Partial results from 1 season with all 3 new roster sets

[rcbuss]

Originally Posted by rcbuss

Thanks for the formula, dougaiton. It looks to be as good as any other estimation of defensive range.

After fooling around with darkhorse's formula, I did some playing around with the numbers at the Baseball Prospectus website. By extrapolating from the Rate, FRAR, and FRAA for a number of extremely good defensive players, I was able to deduce the Rate of a replacement player at each defensive position (per BP). I'm assuming that the "88" used by darkhorse in the original formula was meant to represent the replacement level player at a given position. A potential problem with this is that BP appears to have different replacement levels for each position, which appear to be as follows:

• c: 75
• 1b: 90
• 2b: 80
• 3b: 87
• ss: 80
• cf: 85
• lf/rf: 90

For what it's worth, the replacement level used for defense for pitchers is roughly 93.5.

With this one change, I've taken the liberty of modifying the original formula; we now have:

Range rating (/100) = (Rate - positional replacement Rate) * 4

Another question one needs to ask is: What range of ranges would one like to have? Do you want the occasional shortstop with defensive prowess of historic proportions to be reflected by a range in excess of 100 (~120), or would you rather have 100 be the cap? The answer to that question would determine how you set the multiplier (in this case, 4).

If one used the all-time career high for Rate at each position (e.g., SS - Ozzie Smith, 111) as the basis for setting the range rating to 100/100, one could conceivably come up with a formula for SS range such as this:

SS Range rating (/100) = (Rate - SS replacement rate (80)) * 3.225

This would then allow for extreme performances, such as that of SS Orlando Cabrera for Montreal in 2001 (Rate = 120). His OOTP SS Range for that season would be 129, his only season over 100. Ozzie, on the other hand, would have eleven seasons with an OOTP range of 100+.

What follows is a far from all-inclusive list of best Rate at each position, season and career:

season

• C Ivan Rodriguez, 127
• 1B Keith Hernandez 115
• 2B Bill Mazeroski, 121
• 3B Brooks Robinson, 120
• SS Orlando Cabrera, 120
• CF Andruw Jones, 121
• LF/RF Dwight Evans, 116; Barry Bonds, 115

career

• C IRod, Johnny Bench, 112
• 1B Hernandez, 107
• 2B Mazeroski, 113
• 3B Rolen, 108
• SS Ozzie Smith, 111
• CF Andruw Jones, 111
• LF/RF Evans, Bonds, 104

I've made this a lot more complex than it has to be.

What I'd like to put out there is a modification of darkhorse's original formula:

Range (x/100) = (player positional Rate - 88) * 4

If we make the following assumptions:

• OOTP range scale of 1-100
• league average defensive range at a given position (i.e., Rate = 100, OOTP range = 50) is equidistant from replacement level defensive range (replacement Rate value being positionally dependent, OOTP range = 1) and the best possible defensive range (Rate = 100 + │100-replacement positional rate│, OOTP range = 100),
• replacement Rate varies by position, as previously noted

we get the following theoretical ranges for the Baseball Prospectus Rate statistic for each position:

C: 75-125
1B: 90-110
2B: 80-120
3B: 87-113
SS: 80-120
LF: 90-110
CF: 85-115
RF: 90-110

If we'd like to have a range rating (and, for catchers, an Arm rating) on a 100 point scale, simply divide 100 by the breadth of the range for each position to obtain the multiplier:

C: 2
1B: 5
2B: 2.5
3B: 3.85
SS: 2.5
LF: 5
CF: 3.33
RF: 5

The modified darkhorse formulas for each position would then be:

C: Arm (x/100) = (player positional Rate - 75) * 2
1B: Range (x/100) = (player positional Rate - 90) * 5
2B: Range (x/100) = (player positional Rate - 80) * 2.5
3B: Range (x/100) = (player positional Rate - 87) * 3.85
SS: Range (x/100) = (player positional Rate - 80) * 2.5
LF: Range (x/100) = (player positional Rate - 90) * 5
CF: Range (x/100) = (player positional Rate - 85) * 3.33
RF: Range (x/100) = (player positional Rate - 90) * 5

Using the all-time best Rate performances for each position, as noted in my previous post (season and career), we get the following OOTP range ratings for these performances:

season (Rate, OOTP range rating (Arm, for catchers))

• C Ivan Rodriguez (127, 104)
• 1B Keith Hernandez (115, 125)
• 2B Bill Mazeroski (121, 103)
• 3B Brooks Robinson (120, 127)
• SS Orlando Cabrera (120, 100)
• LF Barry Bonds (115, 125)
• CF Andruw Jones (121, 120)
• RF Dwight Evans (116, 130)

career (Rate, OOTP range rating (Arm, for catchers))

• C Ivan Rodriguez, Johnny Bench (112, 74)
• 1B Hernandez (107, 76)
• 2B Mazeroski (113, 83)
• 3B Rolen (108, 81)
• SS Ozzie Smith (111, 78)
• LF Bonds (104, 70)
• CF Andruw Jones (111, 87)
• RF Evans (104, 70)

Note how rare OOTP ranges of over 100 would be with this system. Because of the lower levels for replacement defense at the skill positions (C, 2B, SS, CF), and the correspondingly wider ranges for Rate, it is particularly difficult for players at these positions to get the equivalent of a range of "10" (91-100). For example, Ozzie Smith's range would've been a "9" in seven seasons, but a "10" in only two others.

On the flip side, Derek Jeter's range ratings would actually improve (using 80 as the replacement Rate for SS, instead of darkhorse's 88): to 30, 25, 28, 35, 23, 15, 15, 20, 5, and 53 (his Gold Glove year) over his 10 full seasons. Even with the use of 75 as the replacement Rate for catchers, Mike Piazza's 2003 Rate2 of 71 would translate to an Arm rating of -8.

[toadfreak7]

Originally Posted by jrdoa

I've run 1 season of each new roster set. I'm trying to figure out which roster I would continue to use to get the most accurate stats possible. After running these, I'm still not satisfied which to use. Very interesting how Halofan and Toadfreak had the same results in the WS. Any, here are some of the results:

Cubbyfan's rosters:
Best team -- StL 103-59
World Series -- Texas beats Florida 4-1

Interesting Race -- Boston beats Toronto by 1 game to win AL East

Strange, or not so strange, stats:
AL had 9 relievers with 100+ appearances; NL had no one over 100 and had 10 over 73.
A-Rod leads league with 67 HR; next closest was Thome, Ramirez, & Ortiz (50)
Cy Young winner -- Ryan Drese with 21-7, 4.38 ERA, and 220k's in 242 innings

----------------------------------------------------------------------------

Halofan rosters:
Best team -- New York Yankees 102-60
World Series -- Minnesota sweeps StL

Interesting Race -- Tampa takes 2nd in AL East with 77-85 record. BoSox finish 3rd at 77-85. Detroit wins Central Division and WSox finish last at 71-91.

Stats:
Suzuki wins batting title easily by .029. Also wins SB (89) by 33. Only 2 players had 40 or more home runs (Sosa and A. Rod). 9 relievers in the NL had over 92 appearances while only 1 in the AL had 90.

--------------------------------------------------------------------------

Toadfreak Rosters:
Best Team -- LA Angels 99-63
World Series -- Minnesota beats StL 4-1

Nothing exciting with races

Stats:
10 players (AL) over .320 Batting Ave.; only 5 in NL
3 players over 40 HR in AL; 5 in NL
1 reliever over 90 appearances in AL while NL has 10 over 90 innings and 3 over 100 inn.

Rookie of Year: Jorge "Steroids" Piedra with .305, 22 HR, and 81 RBI

Just out of curiosity, were you using my Full Minors rosters? Or the old ones?

Yeah, St. Louis tears it up every year.

[GForce]

Originally Posted by dougaiton

It's really, really simple - a quick and dirty estimate. Well, I hope darkhorse (the true hero of historical simming!) won't mind me dishing it out, but IIRC it's as simple as subtracting 88 from BP's rate stats and multiplying the remainder by 4.

For example, Andruw Jones has a career rate of 110, so -88, making 22. Multipy 22 by 4 and you get 88. Bernie Williams has a rate of 98, which would be a 40, but then he hasn't broken 92 (or 16) for 3 seasons, so I'd take it down a few points.

Obviously, you need to weight it in your own opinion. Jeter had an excellent year last year, worth something like 56 by this method, but he's only breached 25 once before.

It's still very subjective, and it does result in far fewer 10s (and far more bad defensive players) than subjective reasoning. I like that though, and I built my file around recreating 'proper' numbers with that in mind. Then again, I preferred to have a logical base to work from and adjust for there. I didn't really have the capacity to think about whether Ken Boyer was a better fielder than George Scott: for OOTP, the problem was not just judging who was better, but by 'how much' they were better.

I may have just found the defensive metric for my Franchise Stars set.

Thanks for the info.

GH