6.6 -05.1 2005 +1.0 +04.3 2006 +5.8 +11.7 2007 +5.1 +08.8 2008 +5.2 +07.5 2009 (so far) Either way he is in the midst of his best offensive season. He has also been a solid WPA player over the years, averaging over 1 win per season over the last 5. Seeing as he is 31 years old, and likely to trend downward soon, I'll be a bit cautious. I'll estimate him at +6 runs offense, -5 runs defense, +20 replacement, +2.5 positional for 23.5 runs per 600 PA. And I'll pay about $8.2M/yr.