A study on the luck factor in SOM

[voovits]

Another thread going on here got me thinking as to exactly how much luck plays a part of this game.
I know multiple leagues have been run in the past on this site where everyone drafts the same exact team over the course of 2 or maybe more seasons to see the variation of wins and losses, but I wanted to take it a step further.
I took the CDROM game and created a well balanced team of 40 players (25 hitters and 15 pitchers) over the course of multiple seasons ranging from 1978 through 2006. They are mostly all good players, with a few utility players.
I took that 40 man roster and gave the [i:9e5f84c12c]exact[/i:9e5f84c12c] same roster to all 30 teams. Making it fair, I manually set starting lineups, pitching rotation, and manager strategy; making them identical for all 30 teams. Taking it a step further, I made all the ballparks be even. 1-8 for both ballpark singles and homeruns, turning off weather effects. I don't know how it effects the game, but I made ballpark doubles (100) and triples (80) the same as well.
So every team is exactly the same. Roster, lineups, settings, rotation, ballpark, and active players/minor leaguers.
League settings are to play with injuries, and to try to limit overusage, but that wouldn't be a major issue. Some of the starters will need some days off and some pitchers will have to skip a few starts throughout the course of the season though, barring injuries). The pitchers hit in this simulation, I don't like the DH and I don't tend to play with it.
The only mistake I made was that the teams have all right handed starters, the only lefties are in the bullpen. A minor issue since all the rosters are identical anyway.

Here are some highlights of the completed 162 game season:

League Standings notes:
3 teams finished with 90 or more wins (Det: 95, Bal 96, Cin, 97)
3 teams finished with 90 or more losses (Stl 90, CHW 91, Atl 94)
Wild Card was won by an 87 and an 86 win team
No teams finished with exactly 81 wins.
16 teams finished within 4 games of 81 wins (between 77 and 85 wins)
Only 2 divisions and the NL Wild card were decided by 2 games or less. 2 divisions had 11 games between 1st and 2nd place.

Team Stats notes:
League batting average: .252
Highest team batting average: .268, lowest: .241
Most runs scored: 728, fewest: 601
Most HRs: 196, Fewest: 143
Most 2Bs: 336, fewest: 249
Most SB: 109, fewest: 66
Most Errors: 127. fewest: 79

League ERA: 3.70
Lowest team ERA: 3.28, Highest: 4.28
Fewest runs allowed: 582, most: 743

Most total games lost due to injury: 231, fewest: 86. League total: 4229

League Leaders notes: (Note the game lists the top 12 in the league leaders)
T. Gwynn (1995) occupies 10 of the top 12 spots, including the top 4. (W. Boggs (1987) occupied the other 2) Gwynn occupied all of the top 12 spots in total hits (he was batting leadoff)
Boggs and R. Sandberg (1992) occupied all of the top triple spots.
J. Clark (1987) occupied 9 of the top 11 slots in the homerun category (Dale Murphy (1985) had the other 2)
K. McReynolds (1988) and Murphy occupy the top 10 in RBI (Murphy was batting 3rd, Clark 4th, McReynolds 5th)
All 5 regular starting pitchers appear at least 1 time in the top 12 in wins (league leader B. Sheets (2004) had 19)
Only 3 of the 5 regular starters appear in the top 12 in losses, with R. Oswalt (2004) appearing 8 times. (League leader, Oswalt lost 22 twice)
4 of the 5 starters appear in the top 12 in ERA (League Leader: C. Swan (1978) 2.33)

So that is a rundown of the stats compiled. I plan on going through each team, singling out 2-3 hitters and 2-3 pitchers and seeing how they vary in stats from team to team. If there is enough interest in the results I will post them. I'll also post complete standings, and the complete stats (team and player) if there is interest.
I did find this experiment quite interesting and I'm considering doing it yet again removing the injury and overusage factor.

[junkfood johnny]

interesting concept. i didnt know you could do that with the cdrom game.

[coyote303]

What an interesting study! It's also interesting that your study and mine that I did in the original thread show the same thing--how much variance there is even when things are exactly the same. The variance was there even though your study was more complex and used the SOM game engine itself and I used a simple random number generator in Excel.

By the way, I didn't see this thread until after I posted my results in the original thread. Fascinating that we were both inspired to demonstrate our thoughts the way we did, in similar yet very different ways!

Coyote

[ClowntimeIsOver]

It's just the bell curve. It's nothing special to SOM -- it's implicit in statistics and reflects natural distribution, which is different from "luck." Skill would equal the ability to "diversify" each roster you create, over several seasons, so that there would be a slight surplus after the "fat ends" of the distribution curve cancel each other out (again, over the course of several teams in several seasons). It's similar to the question of why Warren Buffett and other "Graham-and-Dodds" investors do better over time than a random walk down wall street.

[voovits]

[quote:e66919b103="wachovia07"]It's just the bell curve. It's nothing special to SOM -- it's implicit in statistics and reflects natural distribution, which is different from "luck."[/quote:e66919b103]

I disagree. The bell curve in this instance is created by die rolls, which is luck. All the skill in the world could not force the dice to land on 3-5 on the hitters card and result in a homerun.

The job of the manager is to try to use the percentages to their favor. It's his or her job to figure out the percentages, and set rosters a lineups accordingly in such a way that the percentages give you the better chance over your opponent. That sounds remarkably like poker. The best poker players in the world don't always win the games they play, just like in strat. bad beats happen all the time because of the luck factor, just like in my example above.

This example has all 30 teams perfectly even and there was such a big discrepancy. In theory if I were to do it infinite times, the teams would all ultimately get closer and closer to a .500 winning percentage.
If I took one of the teams and make them better or worse (insert a better or worse manager) the numbers start to skew in favor or against the manager in question. That's the skill part of it, but the luck part is still there as the good manager will not always win the league and the bad manager will sometimes win the league, it's just that it will happen either more or less frequently, depending on which example you use.

[LoopsandRolls]

I am curious - do you see a correlation between injuries and runs scored/runs allowed? An extra 145 games missed on a base of 86 is a large variable.

[voovits]

[quote:c40c447635="LoopsandRolls"]I am curious - do you see a correlation between injuries and runs scored/runs allowed? An extra 145 games missed on a base of 86 is a large variable.[/quote:c40c447635]

Well there is a quick and easy answer to this.
The Yankees missed the most games due to injury (231) and they had a final record of 77-85. They scored 691 runs and allowed 692. Almost even.
Oakland suffered the fewest injuries (86) and they had a final record of 83-79. They scored 659 runs and allowed 665 runs.

That's quite an interesting result. New York suffered way more injuries and yet they scored more runs.
Taking a look at their roster, I see they got 11 starts from starters not in the original 5 man rotation. Dave Stieb (1988) started 7 games and Ed Lynch (1985) started 4.
3 of their 5 regular starters started fewer than 30 games. Mike Witt started only 22 games, so he was surely injured, possibly more than once. Ed Halicki (1978) started 27 games and Craig Swan started 29, so each of them may have gotten hurt. All 3 of them started 30 or more games on most teams. Also note that any injury that occurs is likely a 15 game injury as per the super advanced injury rules (which are MUCH better than the normal injury rules).
In other words, injuries clearly hurt New Yorks pitching.

Florida suffered the next highest number of injuries (222) but they didn't seem to suffer as much.
85-77 record. 650 scored, 597 allowed. They scored almost as much as Oakland, yet gave up far fewer.
I think that a better indication of a possible correlation would be if the backups were not as good. The teams had a lot of good replacements in case of injury.
I'll post the entire roster if anyone wants to know it.