OOGP14 Simulation Accuracy

djawic · 07-17-2013, 11:51 AM

I am a newcommer to OOTP14. When I simulate the 2012 MLB historical season I consistently get these divisional winners: AL East-- Boston by 1 over Tampa Bay; AL Central -- Minnesota by 3 over Chicago; AL West -- Los Angles by 11 over Oakland; NL East -- Washington by 13 over New York; NL Central -- St. Louis by 11 over Cinncinati; NL West -- Arizona bu 11 over San Francisco. These bear little resemblance to the actual results which is very disappointing for a program touted to provide accurate simulations. I am using the "The New Historical Game" option selecting the 2012 season with "single season replaY" with "real historical transactions" and "real historical lineups" and "disabled Player development" and "league expansion" and "financial imports" turned off and all default "picture" options. I have selected no team to manage and am not using "commissioner" mode. Am I doing something incorrectly? Thanks for your help.
djawic

Charlie Hough · 07-17-2013, 02:55 PM

How many sims have you run? First of all, to ensure that you have sufficient sample size, I would run a good 10 to 20 full season sims and then evaluate your results. Ideally, you'd run even more seasons to reach statistical significance and allow you to draw valid conclusions.

A couple of your AL division winners are suspect (Boston and Minnesota), but the others in the AL and NL are all within the expected range of statistical variation.

It all comes down to mathematics, and the difference between OOTP and other sims is that it doesn't 'cheat', penalize players, or otherwise use artificial techniques to influence statistical probabilities and force outcomes to more closely resemble what happened in real life.

It sticks to mathematics and real statistical probabilities, which is actually more realistic than artificially rigging things. In real life, if the season was played again, you would likely see the same sort of variation that you see in OOTP. But, generally, OOTP produces division winners that are accurate or definitely within standard statistical variation.

Also, OOTP allows injuries for players that may not have occurred in real life, so if you're allowing injuries, that can be a problem.

Also, you need to make sure that your game is set up to adjust or weaken players with relatively few at bats or innings pitched. Sometimes a player can be skewed too well or too poorly using the default settings. I tend to use a 200 AB threshold and a 40 IP threshold for adjustments. Then I weaken batters with fewer than 50 AB and pitchers with fewer than 25 IP.

CatKnight · 07-22-2013, 11:40 AM

Yep. No offense, djawic, but the fact you gave us 'games ahead' for each division tells me you didn't run the sim too many times. 1, 2 even 3 times simply isn't statistically significant.

You should actually expect to see a lot of differences from sim to sim. OOTP can be extremely unpredictable in that regard.

07-17-2013, 11:51 AM	#1
djawic Bat Boy Join Date: Jul 2013 Posts: 7	OOGP14 Simulation Accuracy I am a newcommer to OOTP14. When I simulate the 2012 MLB historical season I consistently get these divisional winners: AL East-- Boston by 1 over Tampa Bay; AL Central -- Minnesota by 3 over Chicago; AL West -- Los Angles by 11 over Oakland; NL East -- Washington by 13 over New York; NL Central -- St. Louis by 11 over Cinncinati; NL West -- Arizona bu 11 over San Francisco. These bear little resemblance to the actual results which is very disappointing for a program touted to provide accurate simulations. I am using the "The New Historical Game" option selecting the 2012 season with "single season replaY" with "real historical transactions" and "real historical lineups" and "disabled Player development" and "league expansion" and "financial imports" turned off and all default "picture" options. I have selected no team to manage and am not using "commissioner" mode. Am I doing something incorrectly? Thanks for your help. djawic

07-17-2013, 02:55 PM	#2
Charlie Hough Hall Of Famer Join Date: May 2006 Posts: 3,625	How many sims have you run? First of all, to ensure that you have sufficient sample size, I would run a good 10 to 20 full season sims and then evaluate your results. Ideally, you'd run even more seasons to reach statistical significance and allow you to draw valid conclusions. A couple of your AL division winners are suspect (Boston and Minnesota), but the others in the AL and NL are all within the expected range of statistical variation. It all comes down to mathematics, and the difference between OOTP and other sims is that it doesn't 'cheat', penalize players, or otherwise use artificial techniques to influence statistical probabilities and force outcomes to more closely resemble what happened in real life. It sticks to mathematics and real statistical probabilities, which is actually more realistic than artificially rigging things. In real life, if the season was played again, you would likely see the same sort of variation that you see in OOTP. But, generally, OOTP produces division winners that are accurate or definitely within standard statistical variation. Also, OOTP allows injuries for players that may not have occurred in real life, so if you're allowing injuries, that can be a problem. Also, you need to make sure that your game is set up to adjust or weaken players with relatively few at bats or innings pitched. Sometimes a player can be skewed too well or too poorly using the default settings. I tend to use a 200 AB threshold and a 40 IP threshold for adjustments. Then I weaken batters with fewer than 50 AB and pitchers with fewer than 25 IP.

07-22-2013, 11:40 AM	#3
CatKnight All Star Starter Join Date: Apr 2007 Location: Cleveland, OH Posts: 1,199	Yep. No offense, djawic, but the fact you gave us 'games ahead' for each division tells me you didn't run the sim too many times. 1, 2 even 3 times simply isn't statistically significant. You should actually expect to see a lot of differences from sim to sim. OOTP can be extremely unpredictable in that regard.