Random Number Generator?

Does anyone have any insight or has SOM ever published what algorithm they use to randomize the dice rolls?

Obviously player performance and stats are based on a lot of things, and knowing that half the time your player doesn't really factor into the outcome makes a huge difference.

I picked up Mike Jackson (1998). So far he has faced 183 batters (rolls) and 114 of them have come off the hitters card. If we assume a 50-50 binomial equation, then the likelihood of 114 or more rolls coming up 1,2,3 is 0.0005433465. I won't get into statistical "p value" stuff, but let's say that this result shows that dice rolls, as least for Mike Jackson are very (very) likely not being randomly generated with 50-50 odds.

To put it another way, if you were monitoring a manufacturing floor and saw results like this you would conclude that you have a process out of order.

Any thoughts out there?

I'm not a #'s guy but...

If you used Jackson on a few teams, the roll distribution would even out.

A 1 time use isn't gonna tell much.

LOL - I will agree, you are not a numbers guy.

"A 1 time use isn't gonna tell much." The bottom line is that 183 binomial events showing results that are outside of a +/- 3 sigma grouping is in fact, statistically significant.

What difference would being on a different team make? Randomness across teams or seasons isn't the goal - it is supposed to be 50-50 for each pitcher/hitter confrontation.

The results show that it is not a random event - that something else is having an influence.

But I do appreciate you taking the time to respond. Thank you.

A quick perusal of my 12-team league and I could find only one imbalance that rivals Mike Jackson's 62%/38% after 183 rolls. (Dick Selma 60%/40% after a similar number of rolls (181)) He also was a relief pitcher. Good place to start for conspiracist theories, I suppose. Like, maybe Dick Selma/Mike Jackson is an example of a pre-designated player of a "catch-all" category whose imbalance is pre-determined in order to ensure the total batter/pitching balance of the entire pitching staff will be close to 50%. Why would they do that, you say? Well, maybe the game runs faster if you roll the dice ahead of time before the games start instead of after every batter, I don't know. Instead of distributing batter/pitcher roll imbalance corrections among several players to fix the 50/50 balance being out-of-whack, they just do it to one player. I don't know, maybe.

Thanks Paul - that's an interesting theory. I didn't know SOM messed with that stuff, but this is my first time back since the '90s when we had a league playing their PC game.

As maybe an interesting side note, and it was over 20 years ago, but we had a guy who played the PC game with the dice rolls turned on, and he eventually quit the league because he could tell what rolls were coming next and would unfairly adjust his decisions. Basically he didn't want to cheat but knew he couldn't control himself.

In our newly formed league there has been a lot conversation around why SOM continues to use the cards and dice rolls for their computer simulations. There are far better ways to calculate the impact that pitcher and batter have on one another other than using just the pitcher half the time and vice versa.

Of course last time I posted something like this my team went on a long losing streak...

Mick -- the odd are not 0 that with a "fair" random number generator that you will get this result...granted they are low, but not 0.

You have been very unlucky with Jackson, but if you looked at the dice rolls for him as the sample size increased, you would find that they would move toward a 50-50 split. It might take a lot of rolls, but you'd get there. Look at the hitter/pitcher splits for every player in the entire league and it is pretty close to 50-50 -- with a much larger sample than just Jackson.

There is a chance that there is a problem with SOM's random number generator, but they're pretty established these days.

looking at all of the teams in the league, only two teams have as much as a 51.2% "tilt": the Roos hitters have been on their own cards and the Hounds pitchers have been on the batter's cards that share of the time. Every other team's hitters and pitchers are closer to 50-50.

Your pitchers have been on the batter cards 51.1% of the time, but if you take Jackson out it drops to 50.5%.

Jackson's results are a small sample of the entire league -- I haven't computed the aggregate numbers, but they must be very close to 50-50.

Again, the more chances (rolls) that Jackson has, it is very likely that he would move inexorably toward a 50-50 split.

According to the binomial distribution calculator on the internet MickNuggets calculation of 0.0005433465 for Mike Jackson in his first post is indeed correct. His question is quite valid, given the sample size of 183. You don't need a bigger sample size to question how a sample size of 183 could result in such a low probability of 0.00054333465 that there would be 114 "successes" and 69 "failures" for Mike Jackson. That's all I know.

A low probability, undeniably, Paul...but still not 0. And the larger sample size of the entire league is about 50-50.

A low probability, undeniably, Paul...but still not 0. And the larger sample size of the entire league is about 50-50.

No argument with a larger sample size leading to a 50/50, but, that does not help explain what happened with Mike Jackson and Mike Jackson only, no matter how many leagues and teams are played on a daily basis.

If what happened with Mike Jackson (114/183 rolls on hitters card) only happens once out of 2000 times (approximately .0005433465), then, I would say nothing fishy is going on.

But, If the entire league is about 50/50 and the probability of what happened with Mike Jackson's 183 pitcher/batter rolls is 0.0005433465, then, I'd say someone has some explaining to do (but, only if this imbalance instance happens to more than 1 of 2000 "cards").

Perhaps, Mike Jackson pitched with a high fatigue factor (F0) several times, causing the game engine to roll the dice on the hitter's card more often than normal. If Jackson pitched as an F0 three or four times the probability of say, 114 of 183 instead of 105 of 183 paints a different picture when plugging the numbers into the binomial distribution calculator.