Defense Metric: MLB Defensive History [1901-2012]

[leewileyfan]

I've resurrected a book manuscript of mine, starting with 1901 and attempting to plot a history of MLB defense and how it has evolved over 113 seasons, and growing.

I have described the formulation and approaches in other threads and responded to questions about particular players and their comps. I spent a good part of 2012 editing and updating each season and am still contemplating publication of the finished work, hoping to find a publisher - rather than going the self publish route.

A portion of the introductory material has to do with equivalence and how I address it. Equivalence came up in 2005 in a discussion on a since gone baseball forum where I got into a discussion of all defense metrics, their pluses and minuses. One poster quizzed me at length about seasonal strikeout differentials among pitching staffs and how I addressed the discepancy of opportunity for fielders.

If I go ahead with publication efforts, I'll be seeking a graphics specialist for page formatting and data presentation formats; but here's how one draft looks [in draft] to show positional data for 1901 as it relates to K discrepancies between teams and League averages. This would be an offshoot to the 1901 page itself, which shows a sampling of players at each position, always including the top ones.

Equivalence Applied to Ratings, Defense Runs +/- Average.

Pos. .........Player...Team......Rating......DR......K Diff:LG.......Adj. DR

C........Bergen, Bill......CIN........ .940 ....+24.6.... +50............. +25.4
1B.......Isbell, Frank.....CSX....... .930.....+11.4.... +52............. +11.8
2B.......Williams, J. .....BLA....... .935......+12.2.... -71.............. +11.1
3B.......Collins, Jim......BSX....... .963......+17.3....+54.............. +18.0
SS.......Wallace, Bob....STL........ .974......+27.7... -85...............+26.0
LF........Delehanty, Ed..PHI......... .924......+17.8...-50............... +17.0
CF........Hamilton,Bi. ...BSN........ .907.......+11.6..+28..............+13.03
RF........Flick, Elmer.....PHI......... .949......+26.8.. -50...............+26.0

This is an example only of a page to introduce equivalence and how it would affect rating and defense runs at each position, according to team K's and position played.

The 1901 season listing of defenders would have been presented first, with unadjusted ratings and DR. Of course, the metric itself is fully explained before the posting of season data.

Basic question is: Is this kind of information what you would look for in explaining how a metric treats of game irregularities [like K's]?

Feedback is welcome as are questions. The actual layout starts for me at he beginning of the New Year [Happy to all].

[leewileyfan]

There's been some passing eyeball interest here but no responses; so I figure it would be wise to provide some context to the initial post and why I wrote it.

If I proceed with trying to get this published, I have to address issues from a perspective I believe in; and I have to make my best attempt at presenting and defending it.

My basic mental image in trying to make a mathematically reasonable evaluation of a K is nine inning grid divided into two tiers: the run production grid as the upper tier and a perfectly "pure" conunterbalancing set of "anti-runs" cells on the lower tier.

In a 4.5 run climate, roughly the average, the upper tier model would get .5 in each of its nine run procuction cells to equal 4.5 runs.

A "pure" counterbalancing "out" element would be comprised of individual outs [no double plays]; no outs that might possibly have ripple effects such as base runner advances or runs scoring. The K, being the irregularly distributed out type among teams in any season is seen as that "perfect out." Three of them = .50 for each inning [.50/3 = .1666, or .167]; burt the subject being addressed is equivalence as it applies to defensive opportunity +/- because of K's and their unequal distributions.

I have had discussions with one math-knowledgeable person who says that .15 is an appropriate value per K. My .167 seems a mite high; but getting back to equivalence as the operative topic, I am convinced that, to be truly equivalent, we must also address defensive efficiency ... the relative fielding execution of batted balls in play. I propose that .700 is a fair standard, which would mean that .167 per K takes care only of the raw K differential among pitching staffs resulting in disparate BBIP.

I am using .167*.700 = .1166 = .117 for an equivalent K value impacting fielder positional DR.

The sample page that started this thread was for the purpose of getting feedback on:

1. The importance of such a page at all;

2. The general topic of equivalence;

3. The specific evaluation of a K for the purpose of proper application of equivalence at .117.

Questions, comments, counter positions/values are invited and welcome.

[leewileyfan]

To illustrate how the equivalence subject looks, when applied to my metric of fielder evaluation, the sample below is from a great year for me [as a little kid and blossoming fan ... into full avid BOSOX status], 1941 ... also great because we were not yet involved in WW II and so many greats were still playing.

2B:

Frey, Lonny: CIN; [G]: 144; Rating .941; DR + 3.7; DR Adj. Playing Time +3.5; K's +76; DR K Adj. +4.7

Stringer, Lou: CHI; [G] 135; Rating .957; DR +9.3; DR Adj. Playing Time +8.2; K's -3; DR K Adj. +8.2

Crespi, Creepy: STL; [G]: 145; Rating .933; DR +1.0; DR Adj. Playing Time +1.0; K's +108; DR K Adj. + 2.8

Gordon, Joe: NYY; [G]: 130; Rating .953; DR +8.1; DR Adj. Playing Time +6.8; K's +21; DR K Adj. +7.1

Doerr, Bobby: BSX; [G]: 130; Rating .941; DR +3.8; DR Adj. Playing Time +3.2; K's +36; DR K Adj. +3.8

Bloodworth, Jimmy: WAS; [G]: 131; Rating .974; DR +15.3; K's -9; DR K Adj. +15.2

McCoy, Benny: PHA; [G]: 134; Rating .947; DR +5.8; K's -167; DR K Adj. + 3.1

NOTE: When a position player's pitching staff fans a greater number of batters than League average [ie 1,000 vs LG 850] the difference will be positive [1000-850] = 150; and the equivalence value of the K will be added to calculate net DR. Conversely, a staff with fewer K's will have a negative sum [ie 850-1000] = -150 and the DR will be subtracted from the unadjusted DR.

Benny McCoy's PHA staff recorded 386 K's, while the AL staff average was 553.