OF Arms

So based on a recent thread exploring catchers' arms, I decided it was time to revisit and challenge some of my constant values for arm strength I use when pricing. The following are some of the key historical assumptions that I was making when valuing arm strength:

1) CF is the most important position for OF arms since that is where plays default if the card does not distinguish a fielder (i.e. CF'ers get more chances to use their arm than LF and RF).
2) Similar to catchers I would expect that the primary impact of a good arm is that it prevents runners from even attempting to take an extra base (in addition to reducing the safe chances).

After a recent examination of a couple of leagues I believe some of those assumptions to be incorrect. I now believe the following to be true:

1) RF is the most important position for the arm (or possibly equal to CF in importance). What I found in my research was that players such as Suzuki routinely get 20 assists a season, which on the surface does not make sense, because I would expect very few players attempting an extra base on an arm this strong. In fact if you look at equivalent CF seasons such as Mays, you will find that they are averaging close to 10 assists per season and it isn't until you get to the -3 arms in center that you see assist numbers go up because runners actually are attempting to take extra bases on the CF arm. So why doesn't RF follow this same mechanic?

I believe it to be the option for runners to take 3B on singles. If the hit is to RF the RF'er's throwing rating is with a +2 penalty. So now with Suzuki's arm now at -3, the CF corollary applies and that runners will attempt the extra base at this arm strength, which allows Suzuki to produce assist numbers equivalent to CF'er's with -3 arms.

So while my original hypothesis was that CF'er get more arm throwing chances, I now believe that due to the 1st to 3rd running option RF'ers get the same number of throwing chances as CF'ers and becuase of the +2 penalty runners will go on them much more than strong CF arms.

The flip of this is also true for LF'ers. They see the fewest throwing opportunities because they get a -2 on that 1st to 3rd runner. So its not until we get down to -1 arms in LF that we see larger assist numbers, and therefore you can get away with a much worse arm in LF.

I have always valued RF/LF arms equivalent, but I think I am going to make CF/RF arms equal with a drop off for the same arm in LF.

Have I reached an incorrect conclusion based on limited data? Give me your thoughts.

I've had similar thoughts/questions and have investigated a few things - here are some comments.

Not all undesignated hits go to the cf. Singles and doubles with asterisks - which are subject to the more baserunning decisions max rule - are distributed to all three fielders, I'm not sure of the distribution %'s.

I've looked at pbp data from ATG and 20xx leagues and get the following distributions (lf/cf/rf) for all singles, doubles, and flys

singles 30%/40%/30%
doubles 11%/78%/11%
flys 31%/38%/31%

Similar to my catcher arm simulations with the pc game, I did simulations holding everything else constant but changing outfielder arm strength. I got cf and rf arm strength as being equally important, while lf arm strength was much less important. I suspect for the reasons you suggest.

As always, I start with Pelletier. https://forum-365.strat-o-matic.com/com ... 7&t=639204

Arm value is just the inverse of speed value. Multiplied by 9 across the lineup and divided amongst the three outfielders.

The division I use is 20% LF / 45% CF / 35% RF. These figures were derived from a linear regression run by my Uncle Ross using sim data from the Windows game. He used to teach math at the Air Force Academy so I'm confident in his findings. Although it sounds like barrmorris did something similar and got different results?

Another good arm value thread: https://forum-365.strat-o-matic.com/com ... 6&t=641226

Thanks for the post.

I've been reading this thinking of what to do. My NERP valuing system doesn't have much insight for OF arms and I'm uneasy about redoing the ATG9 set etc (or at least unprepeared) which my prices are a lil based on. Still coming though....

My OF arm value has been 0.25 NERP for each arm into negatives (-1, -5 etc). So Willie Mays' 1(-5) is 1.25 NERP better than a 1(0) arm like Averill. I'm open to having it be relative to position and I'll just do some more manual entry for guys with multi-position flexibility (price for best price----I don't hard-code in value for multi-position except for bench players). Plus open to non-linear, but for my needs really want a "ballpark" estimate for a -1 vs a 0. I'm using 0.25 NERP.....

If say CF was 1.0 or 100%, what would the weights be on for LF and RF? Equally critical, how many NERP does that being 1 arm better than "0" earn? At each position?
-not linear
-MaxPower proposes 20% LF/45% CF/35% RF split for *all value derived for throwing "unit"* (sac flies, 2Bs, SIs, etc)
-MarcPelletier mentions "If the three outfielders have an arm rating of -1, then collectively, the value of this asset can simply be computed with the same unit (0.23) X 9 (to go across the lineup) = 2.07" --- care to share anything about the non-linear "curve"? When you say "if the three outfielders have an arm rating of -1, then collectively...." Do you mean they have a TOTAL of -1, or each -1 so -3 between the three of them?
-Considering some systems gave (significant) value to base-stealers that are Es, Ds and Cs, my

baseline CF give at say -1 arm? Who's to say (0) arm is worth "0" NERP?----this is always relevant.

NERP for LF - CF - RF
(-5)
(-4)
(-3)
(-2)
(-1)
(0)
(+1)
(+2)
(+3)
(+4)
(+5)

?

I consider 1-12 runners "0" NERP, with each rating below that getting that -0.25 NERP again, and each above +0.25 NERP.
1-12 = 0 NERP
1-17 = 1.25 NERP
1-8 = -1.0 NERP

That classic AA *3-6/- (19-15) great base-stealar like '12 Speaker is worth +2.68 NERP. With his speed that and extra +1.25, so 3.83 on raw-running. ****Much lower than the other models I see linked by MaxPower, but that's that "0" NERP unknown middle again. I think a D stealer at 1-11 running shouldn't be "worth" really any extra NERP.... Considering those baserunning weights, is 0.25 NERP per arms ~reasonable? Love the discussion! Thanks

I consider 1-12 runners "0" NERP, with each rating below that getting that -0.25 NERP again, and each above +0.25 NERP.

Pelletier has zero NERP at 10.6 speed rating, with the increments worth 0.23. I understand what you're saying about "is +0 arm really worth zero" and the answer is yes and no. Yes it is worth zero NERP because it is not affecting the runner's speed rating in either direction. But at some later stage in the valuation process you're likely going to be converting runs produced into runs above average, at which point +0 arm may no longer be zero, depending on where you set the average.

NERP Value per 1 throwing arm proposed above?
LF = 0.414
CF = 0.9315
RF = 0.7245

I was doing 0.25 per for all positions. The flat rate for all positions should get changed and this is supa great. I messaged you MaxPower about some of the diminishing returns childsmwc writes about, which definitely upped my loose-ends or overcomplicating. For CFers, my 0.25 was only giving 26.8% the credit this system does----that's jarring and even so much more when you consider a great arm like Willie Mays vs humble (0) or even somebody like Richie Ashburn.

What would a sliding scale look like? When would it become relevant? childsmwc was pretty convincing about the elite arms NOT being run on and rburgh gives an example about a (-6) vs a (-5) etc. childsmwc notes the Mays getting less assists IS still good though, but I was assuming there would be some diminishing return the further you get from that "0" arm?

Here is how I break down outfield arms (the base chart):

Arm Opps Go% Safe% XB Outs Total
-6 90 10% 55.0% 5 4 9
-5 90 20% 57.5% 10 8 18
-4 90 30% 60.0% 16 11 27
-3 90 40% 62.5% 23 13 36
-2 90 50% 65.0% 29 16 45
-1 90 60% 67.5% 36 18 54
0 90 70% 70.0% 44 19 63

Etc.

Here are the assumptions imbedded in the above chart:

1) Opps = the number of opportunities that will call the OF arm into play during the season (this chart assumes CF at 90 chances)

2) Go% is based on the number of those 90 runners that will actually attempt to take the extra base on the specific arm. Further in this example the assumptions are that runners with a 55% chance will attempt the extra base and assumes an even distribution of runners from 17 to 8. At a -6 arm only the 17 runners meet that criteria (i.e. 1/10 of total runners).

3)Safe%- based on the run ratings that will attempt the extra base what is their average safe chance, using the even distribution of runners discussed in #2

4) XB, Out, and Total are all then a function of the inputs above.

This basic chart does not incorporate the Sac Fly chances and two out chances where the runner gets a +2, and since this is for CF we don't have to incorporate adjustments for 1st to 3rd.

To incorporate the +2 opportunities simply identify the ratio (assume 20% for this example) and then take the expected values of the chart above using 80% of the -6 arm and 20% of the -4 arm. Repeat for each arm.

Once you have built this table then assign whatever runs values you wish to the XB and outs. What I don't have is the total opportunities, my experience puts that in the 85 to 95 range for CF'ers to align with the assist totals at various arm strengths etc. but this isn't exact

The key to all of the above is that it is based on assumptions that drive the results which then convert into a NERP or Runs value. Modify accordingly to your preferences or expectations.

The division I use is 20% LF / 45% CF / 35% RF. These figures were derived from a linear regression run by my Uncle Ross using sim data from the Windows game. He used to teach math at the Air Force Academy so I'm confident in his findings. Although it sounds like barrmorris did something similar and got different results?

A while back I noticed that Don Padgett's card has all of his singles, doubles and flyballs hit to centerfield against right handed pitching.
Does anyone know if Strat gives a different distribution for pull hitters, straightaway hitters and spray hitters?
Maybe you should put the better arm in RF against a lefty lineup and LF against a righty lineup.

Here is how I break down outfield arms (the base chart):

Arm Opps Go% Safe% XB Outs Total
-6 90 10% 55.0% 5 4 9
-5 90 20% 57.5% 10 8 18
-4 90 30% 60.0% 16 11 27
-3 90 40% 62.5% 23 13 36
-2 90 50% 65.0% 29 16 45
-1 90 60% 67.5% 36 18 54
0 90 70% 70.0% 44 19 63

Amazing stuff childsmwc! Thank you! The Safe % is off, though?
Arm Opps Go% Safe% XB Outs Total
-6 90 10% 55.56% 5 4 9
-5 90 20% 55.56% 10 8 18
-4 90 30% 59.26% 16 11 27
-3 90 40% 63.89% 23 13 36
-2 90 50% 64.44% 29 16 45
-1 90 60% 66.67% 36 18 54
0 90 70% 69.84% 44 19 63

I've never translated expected runs to NERP and have followed MaxPower/Pelletier's outliine a bit more.

Assuming consistent value changes between each arm increment, do you think valuing one arm value at: LF = 0.414, CF = 0.9315, RF = 0.7245 is about right? Too high?

That seems to read too high in my db, making the great throwers a bit too consistently the best values in the deck.

NERP Value per 1 throwing arm proposed above?
LF = 0.414
CF = 0.9315
RF = 0.7245

This is per season NERP when I've been meaning to talk about per per 216 PAs, like on DiamondDope. Pelletier proposes 688 PAs for the season, whereas Berce's model was built off 648 PAs. That'll affect the numbers a bit. Here are the proposed arms NERP scores based on 688 or 648 PAs.

Assuming 688 PAs for Whole Season
Season NERP per arm NERP per 216 PA (DiamondDope)
LF 0.4140 (20% of 2.07) -----> x 0.313953488372093 = 0.1299767441860465
CF 0.9315 (45%) ---------------> x 0.313953488372093 = 0.2924476744186047
RF 0.7245 (35%) ---------------> x 0.313953488372093 = 0.2274593023255814

Assuming 648 PAs for Whole Season
Season NERP per arm NERP per 216 PA (DiamondDope)
LF 0.4140 (20% of 2.07) -----> x 0.333333333333333 = 0.138
CF 0.9315 (45%) ---------------> x 0.333333333333333 = 0.3105
RF 0.7245 (35%) ---------------> x 0.33333333333333 = 0.2415