50% chance of winning = 1.98x payout of course

No, actually the expected value for the site is (starting BR + 0.01*total amount bet).

And if you really believe in Martingale I welcome you to play on JustDice and ruin the bank!

I simulated a very pessimistic scenario in which a high roller with an infinite bankroll places 10000 maxbets in a row (50% chance of winning each bet). Maxbet adjusts dynamically (to 1% of site BR) after each bet. Site's starting bankroll is 26500 btc.

See below the probability of the site's bankroll after 10000 bets (3600 iterations).



Site can't practically go broke, but BR can decrease substantially. Note the logarithmic x-axis. There is far more upward potential with this type of player....

Note that it is not the max bet that is 1% of the BR, it is the max profit that is limited. When you bet at 98% chance of winning and go for max profit, say 250btc, means that you have a 2% risk of losing 24500btc each bet! You will be ruined sooner than the bank, I guarantee you that...

Very good points above!

I have a long background in advantage gambling and it keeps fascinating me how people keep believing in Martingale and the like as winning systems. Please understand that it is bettors like Nakowa that we should cherish. Real casinos treat these type of players like gods as they bring in huge profits.

Besides having a max bet on the site's end, the risk of ruin for JustDice is infinitesimally small due to the fact that the max bet reduces as the site's bankroll reduces.

On the other hand, investors seem surprised to have lost a significant (6%?) part of their investment. This surprise may be because somewhere it was stated that investors risk max 1% of their bankroll. This may be deceptive as the risk is max 1% for each bet! In fact, it is not unlikely to lose 10-20% of your investment when a big player plays for a long time and is lucky.

No, there was a 5% chance of winning 2000 (based on some crude assumptions). The risk of ruin for the site is much much smaller, provided there is no software flaw that can be exploited.

Yes, but not as more probablw as you might think. His expected return, assuming 100 bets of 100, was 100*100*-0.01=-100. In this case it is equally likely to lose 2000 coins as it is to win 1900.

It's variance.

The total standard deviation after multiple bets is: SD = sqrt(N)*unitSD*(bet size)

He bet with a bet size between 10 and 250. Lets assume a mean bet size of 100. The unit SD is 1 in this game, so after about 100 bets the total SD is 1000. He realised a 2000 coint profit, which is a 2SD event, or probability of about 5%.

I don't know how many bets were made, but it would not surprise me if it were >100. Then the result is even more likely.

No reason to divest. It was equally probable that the house was up by another 2000 coins

This indeed:


And I do this on a separate User Account that I exclusively use for the Bitcoin client. To check my balance, I use http://forum.bitcoin.org/index.php?topic=23123.0

I think it is pretty safe. Probably not 100% safe, but I don't know a better way that is at least reasonably practical.

Wow, didn't know Google Docs was so versatile - but there was certainly some geniality involved as well.

Donated for this excellent tool!

Thanks for that!

What is the font in the advertisement? I like 

Spotlight (search function) works great but it does not normally index system folders.

On my 5870s, I get close to 400 at 900/300Mhz (see my post above). With overclocking I meant going above 900MHz and at higher voltages than AMDOverdriveCtrl allows me to. For me it seems to be not much more than 10% gain, and I doubt I want to risk my hardware for that. It would also require a BIOS flash, which may not be entirely without risk either.

So how much does OC'ing cost in terms of lifetime?  Seems to yield an extra 10% in hash rate,  is it worth it?

Just want to let you know the results of my optimization:

4x XFX 5870 1GB
LinuxCoin
Clocks: 900 / 300, standard voltage (AMDOverdriveCtrl)

phoenix.py run from command-line (not start_mining script)
kernel: phatk (15-20 MHash/sec higher than withpoclbm)

flags:
VECTORS
AGGRESSION=9 (Increasing AGGRESSION makes system slower and does not further increase hashrate)
BFI_INT
WORKSIZE=256 (10MHash/sec higher than with 128)
DEVICE:<X>  (0,1,2 or 3)

hashrate: 395 MHash/sec
temps: 63-78 ^oC (lowest to highest mounted card)

Cards are on a 890FXA-GD70 with a 1200W CoolerMaster PSU in a closed HAF X at 25 ^oC ambient temp.

If you read my post above you see that my temps are good and stable. Just want to stress this as a lot of people keep saying that quad cards in a case is impossible due to heating. I don't even need a table fan or anything. On the other hand, I did not overclock drastically, just the 900/300MHz trick with AMDOverdriveCtrl.

I won't stick with Linuxcoin either. Wouldn't say it sucks, it's very good to get going quickly but it's not very practical if you want a bit more than that.

I get better results with poclbm, seems to fluctuate more with phatk.

I read that LinuxCoin comes with SDK 2.4, and 2.1 would give better results. Anyone care to share how can I install this myself?

I have this one with 4 x 5870, max. temp = 78.5 degC.