Topic: Just-Dice.com : now with added CLAMs : Play or Invest - page 145. (Read 454769 times)

What are you trying to say?

It was interesting watching 'percent' play last night.  He or she had massive swings, 10 wins or losses in a row betting 60 btc were quite common.  Balls of steel.  What was more interesting was watching the reaction of the investors.  Some claimed the bettor has 'found a system' and was exploiting it, or betting small then big when 'variance was in their favour' or 'they knew they'd hit a streak'.  Anyone with a passing familiarity of gambling mathematics would know how silly that talk all is.  

Other investors day traded, moving their money in or out when they felt 'percent' would hit a losing streak.  This is nonsense too of course, as no one can predict such a streak and the only rational action is to stay invested.  'percent' of course ended up losing a few thousand bitcoins, as compulsive gamblers usually do.  

As an investor big swings down when players get lucky have been painful, but it's also a given (assuming JD is not a crooked site, by fault or design) that there will be big swings upwards.  The laws of mathematics haven't been violated.  There is a 1% house edge.  Casinos the world over run French roulette wheels with a 1.5% house edge and they're not closing their doors in bankruptcy.

Watching 'percent' play was interesting.  Part of me was glad to see bitcoins returning to the site.  Part of me was amazed that someone just dropped the equivalent of US$300k.  That's a 30 year mortgage in many parts of the world done in a couple of hours.  

Maybe relevant.

This was the assumption implicit in this statement; but considering that the max bet was ~75 before he was playing and up to this point he's been betting 50, 100, 200, 300; it's not the craziest assumption in the world. The only thing that might stop him from losing that much in a day is his balance on a particular account going to zero, but I watched him have a BTC6k swing yesterday before leaving up 500 so I'm not sure where the bottom of his roll is.

You're assuming he can stop.

You assume he would make all the same bets except at 4X the amount.  He would have likely made different bets and would have stopped himself before reaching 12k in a single day.

The prior earnings of any player occur independent of future earnings. I guess you're saying something like if the whale has a larger roll than the house, he can absorb enough losses to eventually bankrupt the house, but while there is always a non-zero probability of this happening regardless of house or whale roll size, I think that the odds are substantially against it as a result of the house edge. Long run, whale should always bankrupt eventually unless he can quit while he's ahead. With the lower limit today he shipped back BTC3k, there is certainly a risk of this happening. And to think that if we hadn't lowered the limit, he would have shipped back 12k .

But that limit only applies to a single bet, not the cumulative earnings of a whale.

EDIT: to clarify, as the whale keeps playing, then bankroll goes down, and so does maxbet. the limit of that sequence is still a broken bank :P

The max profit is a rolling calculation against the size of the bankroll. People divesting means lower limit.

You're right, I was wrong. 

Those are not the only reasons. Kelly criterion is about maximizing growth. Individuals (investors) however might also have a (differing) view on risk management, i.e. they are willing to accept less-than-optimal growth in trade for a lower chance to lose everything/a lot.

And another point: simply calculating the optimal value according to Kelly assumes a static system which j-d is certainly not: assume bankroll size of a whale stays the same, but as a result of normal variance, the bank is down substantially and investors get scared and start divesting. While the whale wouldn't have been able to break the bank before, he might be able to do so now.

Both are valid reasons to stay below the Kelly value.

In a totally perfect world I agree that the Kelly criterion should be followed. However, there are two things that need to be taken into consideration in this situation.

-The possibility of some flaw in the RNG.
-The possibility of Nakowa cheating.

If there was a way for us to 100% know neither of those were happening, the Kelly criterion should obviously be followed. FreeMoney explains this very well in his last post.

Unfortunately there is no way for us to be sure that there is no flaw or no cheating going on. Having a max bet lower than the Kelly criterion gains the investors equity because nakowa will not be able to win as much by cheating before it statistically becomes obvious that he is doing it.

If you think the chance of cheating/flaw are >0% the optimal max bet will be somewhat lower than what the Kelly criterion dictates. The chance of something bad going on would need to be pretty high for the max bet to be adjusted as low as 0.25%, but I think a good case could be made for it to be somewhere around 0.6-0.7%.  

How does (1*0.505-0.495)/1 = 0.25? I get 0.01, which is where it was before.

I assume you stopped reading at the heading, "Reasons to bet less than Kelly".

There is an article in a swiss newspaper:

http://www.20min.ch/ro/multimedia/stories/story/Un-joueur-decroche-le-jackpot-en-Bitcoin-22694555

Freezing new investments is dumb.  Just because the current investors lost money, there is NO REASON you should stop new investors.  An investment is just that, an INVESTMENT.  It can RISE OR FALL.  If you invest any lose, you dont get to bitch and say that no one else gets to invest until you win it back.

+1 most investors are falling for the same fallacy that gamblers do.

Meh this is terrible. I as an investor am very dissapointed. Kelly's criterion says that we should allow up to ~25% of invested be risked so 1% was already very conservative. The wales wont like it and us investors will get a lower ROI.

Too bad so few people understand game theory and risk. The site was doing fine without the lowering the amount, cmon it was just getting started and even though profits went down, 30k BTC was invested while the site was just getting started.

Hope the site owners comes back to their senses after having read this:
http://en.wikipedia.org/wiki/Kelly_criterion

Interesting. Thanks for posting this.

But I wonder, isn't reducing max betsize only a moderately effective solution? Say *max bet size* is reduced to a tenth of its previous value, but as a result *number of bets increases tenfold* (not completely unlikely). Then, according to your formula, return volatility is reduced only by a factor of about 3 (/10, *sqrt(10)).

Paradoxically I don't like the max bet reduction and now I'm sending dooglus more money (but not 4x more).

People are never going to agree on what the fraction should be, but if it is higher then anyone can reduce their exposure, but increasing it is harder and requires more trust.

I think a good solution would be to have three places for money, balance, 'ready to invest' and 'invested'. Then let users select a fraction of 'ready to invest' + 'invested' which should always be in 'invested'. It needs to be more intuitive than that I think, essentially it boils down to letting investors manage their own max bet by automatically shuffling money in and out.

With a system like this a 10% max bet (of 'invested' of course) would be reasonable. To get the old way players would just chose 10% of their 'ready to invest' money to be invested, to get the current, 2.5%.