Defensive Impact

[Tyrus4189Cobb]

Gather 'round children, it's time for another concoction from the wonderful whizzical lab of Tyrus. This time I'm tackling defense. It isn't my strongest ever, so I'm looking for feedback (of the nonabrasive variety). I don't feel I have anything to lose as a novice of the site and an even more novice of statistics. This one is called Defensive Impact. Here goes:

Defensive Impact seeks to tell you how many plays a player contributed above/below league average. It doesn't try to assign what these plays would be worth in runs saved or outs made. It is strictly plays made (though you could say outs because virtually all basic fielding statistics are recorded with respect to the out).

I used basic statistics (all the sabernuts just clicked out of this thread). I'm not a genius and I don't have the resources for things like UZR. I have less of a problem with traditional fielding stats than others on this site. If a guy makes a play, his FP is going to go increase. If he has poor range, then he won't get to the ball and his FP doesn't benefit. Errors are subjectively issued, but not to the point where guys are being ridiculously charged without reason. 98% of errors scored by scorers are quite indisputable. If a guy makes a defensive misplay, he may not get the error, but he doesn't benefit from making the play, which is what this statistic deals with. Furthermore, only infielders (1B, 2B, 3B, SS) can be evaluated by this metric. I'll see how it goes before designing adjustments for the others. For the sake of accuracy, I go to two decimal places.

Here is Dave Concepcion's basic 1977 fielding stats along with he league and two fictional players. Team chances are those strictly of the infield (1B, 2B, 3B, SS).

1977	CH	PO	A	E	FP
Concepcion	781	280	490	11	0.9859
Smith	850	300	518	32	0.9624
Jones	740	260	475	5	0.9932
team IF	3563	2177	1332	46	0.985
league SS	10150	3386	6384	380	0.9626
league IF	46041	27491	17497	1053	0.977

Let's first make some observations: Concepcion fielded .2192 (21.92%) of all the Reds' infield plays with an efficiency of .9859 (98.59%). Thus, he efficiently fielded .2161 (21.61%) of all the Reds' infield plays (this would include the infielders' ventures to shallow outfield, but it doesn't matter because it's all being done by infielders). Concepcion fielded .0031 (.31%) of plays below perfection (he fields .2161 our of a possible .2192). 0.31% of all IF plays comes out to 11.05 plays (.0031*3563).

Concecpion fielded .2912 of all plays, a number I'll call IC (Infield Chance Portion). IC simply equals chances divided by IF chances. In this case, 781/3563=.2192.

Doing the same for the entire league, we find that the league SS fielded .2205 of all league IF plays with an efficiency of .9626. Thus, the league SS efficiently fielded .2123 of all plays. This comes to a difference of .0082 (.2205-.2123), meaning the league SS fielded 377.54 plays below perfection (though this is irrelevant).

Let's say the league SS was embodied in a single player in the Reds' infield. He fields 0.82% of all plays below perfection. Put into the context of the Reds, 0.82% below perfection equals 29.22 plays.

Concepcion, remember, only fielded 11.05 plays below perfection meaning he fielded 18.17 (29.22- 11.05) plays more than a league average guy in his shoes in 1977.

The formula for Defensive Impact is thus:

DI=league impact-player impact
Impact=[IC-(FP*IC)]*IFCH
where IFCH s total infield chances and IC is infield chance portion (chances/IFCH).
-----------------------
Let's swing back to Smith and Jones. I invented the two as replacements for Concepcion had they played SS for the 1977 Reds in his place. Smith is a guy who makes a lot of plays yet a lot of mistakes. Jones is a guy who plays more conservatively with much more prowess. If in some tiny chance you are still on-baod with this presentation get ready for the turn-off. Here are their numbers again, DI and IC also calculated:

		CH	PO	A	E	FP	IC	Imp	DI
Concep		781	280	490	11	0.9859	0.2192	11.05	18.17
Smith		850	300	518	32	0.9624	0.2296	33.85	-4.63
Jones		740	260	475	5	0.9932	0.2077	4.99	24.23
team IF		3563	2177	1332	46	0.985	-	-	-
league SS	10150	3386	6384	380	0.9626	0.2205	-	0	
league IF	46041	27491	17497	1053	0.977	-	-	-

Smith makes a lot of plays, but his errors hurt him. That's good, I don't want a guy to have a good number if he keeps making mistakes. Jones, on the other hand, has a DI of 24.23. Obviously this metric heavily favors efficiency.

That's it. I appreciate any help or thoughts (try to keep them kinda nice, eh?)

[leewileyfan]

If he has poor range, then he won't get to the ball and his FP doesn't benefit.

I just picked up on this point. A player with limited range, compared to a rangy fielder, does benefit in fielding percentage, sometimes compounded, because fielding percentage does not penalize him for assists [or PO] not executed. Nor is he cahrged with an error on balls he cannot get to. The rangier fielder, who may get to 30 more balls in play, may also make a few errors; and in that regard, fielding percentage is unforgiving.

As for the specifics, Concepcion/1977, my metric has him about 12 defense runs better than average 1977 SS, which is then diluted a bit to account for his playing time relative to team total innings at SS. If he played around 125 games, that would bring his runs to + 9.25 DR > AVG. You have 18+ plays; and my metric at SS converts raw data to about a .374 ratio to plays. 18+ plays @ .374 = +6.73 runs. There's no real gap in conclusions; but the method of getting there is quite different.

I like to differentiate chances for fielders, with distinct weightings for assists as opposed to PO [and, beyond that, position-specific].

[Tyrus4189Cobb]

I just picked up on this point. A player with limited range, compared to a rangy fielder, does benefit in fielding percentage, sometimes compounded, because fielding percentage does not penalize him for assists [or PO] not executed. Nor is he cahrged with an error on balls he cannot get to. The rangier fielder, who may get to 30 more balls in play, may also make a few errors; and in that regard, fielding percentage is unforgiving.

As for the specifics, Concepcion/1977, my metric has him about 12 defense runs better than average 1977 SS, which is then diluted a bit to account for his playing time relative to team total innings at SS. If he played around 125 games, that would bring his runs to + 9.25 DR > AVG. You have 18+ plays; and my metric at SS converts raw data to about a .374 ratio to plays. 18+ plays @ .374 = +6.73 runs. There's no real gap in conclusions; but the method of getting there is quite different.

I like to differentiate chances for fielders, with distinct weightings for assists as opposed to PO [and, beyond that, position-specific].

All greats points as usual, Lee. As for differentiating chances, I did not because my metric deals strictly with plays made. A putout or an assist is still a play. Had I gone with outs, I probably would've assigned a weight for each (a putout is always one whole out and an assist is a specific portion of an out).