that’s not good. I’ll e-mail you.

I have write-ups on four things:

PCA (my win shares equivalent rating system)

PythagenMatt Game Tracker (think Pythagorean W% but done for each individual game instead of just on the seasonal run scoring totals…it’s statistically significantly more predictive than regular Pythag in-season though the improvement in accuracy is relatively small…it’s great for looking at in-season trends though)

The Fiato/Souders Matrix (the math theory behind this is actually very simple and can be summed up in one sentence: for any game, for any team, the runs that team scores are equal to the league average run scoring rate plus the perturbations in run scoring context caused by the many interwoven variables that can impact scoring…everything from team offensive ability to opposing defensive ability, opposing starting pitcher, ballpark, home plate umpire, weather conditions, etc…basically the matrix is just a HUUUUGE set of linear equations…solving the matrix solves for all of the variables I can identify all at once)

Defense Neutral Run Average (my DIPS pitching metric…based on the idea that pitchers as a group have little control over the results of batted balls in play…but that individual pitchers have quantifyable impacts on batted ball results and those are defined as the differences between event rates they allowed and event rates the team allowed when they were not pitching)

I’ve done other work that I still need to write up…lots to do…little time to do it.

And hey Jon…if you’re interested we can discuss occasional contributions on the sabermetric end of things…I try to boil things down to simple common sense as much as possible just because very few people have the time or energy to be as geeky as I am…LOL I don’t know how much demand there is in the Mariner blogosphere for yet ANOTHER saberdweeb…but…there ya go.

Matt

Interesting thread.

If you ever want another outlet for publishing your stuff feel free to drop me a line.

Bottom linr on this topic…I can understand you wanting to use the statistics at your disposal…but just be aware that there will be significant biases against sluggers and for part time glove men.

W% = RS^X / (RS^X + RA^X) where X = ((RS + RA)/G)^0.295

You can use the same logic to compare a player to his average expectation and come up with a winning percentage figure (this is the principle behind Bill James’ Offensive Winning Percentage).

Similarly, you can use UZR data if you know what the average run created per unit playing time (inning?) rate is for the position – I’ve calculated those values from the historical database if you’re curious).

You would take the player’s UZR runs (-20 for Dunn in ’08 for example) and add them to his expected runs (a left fielder should creat about 30 pure runs in a full season…Dunn played about 98% of a full season) to get a net runs figure. That’s about 9 runs.

Then you need to figure out what the replacement level is for left field defense (you’re going to discover that the worst defensive left fielders are not all sluggers, but even if they were…that’s real information you need to use in your analysis to describe how important defense is t the position…if teams readily except bad fielders only when they’re plus plus bats…that argues that a -20 defensive rating WAAAYY overstates how teams value defense)…

The fielding replacement level for corner outfielders is on the order of 18 runs created. So Dunn is about a -9 run above replacement fielder…that’s about 0.9 of a win below replacement…not 2 wins.

Unfortuantely…I got my values of what the replacement level is, and what the average runs produced by position are by creating a full win-shares-like sabermetric tool using more modern approaches than James did…that’s obviosuly not a quick back of the envelop approximation for people who don’t want to spend years pouring over the database.

BUT…you can make a rough estimate of those values yourself by looking at the average number of plays made by each position and the relative value of those plays using linear weights.

For example, the typical left fielder makes 300 plays in a full season and each of those plays is worth about -0.1 runs (30 runs total value). This isn’t perfect but it does paint a rough picture…except you don’t do that with first basemen (their putouts are mostly not plays…you need to look only at unassisted putouts and assists) and catchers (their PO are mostly Ks…you need to remove the Ks). You also need to be aware that double plays, catcher assists and outfield assists often have more value than the typical out. But you hopefully get the general idea.

So anyway…the defensively is unfortuantely not simple…another reason why there’s a major problem when you start combining different scales together as though they were linearly comparable.

You general point here is a valid one though…we can gain 20 or 40 runs by the changed outfield defense alone…maybe more….just adding another sabermetric viewpoint you don’t often see.

The justification of doing this is that the replacement level bat will be an average fielder, but this is fatally falwed logic. The idea isn’t to pick some arbitrary net player and describe this amalgum as the replacement level…the idea is to figure out what level of production can be replaced for “free”. It’s demonstrably UNTRUE that the replacement level for defense (by itself) is .500. Every team routinely carries players who produce at below average output defensively AND offensively…even championship caliber teams carry these guys.

The proper way to describe players is by comparing their offense to the replacement level for offense and their defense to the replacement level for defense. The net will not be a single replacement player’s likely production, but it WILL more accurately scale offense to defense in terms of the real value of each for a team.

When you do this, you end up describing Raul Ibanez as a 3 win hitter and a 0 win fielde (three net wins)…of course a lot of other bad fielding left fielders will move up in value too…but the math will change significantly…and you won’t get ridiculous looking lists that claim Raul Ibanez is a comparable valued commodity to a part time all-fielding no-batting 4th outfielder (this has actually happened this off-season) or Eric Hinske has the same value as Adam Dunn. Both of those statements should immediately make you rethink the way you’re using fielding metrics…because they’re so far out of joint with traditional rational thinking about the game that they demand immediate attention.

No one in the game today would listen to you if you tried to claim that Raul Ibanez and a part time fielder were the same value or that Adam Dunn is the same as E5 Hinske. Not one man in the game would believe that. NONE. Dunn is clearly and identifiably a more valuable commodity even with -20 runs of defense in left field. Ibanez got 11 million per year while the 4th outfielders of the game get 1 mil if they’re lucky…FOR A REASON. And no…that reason is NOT that GMs are all a pack of morons who can’t use the new wave of sabermetric tools.