Maximizing Run Differential

Has anyone run a computer simulation to determine the best combination of 9 or 10 pitchers and 15 hitters to maximize run differential for a typical $80 million auto league (no DH) for the ATG9 all eras data set?

If so, are you willing to share your results? I'm just curious because there's a lot of disagreement about the ideal ratio of amount spent on pitching vs. hitting to have the best of both worlds, where you have high value hitters who score lots of runs, and at the same time don't give up a lot of runs by having good pitching backed by a solid defense.

This is the ultimate quest, the Holy Grail of Strat-O-Matic immortality.

Full disclosure, I have tinkered with roster construction quite a bit over the years, never ran any computer simulation however, just relied heavily on Diamond Dope (thanks Adrian!). It's a moving target for sure, especially when new versions come out and salary adjustments occur, and you have to pick a particular ballpark to experiment with. But ultimately you keep playing the same game with similar rosters, which can get dull and stale. I'm nearly done, down to my last teams, so I wanted to poll the audience and see if anyone has the answer to the big question.

You're talking about a question that I've thought about a lot. Here are my quick reactions:

I'm not sure how you would resolve this by simulation without recreating the SOM game engine. It could be simplified some, but it's too big a job for me to tackle.

If I understand what you are saying, you have approached the question by trying to experiment with your own teams. How many teams do you run at each salary split? It seems to me that you would need to run quite a few teams at each split to get a good fix on win %, given that the standard deviation for any expected win number is about 6 games. For example, a team with a true win probability of 55%(89 wins) will actually win between 83 and 95 games 70% of seasons played. Then consider that an identical team playing in two different leagues will have different win expectations based on the make-up of the rest of the league.

I've looked at Barnstormers data. Looking at roughly 900 teams at the $80 m salary level, the average salary split is about 59%/41%. When I stratify the data by wins, I can't discern much difference between successful teams and unsuccessful teams. I'm sure the problem is that there are just too many other factors.

Another approach is just to think about which players influence the outcomes in SOM and the associated probabilities.
For a given at bat, there are 216 possible rolls. 108 on the batters card and 108 on the pitchers card. Of the 108 on the pitchers card, 30 are X-chances, of which 2 belong to the pitcher. So, of the 216 rolls, 136 are determined by a position player and 80 are determined by the pitcher. That's about 63%/37%. In a non-dh league some of those batters are pitchers so I think the % split gets closer to 60%/40%. Other factors like throwing, running, stolen base ratings, pb, wp, holds, balks, etc may move this a little but I doubt if it is by much.

And for those of you that do not like this much quantitative analysis, rest assured that my win % is only a solid.500 (but rising).

The really fascinating cases, the ones I try to learn the most from, are the fortunate teams that led their respective leagues in both runs scored and runs allowed. I've had a couple of these. Hard to duplicate, either because you don't get the same roster of players the next time around, or the league makeup (especially ballparks) is significantly different, either more hitter friendly or tilting heavily in favor of pitching and defense.
Complicating matters is even if you you "borrow"(copy) someone else's max run differential roster exactly down to the bench players, you may have vastly different results because of manager settings.

One would definitely think that the golden ratio would be around 60/40 because pitchers determine only 10 out of 27 rolls. But the fact is middle class pitchers are simply overpriced. So the correct strategy is almost always closer to 4 Aces, and the correct % spent on pitching is often greater than 50.

for an 80 million dollar league:
35% Pitching
65% hitting

And you better optimize and get the highest value players for your league.
Example, if you have 75-80% righties in your league you want Don Sutton and Joel Pinero on your team.
If you have a league with poor catchers, you want the top base stealers.
If the average team is more than 50% pitching then you can reduce the quality of your defense to increase your offense.
Try to work it so the managers are inexperienced, that is a big help.

Barnstormer 102 win 80 Million team lead league in run dif 35.7% pitching
https://365.strat-o-matic.com/team/1678222
Barnstormer 99 Win team
33.4% pitching
https://365.strat-o-matic.com/team/1678373

So there you have it..

I agree with Max on this one. I’ve won a bunch of rings focusing on being first in pitching, with offenses ranging anywhere from 6th to 12th.

If you want a lot of wins, make a Polo team. If you want a lot of rings, play in Braves. Having a solid, balanced team just means you finish sixth in offense behind a Coors, two Polos and two Navins, and sixth in pitching behind a Braves, two Forbes and two Petcos. There’s a good reason that neutral teams, like neutral parks, are generally suboptimal.

My best $80 mil off/def effort was 2nd in offense and 4th in pitching in Crosley. The salary construction I used I think mimics other solid, well-rounded teams ($28.5 pitching, $51.5 hitting). FWIW, when I started playing in 2002, Riggo told me to “learn to build high scoring Petco teams.” I don’t know if in the last 20 years I ever had any better advice.

https://365.strat-o-matic.com/team/1656370

Again, however, this was an outlier for me. I generally favor high dollar pitching at +$40 mil.

The best +/- from each 2022 Event 2 league:

+192 https://365.strat-o-matic.com/team/1678054 45% pitching
+218 https://365.strat-o-matic.com/team/1679445 42%
+153 https://365.strat-o-matic.com/team/1674194 51%
+202 https://365.strat-o-matic.com/team/1681759 36%
+179 https://365.strat-o-matic.com/team/1681682 37%
+299 https://365.strat-o-matic.com/team/1670753 43%
+125 https://365.strat-o-matic.com/team/1680906 32%
+184 https://365.strat-o-matic.com/team/1681768 29%
+190 https://365.strat-o-matic.com/team/1679816 39%
+182 https://365.strat-o-matic.com/team/1681063 39%
+118 https://365.strat-o-matic.com/team/1681598 45% (me)
+176 https://365.strat-o-matic.com/team/1681644 38%
+209 https://365.strat-o-matic.com/team/1681522 42%
+ 80 https://365.strat-o-matic.com/team/1681896 39%
+147 https://365.strat-o-matic.com/team/1682178 36%
+139 https://365.strat-o-matic.com/team/1679402 40%

The 10 teams with the most wins in ATG9 auto-leagues:

+386 https://365.strat-o-matic.com/team/1638745 30%
+291 https://365.strat-o-matic.com/team/1608230 35%
+295 https://365.strat-o-matic.com/team/1679293 52% (me)
+182 https://365.strat-o-matic.com/team/1608377 38%
+215 https://365.strat-o-matic.com/team/1628480 57%
+245 https://365.strat-o-matic.com/team/1657945 31%
+156 https://365.strat-o-matic.com/team/1633150 32%
+166 https://365.strat-o-matic.com/team/1636252 40%
+283 https://365.strat-o-matic.com/team/1666797 37%
+208 https://365.strat-o-matic.com/team/1668431 39%

Clearly there are many ways to build great $80 teams. Unfortunately none of this data tells us which % leads to the best results most consistently. But again, FWIW, my numbers consistently tell me that a cheap offense holds more value than a cheap pitching staff.

Has anyone run a computer simulation to determine the best combination of 9 or 10 pitchers and 15 hitters to maximize run differential for a typical $80 million auto league (no DH) for the ATG9 all eras data set?

Theoretically, it sounds possible. But only if all of the external variables (ballpark, opponent teams, opponent ballparks, strategies, etc.) are held constant. In other words, each league has a different optimum roster. That would explain why the top performing teams often employ widely different strategies.

Before the repricing, there were some obvious bargain players that would be part of the optimum team. But everyone quickly caught on and Joel Piniero became an open secret. In ATG9, players are at best a situational bargain (Bagwell in a league with 50% left handed pitching).
Of course, there are some universal ways to improve run differential such as batting your better hitters at the top of the lineup. But this is also well known and all managers are using a somewhat efficient lineup.

Barnstormers 2018-2021
$80 million - no dh
wins # teams Avg. % on position players
0-69 89 57.7%
70-74 114 57.6%
75-79' 165 58.8%
80-84 221 57.6%
85-89 171 60.1%
90-94 91 59.3%
95-99 40 60.8%
100-104 19 61.3%
105+ 3 61.0%

All 913 58.7%

of the 62 teams with 95+ wins the minimum spent on position players was 48% and the maximum was 76%