Just-Dice.com : Invest in 1% House Edge Dice Game

[bobboooiie]

Edit:  It appears that the amount invested has remained fairly stable for about a week now.

Surprising, but true.

It seems the "Hey, this thing works; I'm in!" effect is equal and opposite to the "Nice; I'm no longer in the red; quick, let's get out before celeste comes back" effect.

Yeah, prepare yourself for "why dont we raise maximum win" week

[iANDROID]

Code:

[b]16:37:26 (54367) <roflmao> PENIS?[/b]
[b]16:38:02 (54367) <roflmao> testicle{{{{{ha[/b]
16:38:18 (7791) <Keven> lordsonkit 最好有个地方大多人都说英语,为样才快.
16:38:21 (54221) <daffy> Chat is a clusterfuck today
16:38:31 (4764) <uvwvj> test
16:38:34 (886) <iAndroid> Yep
[b]16:38:47 (54367) <roflmao> I AM BIG PENIS[/b]

Sometimes on J-D, someone is crazy.

[infested999]

Historical site profit graph:

[wolverine.ks]

any way to get a graph just like ^, but overlay colors to represent standard deviations away from the expected value? I.e. green along the EV line, changing to yellow and red after getting 1 and 2 standard deviations away.

any suggestions how I could do something like that?

[dooglus]

The red line is the theoretical profit, assuming no variance?

Yes.  It's just 1% of the total wagers.

any way to get a graph just like ^, but overlay colors to represent standard deviations away from the expected value? I.e. green along the EV line, changing to yellow and red after getting 1 and 2 standard deviations away.

any suggestions how I could do something like that?

I expect it would be easiest if I generated it.  Do you know how to calculate it?

[infested999]

The red line is the theoretical profit, assuming no variance?

Yes.  It's just 1% of the total wagers.

any way to get a graph just like ^, but overlay colors to represent standard deviations away from the expected value? I.e. green along the EV line, changing to yellow and red after getting 1 and 2 standard deviations away.

any suggestions how I could do something like that?

I expect it would be easiest if I generated it.  Do you know how to calculate it?

Rannasha calculated the formula for find 1 Standard Deviation in this post: https://bitcointalk.org/index.php?topic=242962.msg2778245#msg2778245

I was wrong

[Rannasha]

The red line is the theoretical profit, assuming no variance?

Yes.  It's just 1% of the total wagers.

any way to get a graph just like ^, but overlay colors to represent standard deviations away from the expected value? I.e. green along the EV line, changing to yellow and red after getting 1 and 2 standard deviations away.

any suggestions how I could do something like that?

I expect it would be easiest if I generated it.  Do you know how to calculate it?

Rannasha calculated the formula for find 1 Standard Deviation in this post: https://bitcointalk.org/index.php?topic=242962.msg2778245#msg2778245

That equation is incorrect. I corrected it a few posts later and computed the expected standard deviation with a db-dump dooglus made:

https://bitcointalk.org/index.php?topic=242962.msg2778468#msg2778468

It's actually pretty easy (math/data-wise anyway) to keep the standard deviation of expected profits (and the corresponding z-factor of the achieved profit of the site) updated in real time.

[Oleander]

You might be interested to see the results of this simulation that I ran a few days ago. It basically takes as a starting point that there is a whale with a bankroll comparable to the house's bankroll, and the whale bets the maximum until one side or the other goes broke (i.e., the most extreme and aggressive whale scenario theoretically possible).

Your code assumes that the whale always makes a maximum bet at 2X odds... as opposed to a smaller bet at higher odds for the same maximum profit, which would lead to higher variance.  So this is not the most aggressive whale scenario.  The database dump shows a number of largish bets at high odds.

As for "the house always wins in the end", can we quantify "in the end"?  Could be a long time, if a large drawdown reduces the maximum profit to the point that the whales leave.

[dooglus]

Do you have a version that is less transparent?  I'm finding it hard to read.

[Rannasha]

Do you have a version that is less transparent?  I'm finding it hard to read.

Oh, didn't know there were different styles for this forum, I just used a simple online LaTeX generator.

The equation reads:

Sqrt( Sum( c_i (0.99 b_i)² p_i (1 - p_i) ) )
where you sum over each line from the db-dump you posted (bets-by-bet-and-chance.txt), taking the count, the bet-size and chance.

If you instead have a dataset containing each individual bet, you remove the c_i from the equation and sum over all bets. At the end, take the square root.

This value can be easily updated in realtime by storing the sum (before taking the root) and adding (0.99 b)² p (p - 1) to it each time a bet is made. Then to display the expected standard deviation, simply take the square root. From this you could calculate the z-value by dividing the difference between the current house profits and the expected house profits (= 1% of amount wagered) by this expected standard deviation. If JD results are a perfect normal distribution, then any z-value between -2 and 2 is within the 95% confidence interval.

[infested999]

The red line is the theoretical profit, assuming no variance?

Yes.  It's just 1% of the total wagers.

any way to get a graph just like ^, but overlay colors to represent standard deviations away from the expected value? I.e. green along the EV line, changing to yellow and red after getting 1 and 2 standard deviations away.

any suggestions how I could do something like that?

I expect it would be easiest if I generated it.  Do you know how to calculate it?

Rannasha calculated the formula for find 1 Standard Deviation in this post: https://bitcointalk.org/index.php?topic=242962.msg2778245#msg2778245

That equation is incorrect. I corrected it a few posts later and computed the expected standard deviation with a db-dump dooglus made:

https://bitcointalk.org/index.php?topic=242962.msg2778468#msg2778468

It's actually pretty easy (math/data-wise anyway) to keep the standard deviation of expected profits (and the corresponding z-factor of the achieved profit of the site) updated in real time.

That image can't be any less transperant

[nimda]

Can't you just open the image in a new tab?

[Otoh]

One week after having started to back the house bankroll I am extremely happy with my share of the edge so far, thanks doog & Deb!

[dooglus]

One week after having started to back the house bankroll I am extremely happy with my share of the edge so far

It has been a good week.  Please don't be too surprised when we have losing weeks!

[Otoh]

One week after having started to back the house bankroll I am extremely happy with my share of the edge so far

It has been a good week.  Please don't be too surprised when we have losing weeks!

I've seen the Historical site profit graph so won't be surprised, I'm still relying on investor's fallacy to help me along though, that & astrology

[molecular]

One week after having started to back the house bankroll I am extremely happy with my share of the edge so far

It has been a good week.  Please don't be too surprised when we have losing weeks!

Also you shouldn't be surprised if we have another good week.

Unless of course you have more information than we do.

[romerun]

celeste has gone for good ?

[RationalSpeculator]

Whats to stop "investing gamblers" from taking on a high risk with a 1% edge instead of gambling with -1%?

I think the thing that stops them is that investing is too passive.  Gamblers want to pick their odds and click the bet button.  They don't want to deposit, invest, and wait for someone else to place the bet for them.

To take an extreme example, what happens if 10 people are invested in the website with 10 btc each (100 btc is physically on the website), and claim to have 990 in reserve bringing the websites total bankroll to 10,000. A winning max bet of 100 will wipe out all the investors reserves and the websites bankroll is now bust.

People who risk too high a percentage of their online coins will find that they very quickly go bust.  People will realise this and either not do it, or quickly stop doing it once they realise that it's not working out for them.

The way I see it:
1) Real investors will have their % of bankroll diluted by people gambling on investing.
2) Real investors will either accept this lower return, increase their own risk, or withdraw.
3) Overall the website's bankroll will probably be under more risk.
4) Wagered volume decreasing as people gamble on the investing side with a 1% edge instead of a -1% edge.

I'll try to counter those 4 points.

1) Big investors who currently can't justify keeping 99% of their coins inactive in my cold wallet "just because" will now be able to invest in the site.  This will dilute existing investors, but increase max profit, attracting bigger players.
2) Increasing risk is dangerous, and probably ruinous if you don't really have the offline coins you claim to have.  Investors withdrawing balances extra investment from point 1.  A balance will be reached.
3) Currently 1% of the online bankroll is under risk per roll.  The whole point of this proposed change is to allow that to be increased by having investors be able to keep control of some part of their investment that isn't actively needed (yet).
4) I don't know if that would actually happen.  Gamblers don't appear to be able to stay invested for long.  It's too passive and boring for them.  It lacks the buzz you get from clicking the 'bet' button.

note: its entirely possible I'm misunderstanding part of this discussion

It seems to me that you've understood it all clearly, and made some good arguments against it.  Thank you.  I'm by no means committed to taking this course, and appreciate you helping me explore the idea.

I'd like to put in another vote to implement this system. The counterparty risk is way too high to rationally justify allocating a considerable amount of ones capital to just-dice. And time is not making it any better. As the total funds increase the motivation to run away with the funds, or the chance of getting robbed, are also increasing.

What do you think dooglus?

[Otoh]

House tops 2,000 BTC  



Albeit short lived, for now...

[mechs]

Check out this post under services if you can program and are interested in making some coins:
[WTB]  An Auto-Invest/Divest Bot Script for Just-Dice - 4 BTC Bounty
https://bitcointalk.org/index.php?topic=263522.0