Deck Shuffling

[Demagog]

The point of this isn't really to see if shuffling is biased, it's to see how it's biased. Perhaps card number twenty in a thirty card deck has a tendency to be somewhere between six and ten cards deep in your deck when it's shuffled. Information like that would allow a person to better fine tune a deck.

Anyway, if anyone wants to help, you could make a deck in which ever card is different, post the deck here (text format preferably, in the order the cards are in), and then post the order in which cards were drawn in games you played with it. I find a combination of healing and drawing let's you get through a thirty card deck rather quickly while keeping you alive, against level 1 AI. I'll convert your cards into what's needed for the data collection. What cards they are really doesn't matter; they just have to all be different.

I've completed nineteen games so far... some of the data entry is really tedious...

[Scaredgirl]

I've completed nineteen games so far... some of the data entry is really tedious...

You do know that you need thousands of games before this experiment gives any relevant statistical information, right? Less than that, and the margin of error is way too high.

[unionruler]

So far I haven't seen zanz admit to devourer bias, which is clearly a more pertinent AND obvious issue than this. Unless you've really gotten eternity last card every other game that you play an anti FG rainbow, maybe you should go collect statistics for that first? 

[Demagog]

Well if the function-copying thing in excel were better this would be a lot quicker... I wrote a formula for a few boxes in a tab to retrieve the number that card was drawn from the list in another tab, but the only way I can get the formula into another tab is to to either rewrite it from scratch (it's really long, about thirty "if" arguments), or copy it then go back through the formula and change certain numbers, but that still takes forever.

If you want to see what I mean, go look at column M in the "mode" tab. Hopefully this forum lets you attach files...

[Malduk]

I've completed nineteen games so far... some of the data entry is really tedious...

You do know that you need thousands of games before this experiment gives any relevant statistical information, right? Less than that, and the margin of error is way too high.

I dont want to spoil the party, and it would be interesting to see the results I agree, but actually, you need millions of games if algorythm is using 32 bit randomisation and even then you might not be able to draw a conclusion. You just cant do it manually.

The ONLY way to get to any conclusion in this life time is to get randomisation code, program a simulator, run all combinations, and see which shuffling results are missing.

[Demagog]

I think you're missing the point... I'm trying to see if certain cards have a tendency to stay close together or go to a certain place in the deck when shuffled. If, after a hundred games, the average position of a card is far enough away from fifteen, one might be able to conclude that the card's positioning is biased in a certain direction. If two cards happen to be somewhat close together often enough, one might be able to conclude that those cards will continue have a tendency to be close together.

So I'm not looking at the deck shuffling as a whole, I'm looking at it card by card.

[Scaredgirl]

I think you're missing the point... I'm trying to see if certain cards have a tendency to stay close together or go to a certain place in the deck when shuffled. If, after a hundred games, the average position of a card is far enough away from fifteen, one might be able to conclude that the card's positioning is biased in a certain direction. If two cards happen to be somewhat close together often enough, one might be able to conclude that those cards will continue have a tendency to be close together.

So I'm not looking at the deck shuffling as a whole, I'm looking at it card by card.

That's not how statistics work. Hundred games means nothing. Margin of error is too high.

[Malduk]

I think you're missing the point... I'm trying to see if certain cards have a tendency to stay close together or go to a certain place in the deck when shuffled. If, after a hundred games, the average position of a card is far enough away from fifteen, one might be able to conclude that the card's positioning is biased in a certain direction. If two cards happen to be somewhat close together often enough, one might be able to conclude that those cards will continue have a tendency to be close together.

So I'm not looking at the deck shuffling as a whole, I'm looking at it card by card.

It is actually the same thing. One shuffle is one string of cards. I do get what you're trying to do. Everything I wrote does apply.

[teffy]

I'm trying to see if certain cards have a tendency to stay close together or go to a certain place in the deck when shuffled. [...]

Perhaps this can be even expected , that a combination of 2 cards exist, which has a tendency to stay close together in 100 games, I mean, there are 30!/(2!*28!)= 435 combinations.
Be careful with statistics!

Zanzarino uses the Fisher-Yates algrithm to shuffle the deck twice.


P.S. I often think that the election of the FGs and T50 is not random.