Topic: Just-Dice.com : Invest in 1% House Edge Dice Game - page 61. (Read 435357 times)

Yeah the more values you add, the 3's and 6's are going to tend to come into line. However, that's not because 6's are somehow more likely, or 3's any less. It's just because as you add more values, the discrepancies become relatively smaller.

Try this: what happens if you add 5000 to all the results? E.g. the fair die is just doing it's job being perfectly fair now. Your new totals are:

1-7000
2-7000
3-8000
4-7000
5-7000
6-6000

Now the 3s and 6s are still 1000 away from where they should be, but as a percentage they're only about 15% out of line instead of 50%. It is only in this sense that things eventually "come into line".

SebastianJu, I've said it before (and I'm not going to read the 100+ pages I missed already), the distribution of results from dice rolls on the site is uniform.

It's this uniformity that allows investors to have confidence in getting a return.

It is also this same uniformity that convinces gamblers to have confidence in whatever strategy they wish to apply.

I've been both. But don't mind me, I rely on weird stuff to gamble or invest.

I've stopped gambling (or investing) for the past 2 or 3 months on games, and have been day-trading / week-trading BTC on the various exchanges. Doubled my coins twice.

But it's too slow, I think I'll go back to JD and play a little bit.

Ah I see. I tried to search for a contact form or anything to check who owns it. Oh well, nevermind my comment then, haha.

I still have a hard time understanding it. I see that random values have no past. Its random. But random doesnt look random anymore the more values you have. So even when you throw a perfect dice 10,000 times and you get these results:
1-2000
2-2000
3-3000
4-2000
5-2000
6-1000

then the more values you add you can be sure that the 3 and the 6 will be thrown nearly the same amount of times like any other digit at a big enough amount of throws. So that means there IS a force that enforces that result. That doesnt mean you cant get results like above. Since its random but something works sorting it.

If just-dice for some reason gets 10 or even 100 times more games per day then you can be pretty sure the real chart will go more and more smooth to the 1% house advantage. It will become harder and harder for random to break that the more values are involved.

When i see the current chart i "think" its inevitable that it will drop below the 1% line in a timeframe. Maybe 1 week, but might be months too of course because, like one wrote, there is no past. But it will go to 1% at one point because stochastic's say so. So one could guess there is the possibility of the dice results above. But the chance is very slim because something works against. So why should it be different here?

Saying that... i dont see in the chart how one could invest trying to exploit that safely without losing more at the end than profiting since its way too random.

I only want to say the more games are played the higher the chance it will meet the 1% line.

I think i have to keep it invested. Even though it looks dangerously away from the 1% line and i feel like the force of more values will force it to break the 1% line again... making the invested bitcoins lose quite much. But i cant be sure that it wont grow the next month at this high level.

Its probably useless to think about. 

Many disappoint.

It's true; I made doge-dice.com.

It's pretty much just a clone of JD but using dogecoin instead of bitcoin.

it's both run by dooglus aka dogelus 

Hey Dooglus,

Did you see that doge-dice.com cloned your website design? https://bitcointalk.org/index.php?topic=374197.new;topicseen#new

I would be pissed off if I were you. Anyway, just wanted to let you know.

I think he's referring to random walk crossing times, or brownian motion hitting times. The profit might be able to be modelled as such, although I haven't been able to.

As an aside, my best model of the change in profit (as a percentage of amount bet) per time period is a stable distribution, with parameters varying depending on the total bet per time period and the amount of time per time period.

Alpha is below two for all time periods up to 24 hours, so the 24 hour change in profit is definitely *not* normally distributed. As a rule of thumb, the 90% confidence interval for daily change in profit is about 9% of the total bet (based only on available data, ie since 11th October).

No. "The dice have no memory." 100 wins does not change the probability of a loss on the next roll (although it may indicate loaded dice, but just-dice is provably fair).

Apart from many mathematical considerations, I think the most difficult proof for an investor is to still believe in the website when a whale arrives, like Nakowa, and starts to win. Have a look at the profit chart then, THAT is something scary.

The 'when' can be tomorrow, it can be in 1000 years (hypothetically). The fact remains that investing is, at any point in time, still a gamble with a positive expectation value.

There is no 'random pressure' to break the line that rises with each bet.

Easy / not easy is relative to ones experience And a Martingale strategy performs better than a straight-up bet under the right conditions. Most people running Martingales don't meet those conditions, but that's besides the point.

The time intervall between crossing can be very high but the longer the time the higher the chance of breaking the line. So if i see the current chart i fear that i will soon lose a big part of my investment (i invested a lot more now) since its inevitable that the line gets crossed. Only when is the question.

In theory i should be able to simply wait and it will grow over time. What i wonder is if the past now doesnt work against me since the random pressure to break the line again is there and is rising with each bet.

That mathematics really isnt easy to understand. Thats probably why still so many people believe in martingale.

If you consider betting as a random walk, then the profits should cross the 1% line infinitely many times. But, the time-interval between 2 crossings can be arbitrarily large.

Regression towards the mean happens when you look at the relative difference of the profit compared to the expected profits. If the site would be at 2% profit at 1M wagered, and would hit an exact 1% profit over the next 1M wagered, the total would add up to 1.5% profit over 2M wagered, so the percentage has moved closer to the mean, but the absolute difference has remained the same: The site is still up 10000 coins more than was expected.

Profit% is total_profit / total_wagered.

Actually there's the notion of regression to the mean given a large number of coin flips.  That's not the gambler's fallacy.

http://en.wikipedia.org/wiki/Regression_toward_the_mean

However using that as an investment strategy is probably more like gambling or day trading than a proper investor should be worried about.  You just end up obsessing about tops and bottoms instead of letting your money grow slowly over time.

Also, I believe the profit % is just (current_total_profit / current_total_investment) so the number on the main page isn't really an accurate representation of where we're at relative to the theoretical edge from a statistical perspective.

Thats what i would think too... but then... stochastics say that a 1% house advantage should lead to having the profits going along a 1% line. It happened in the past too. So shouldnt it go around the 1% line all the time? If it does then it might not go below 1% in the next week but at some point it will again.

Random doesn't have a past.

Divesting when profit is above expectation is another form of Gambler's Fallacy. The outcome of future bets does not depend on what happened in the past.

Im not a pro in stochastics so i ask myself if it makes sense to divest when the profit is so high above the 1% profit and invest once its below the 1%. Normalle the profit chart should flow around that line. So it might be possible to raise the profit that way.
On the other hand... random doesnt have a past. It could be seen as now starting so it should move on from the same amount with 1% profit. But with the past values it shouldnt... it should go way down now.
What view is correct on this?

Small correction: The max profit is 0.5% of the house bankroll. So that's a max bet of 0.5% at a x2 multiplier, but very high payout bets have much lower max bets.

Max bet means a better can maximum bet a half percent of the house bitcoins in one game? But the house advantage is still 1%? Then this only means the player would play twice and it would be a bit less variance... not really less bitcoins played...