by Mean Dean » Sun Sep 23, 2007 9:30 am
First off, it's literally impossible that there is a "cumulative"/"more than the sum of its parts" effect to having more than one good OF arm. This is a dice game, with specific rules, where only one OF at a time makes a throw, and the success of that throw, by the rules, has nothing to do with who the other two OF are. You can't talk about intangibles in Strat-O-Matic; that doesn't make any sense.
I highly doubt that the ATG games and 200x games are completely different creatures, but arms certainly could be more of a factor in ATG; I don't know or care. I can't even rule out the possibility that in ATG, three -5 arms, added together, would save 80 runs compared to a team of awful (not average, but awful) arms. According to this 2006-based study, the difference between a CF -3 and a CF +3 is 19 runs; the difference between a LF -5 and a LF +1 is 12 runs; and the difference between a RF -3 and a RF +3 is seven runs. That adds up to 38 runs, so if arm is more important in ATG [i:28f5b892a3]and[/i:28f5b892a3] they were all -5 [i:28f5b892a3]and[/i:28f5b892a3] you were comparing them all to +3 or higher, who knows, maybe you could get it to 80. However, in 2006, when again a LF -5 saves only 12 runs compared to a LF +1, there is just no way a team of -5s is saving 80 runs compared to any realistic alternative. And you can't get three -5 arms in those games anyway.
None of this, BTW, contradicts the evidence you do provide about teams with good arms doing well. All that matters there is that the pricing system [i:28f5b892a3]undervalues[/i:28f5b892a3] the arm. If, hypothetically, arm isn't being considered in the pricing at all, and less hypothetically, a -5 arm is worth about 10 runs compared to a 0... then absolutely, over a long period of time, in very competitive leagues, once everything has evened out, you will see the teams that have the guy with the -5 arm do a little better, because that team picked up 10 runs (about a win) for free. I'm more interested right now in establishing how many runs the arm affects, simply because that is something we can know, whereas we don't know the pricing system (and also because the information applies to draft leagues as well.) But yes, ultimately the question for TSN purposes is putting together the best team under the salary cap, and if the pricing system underrates the value of the arm, then you would want to go for good arms in TSN. It doesn't have to add up to half a point off your ERA; it just has to have its effect, whatever it is, undervalued by the salary system.
[quote:28f5b892a3]I'd like to add that there's one factor not currently being considered, and that is pitching. A defense with great range and arm strength is going to benefit a team more when their pitchers are "contact" pitchers rather than strikeout pitchers. J. Santana might get just as many outs as Maddux, but the ball is hit in play far less.[/quote:28f5b892a3]Highly doubtful. All pitchers have the same exact X-rolls. And all SINGLE, DOUBLE and TRIPLE results go to the OF. So the only distinction I can see here is that the arm will matter more if more SINGLE, DOUBLE and TRIPLEs are being rolled -- in other words, against worse pitchers. But as discussed before, I don't think that changes the [i:28f5b892a3]percentage[/i:28f5b892a3] of team run prevention that the arm is responsible for (the arm will represent a larger piece of a larger pie), which I think is the larger concern.
Now I'm gonna talk about catcher's arm. Same type of study as before. The catcher here was Paul Lo Duca, who is a T-11 (a high number -- i.e., he will make a lot of throwing errors), so I don't think I can be accused of underrating the arm's value here. Lo Duca's arm is normally +1. When I changed it to +4, the team in 50 seasons gave up an average of 658 runs, or eight more than it did with nothing altered. When I changed it to -2, the team in 50 seasons gave up an average of 640 runs, or 10 fewer than it did with nothing altered.
So the summary of the arm studies:
C range 1: -13 runs
C range 3 (original): 0
C range 5: +7 runs
C arm -2: -10 runs
C arm +1 (original): 0
C arm +4: +8 runs
LF arm -5: -6 runs
LF arm -2 (original): 0
LF arm +1: 6 runs
CF arm -3 (original): 0
CF arm 0: +2 runs
CF arm +3: +17 runs
RF arm -3: -2 runs
RF arm 0 (original): 0
RF arm +3: +5 runs
My next studies will examine the effect of manager settings. For instance, in the next study, I'll sim the seasons with "very aggressive" and "extra conservative" basestealing settings, and see what SB and CS result.