Matrix Games Forums

I couldnt have said it better myself....

ORIGINAL: jacmar

The point I am making is this....Its true, that if a replay is accurate, you know who is going to win if the season plays out accurately. The two ideas that I like to do are as follows...Take a real team from any season and manage the team against the rest of the league. Then see how I do compared with the real team. If the game isnt accurate, how do you know how you really did?
The other option I like is to take a number of great teams from different eras and play it out to see which team would win. Judge, there is no way one can know that from history, because those teams never really faced reach other. If you dont have an accurate foundation, there is no way to know if you are getting an accurate outcome.
I am not criticizing Puresim in any way. Ive said numerous times that it is a nice game, and fun to play. But if you are given all of the teams from every season, and given all of the stats that go with it, I would expect to see some realisitic results if seasons were to be played out. I played out a few seasons, and didnt see anything close to being accurate. Its as if all the teams and players are equal. Thats fine if you want to see how you would do against another player or against the computer on equal terms. But if you are following the 1927 Yankees in a league with the 1986 Mets, and notice at the end of the season that Rafael Santana has a higher average than Babe Ruth, something isnt right.

Well...I asked you to enlighten me, and you did. You've just successfully explained why someone would want to realistically replay...so I stand corrected. [&o]

OK Judge...I think as you get more and more familiar with the game of baseball, you will see why some of us would like to see what would happen if Team A played Team B. Most of us grew up following baseball and argued all our lives about who was a better player, and what our team would have done if they played in a different era. With accurate replays, we can get an idea about these hypothetical questions.

ORIGINAL: jacmar

I agree it is a nice game, and a good simulation, but not realistic. I played out some complete seasons, one being 1955 and 2 for 1919. In the 1919 seasons, which were actually won by the Redlegs and Whitesox, neither team was anywhere near the top. One of the replays actually had both teams finishing last. Im sorry, but that just isnt realistic. Its a fun game, and I would reccommend it for anyone who likes baseball. In a fictional way, its great. I just feel that if one played out a given season 100 times, maybe a couple of times the teams that actually won would finish first.

Well, there's really only one way to find out. So, last night I set up a continuous replay of the 1955 season, and let it run unattended until the end of the 2054 season, giving us 100 years of data.

I've uploaded the seasons page here: http://www.angrytree.com/puresim/seasons.html (I might upload the rest of the almanac at some point, but it weighs in at around 300MB).

Here's a quick summary of what I found:
The Yankees (96-58, AL Pennant winners in '55) won 43 pennants.
Cleveland (93-61, 1 game back in '55) won 29 pennants.
Boston (84-70, 12 games back in '55) won 20 pennants.
The White Sox (91-63, 5 games back in '55) won 7 pennants.
Detroit (79-75, 17 games back in '55) won 1 pennant.
Kansas City, Baltimore, and Washington did not win any pennants.

In the NL:
Brooklyn (98-55, NL Pennant winners in '55) won 48 pennants.
St. Louis (68-86, 30.5 games back in '55) won 31 pennants.
Milwaukee (85-69, 13.5 games back in '55) won 12 pennants.
The Giants (80-74, 18.5 games back in '55) won 7 pennants.
The Cubs (72-81, 26.0 games back in '55) won 2 pennants.
Philadelphia, Cincinnati, and Pittsburgh did not win any pennants.

In real life, the Dodgers beat the Yankees in 7 in the World Series. In the one hundred seasons played in the replay, the Dodgers and Yankees met in the series 19 times, with the Yankees coming out on top 11 to 8.

So I take it with replays you look for this http://www.whatifsports.com/locker/default.asp

Several thoughts:

1.) There is the 'half-season' problem; if Joe plays 70 games for St. Louis and 90 games for Cincinnati... in PureSim he'll be loaded on Cincinnati for all 154 or 162 games, which will cause Cincinnati to do better (having him available more of the season) and St. Louis to do worse (having him unavailable).

2.) There is the 'didn't play all year' problem. If Tom is brought up to the bigs in June, but plays at a great clip thereafter, in PureSim he'll be placed on the big-league roster all season long. (Likewise, you might get a key player missing parts of the PureSim season... or a real-life manager which might have decided to play inexperienced players in September, costing himself ten games which his PureSim counterpart doesn't.)

3.) Lineup construction can cause similar differences, places where the 'right' lineup choices IRL versus the 'wrong' lineup choices in PureSim will cause the PS team to perform worse than its RL counterpart... and vice versa.

4.) Even with the 'closer' checkbox, there are still 'pitcher usage' problems, which I expect you're running into much more in that 1919 replay than in the 1955 one. If the real-life teams-of-that-era rarely went to the bullpen, but the PureSim ones go more often, the bullpen strength will contribute to performance.

5.) There are random numbers involved, both in the real world and in PureSim. A real-world team could have performed better than you'd have expected based on its runs scored and runs allowed, or vice versa.

...

I don't think you can draw any conclusions from a single season, you need to run the sort of thing AngryTree did to start drawing any real conclusions. He found that 45.5% of his pennants were won by the "correct" team, which means that only in 20.7% (.455 * .455) of his seasons, *one in five*, would he see the "correct" World Series pairing.

I think the next step would be looking at why St. Louis appears to have had trouble in this year.