Looks pretty strange, doesn't it. The top team in DP% is mine, but the #2 team has a D+ guy at 2B and a C+ guy at SS. So clearly I'm not capturing all of the information that we need or my DP% numbers are way off. But here's some quick correlation data:

correlation of DP turned and MI Rg				0.357933
correlation of DP turned and IF Rg				0.08922
correlation of DP turned and 2B Rg				0.37704
correlation of DP turned and SS Rg				0.231536
correlation of DP turned and 3B Rg				0.06136
correlation of DP turned and 1B Rg				-0.39596

So, your 2B range clearly does have an impact... I'd say almost a 38% correlation is significant, even if it's not that strong. SS range is also significant though not as strong. 3B range is basically unrelated. What I don't get is 1B range. There is a strong negative correlation between DPs turned and 1B range. Why is that? Maybe it has to do with fielder's choices - you get a lot of 3-6 and 3-4 plays that a poor range team would not have. But I am really not sure.

Clearly if you've read Win Shares then you remember that the biggest impact on fielding in general and DPs specifically is the % of innings thrown by LH pitchers. I am going to need to figure that out at some point. Unfortunately it's not available in the team stats so I'll have to figure it by hand for each team. (Incidentally, the #2 team above - Wasted Players, Wasted ABs has only one LHP on their team)

Another factor is the park factors. I'm estimating doubles/triples and using that to derive runners on 1st. If a park has a big + for 2B/3B then the numbers will be skewed. Ideally I could find some way to account for that as well. Maybe I'll just plug it in and find the correlation to see if it means anything.

What else could be skewing the numbers? I'm not sure. This is all a house of cards anyway, using estimates to derive estimates and then trying to base conclusions on weak correlations. But it's a start.

We do also have K/9 and G/F data for each teams. I could try to modify the expected DP% based on those.

Speaking of G/F data, I'd love to figure out if there's any way to actually build ground ball or fly ball pitching staffs, has anyone ever looked into that?  My compadre toadee gathered some data awhile ago that indicated that G/F ratios were completely random, which would be pretty disappointing.  It would be nice to be able to build specific staffs and then tailor your defense and park to them.

Posted by alleyviper on 7/21/2010 8:35:00 PM (view original):
Speaking of G/F data, I'd love to figure out if there's any way to actually build ground ball or fly ball pitching staffs, has anyone ever looked into that?  My compadre toadee gathered some data awhile ago that indicated that G/F ratios were completely random, which would be pretty disappointing.  It would be nice to be able to build specific staffs and then tailor your defense and park to them.

I remember reading somewhere that the determinant of a groundball or flyball is actually the batter. So you can't really build a ground-ball pitching staff and then put a great infield defense (and lousy outfield defense) around them.

jfranco77....thanks for all the work you have put into this...great stuff.

I didn't forget about this... the season ended and my team finished 97-65, won our division and got knocked out in the first round of the playoffs. My team DID finish 1st in the 'percentage of DPs turned' category. So this was an interesting experiment but definitely didn't produce the results I expected. I'll post some data next week. I might also go back and triple-check my data... I double-checked everything and it all looks right, but it feels wrong.

NOTE - data re-entered on 8/9... found some errors.

Team	2B Rg	SS Rg	3B Rg	1B Rg	MI Rg	IF Rg	DP Turned %
Who Needs Greenies?????	1	2	4	10	3	17	7.1%
Wasted Players, Wasted At Bats	3	5	6	3	8	17	9.4%
The Seven Year Itch	5	8	11	5	13	29	8.1%
The 700 Club	4	8	3	4	12	19	8.5%
Se7en Gluttony	1	3	1	4	4	9	9.8%
se7en	5	9	12	10	14	36	6.6%
Rumble Strip 22	11	8	1	5	19	25	7.7%
Power Pack #7	7	8	11	11	15	37	7.3%
My Bike Likes Ike!	6	11	7	12	17	36	7.4%
MXC Prime Minister of Exclamation Points	3	8	11	12	11	34	8.6%
MayB1Day	10	5	12	6	15	33	9.6%
Managers Afternoon Delights	12	12	12	12	24	48	10.5%
Lucky Seven Dilligafs	5	10	7	7	15	29	8.3%
Lucky and Good	6	5	12	12	11	35	8.7%
Lucky 7s	3	1	5	12	4	21	7.0%
Kids in the Corner	5	5	7	11	10	28	8.0%
Hoochie Coochie Men	6	4	12	4	10	26	9.5%
Hex in the Citi	2	7	5	7	9	21	7.3%
Gobble Up The Hobbled	10	8	2	2	18	22	8.8%
Elsa is in the Lou (Gehrig)	9	1	2	12	10	24	7.9%
Do the Roar	10	12	2	2	22	26	8.9%
derschwanz7BLOKKEN	11	10	7	7	21	35	9.7%
7 Pitch Count	2	6	3	4	8	15	7.6%
7 & 7	5	4	11	12	9	32	7.3%

Posted by jfranco77 on 8/6/2010 3:31:00 PM (view original):
I didn't forget about this... the season ended and my team finished 97-65, won our division and got knocked out in the first round of the playoffs. My team didn't even finish 1st in the 'percentage of DPs turned' category and all the correlations went down at the end of the season. So this was an interesting experiment but definitely didn't produce the results I expected. I'll post some data next week. I might also go back and triple-check my data... I double-checked everything and it all looks right, but it feels wrong.

correlation of DP turned and MI Rg				0.256582
correlation of DP turned and IF Rg				0.038481
correlation of DP turned and 2B Rg				0.309963
correlation of DP turned and SS Rg				0.124468
correlation of DP turned and 3B Rg				0.090367
correlation of DP turned and 1B Rg				-0.39464

Those #'s look right to me, or right enough.    I don't get why the high range of a 1b has a negative correlation to turning DP but everything else looks kosher.

yeah, the negative correlation at 1B has got me completely stumped. i'm re-doing a lot of the formulas in the spreadsheet to see if i missed something up. but even when i find other errors, it doesn't go away. here's the revised/corrected version. it appears to make a bit more sense.

correlation of DP turned and MI Rg				0.354447
correlation of DP turned and IF Rg				0.119763
correlation of DP turned and 2B Rg				0.396689
correlation of DP turned and SS Rg				0.204888
correlation of DP turned and 3B Rg				0.111978
correlation of DP turned and 1B Rg				-0.37407

OK, here's some total linear weights info. My 7 hitters made 174 + plays over the season (plus 4 from my pitchers). The total runs saved was 121.7 (the average was 0.68 runs per play).

My team actually gave up 592 runs. So without the defense we would have allowed 714 runs, for a savings of about 17 percent. That's assuming that the LW numbers would have come true, in some cases we would have allowed more and in some we would have allowed less.

Here are the individual player totals:

CF Ashburn: 35 + plays, 14.3 runs saved, average of 0.41 per + play. 1 error, 12 hits and 22 outs.
RF Bernie: 27 + plays, 17.0 runs saved, average of 0.63 per + play. 1 error, 8 hits and 18 outs.
LF Douthit: 13 + plays, 6.0 runs saved, average of 0.46 per + play. 1 error, 4 hits and 8 outs.

2B McPhee: 29 + plays, 23.5 runs saved, average of 0.81 per + play. 27 outs and 2 DPs.
1B Anson: 27 + plays, 22.3 runs saved, average of 0.83 per + play. 26 outs and 1 DP.
SS Long: 24 + plays, 18.8 runs saved, average of 0.78 per + play. 3 errors, 19 outs and 2 DPs.
3B Latham: 19 + plays, 17.9 runs saved, average of 0.94 per + play. 18 outs and 1 DP.

Interesting results jfranco, also thought it was interesting that you started this thread right after I posted my Flashing Leather thread, I have a feeling my idea inspired you to try one of your own. Which is cool, and I like the way you analyzed at the datas.
I haven't updated my thread but I will post the results, some surpring and unexpected results.
And guess how many wins that team got me... 97... yep, same as yours, thought I did rest my ace pitchers and some tired fielders for the last 3 games which I lost.
Waiting for the WS which is starting this PM...
Now I'll post some raw numbers on the thread I had started, but no analysis though, I leave that to experts like you and JohnGPF, God bless...

Franco, I wish I had the time to offer you more help on this experiment, but if you are going to enter a league ( or start one on your own, or enter one of mine) and would like me to draft a team to your specs, I will be happy to do so. I have had a couple of instances in which I had underperforming players that I anticipated more O from, waived them for D specialists, and the team improved. Even though I had no data, I did have a respect for RRF in the middle infield.

Thanks mr. fattkatt... even though I have data and it seems to be saying that range might not be worth it, I'm not convinced. Maybe "not worth it" is the wrong description but it just seems like the runs these guys saved on defense could have been purchased for less money on offense. McPhee had a .771 OPS# and 720 PA/162 for 5.76mil. Let's say that his fielding was worth 50 points of OPS# (29 hits over 700 AB would be about 0.040). I could have had Charlie Gehringer '28 for 5.72mil, and his OPS# is 0.831 for 727 PA. His fielding is C-/B (McPhee was C-/A+). That would still put Gehringer ahead (.831 vs .821 (.771+.050)) and it's not like B range is total crap.

But still. There's something else that makes teams with really good range play better. Maybe it's the DPs. You can get lucky and get a top-5 team in DP% with low range or you can lock it down with a great range team.

In my (not as humble as it ought to be) opinion, you are in the seminal stages of something that could be extremely valuable. Common sense tells us that you are correct, a little less D for more O would have been beneficial, but where is the line of demarcation? We would need a few teams over the course of a few seasons to figure it all out, but wouldn't it be wonderful to know? I have had good success with pre 1920 players with good D ratings, guys like 1917 Buck Weaver. I play him at 3b, he generally doesn't generate an abnormal number of plus plays, but all of my teams with him at 3b have done well. At any rate, if you need a volunteer to give you a "control" team, I am available.

The issue with a "control" team is that I think (I might be wrong) the data is more accurate and valuable when all the teams in the league play the same 4 infielders almost every day. So it's really more of a control league. Which I just happened to create accidentally when I set up the Managers Dilemma/Delight league.

I'll probably run the data for the Favorite Player league when the current season is done. That should be interesting too since the players will be similar, maybe even identical in some cases.

If you are itching for something to test, I'd put together a team of insane range pitchers and mediocre range infielders. Maybe the pitchers' range really is underpriced. I'd be curious to see the average LW runs saved by pitchers but for it to be significant you'd probably need a good 20-30 + plays in a season (in total).

I realize this is an old thread, but it's utterly fascinating.  In skunk's lastest "Choose your own theme" league, my theme is all A+ range players at every position.  I actually drafted A+++ players at 1B, 2B, SS, 3B, and CF.  Additionally, not only are all my pitchers A+, but several are at what used to be the highest range value (3.70), and I also have three of the crazy 2010 pitchers with significantly higher ranges.  I'll post updates here if anyone is interested, although I'm afraid I won't be able to crunch numbers the way jfranco can.