DIE ROLLS

[The_Inevitable]

Has anybody actually run hypothesis tests on the data set used? I'd be interested in seeing the results.

There's a dice analyzer in the Plug-Ins section, a lot of people post their results if you want to look for them. The probabilities always end up where they should.

But still, the data set is static. Should be able to analyze the whole thing.

[Michael Kowalson]

Did you decide to look for doubles before you rolled those dice, or did you roll them and the see if you could find anything odd?

Wasn't looking for doubles at all - just noticed it when recording them. Thought it to be very odd.

[Michael Kowalson]

I'm well aware of how to do a proper and ethical statistical analysis - used to do it for a living - and I'm saying - over 20 games - now at 23 and the odds are totally NOT matching up.

I have checked random.org and rolled 400 rolls of 2 dice and found that doubles (two of the same die) occurred 38% of the time. That is far above statistical probability. If you read their explanations and they make sense but the actual probabilities don't bear them out well. point is there is an issue that needs to be looked at.

for someone that knows 'how to do a proper...statistical analysis,' you sure don't understand the law of large numbers. Or, you are choosing to ignore it, which would be even worse.

I understand the laws of large numbers - as well as statistical sampling - which I am nearing. Agreed I was nowhere near the correct statistical sampling required - although I am nearing and there is a disturbing trend int he multiple rolls. that in a 3-2 auto attack - the differentials in each attack remain far too similar - such as a 5-5-4 vs. 6-5 then a mysterious 3-3-2 followed by a 4-3 the differential remains the same - in a true random world with dice the differential should differ far more frequently.

[Michael Kowalson]

where is this dice plug in?

[MeDeFe]

Did you decide to look for doubles before you rolled those dice, or did you roll them and the see if you could find anything odd?

Wasn't looking for doubles at all - just noticed it when recording them. Thought it to be very odd.

And exactly that is the problem. You take a sample of random numbers and look for oddities, I can tell you: you're bound to find them in almost every sample you take, either an excess of one or two numbers compared to the others, a lack of one or two numbers compared to the others, or you'll notice a lot of streaks of one number, or, as you found, a lot of doubles. It could be anything. Try it sometime, take samples of 100 rolls or so and see what statistical oddities you can find. There are so many things to look for that you'll practically always find one of them.

[Fruitcake]

I have always had a gift for spotting trends and seeing life in mathematical terms (some would say how sad, sometimes I tend to agree)

However, what this means is I quickly learnt, almost as soon as I was introduced to ERNIE (Permium Bond random prize generator) in the early sixties when I was a young boy, that the two mathematical laws of probability and averages cannot be applied to any computer based randomisation. By the very nature of the software driving it, computer based randomisation must leave those rules, at times, to survive.

In turn this means you should spend a great deal more time watching not just the results of rolls, but the scores while attacking. If you watch carefully, you will see trends apear, it is at these times, that clear and precise thinking is required.

Some of what has been said here is bang on the button, although some of the logic doesn't bear witnessing.

Just apply the age old rules...always attack at the beginning of the game with even number armies etc etc...be careful in your attacking, and plan your moves.

But seriously....

The Gods of Dice are capricious swines at best, treacherous bas**rds at worst.