I just discovered that I made an error verifying the +EV play.

It isn't a 30% edge for the player in the worst case.

It's a 32.26% edge for the player.

I just confirmed that the site in question does offer +EV bets for the player. In the worst case the edge is 30% for the player.

I fully expect the site to fix this when subSTRATA tells them about it.

Right, and he also said it doesn't involve a jackpot. Sometimes jackpots can be +EV for the player - maybe it's being paid out of the site's pocket as a promotion, or maybe it's a progressive pot that fills over time from other players making losing bets.

But if there's no flaw and no jackpot then I don't see where the +EV is going to come from.

No, it's worse than that. They missed the deadline and didn't post any update at all.

"We're working on the databases, this function will be available soon."

It's not impossible at all. DaDice for instance has a jackpot that pays out 20 million satoshis for a 1 satoshi bet, and hits about 1 in 4 million rolls. That's massively +EV (except that the jackpot isn't at all provably fair, and it is really easy for them to cheat).

subSTRATA, if you want to PM me the details I'd be happy to check it out and confirm or deny. If you did find a problem I hope you would offer it to the site itself for a reasonable price first.

Edit:
But +EV is really all you need. Play long enough, apply proper bankroll management and you will win.

(Assuming of course that it really *is* +EV).

There's always luck involved and so it's possible to have short term losses. Think of counting cards at blackjack for instance. It's possible to have long losing sessions, but the odds are in your favour if you do it right, and so you will eventually end up ahead.

Edit2: Lots of people are saying it's not possible to find a +EV strategy. I wouldn't be so sure. There can be all kinds of problems with the implementation of a dice game. PRC dice used to roll high numbers more often than low numbers due to a bug, and so it was possible to have a +EV game just by rolling 'hi' every time. I can imagine a game where your payout is rounded up to the nearest satoshi rather than down, such that a 60% bet ends up paying 2x for small bets. That would be +EV. And so on.

+EV means "positive expected value", and basically just means that the odds are in your favour rather than against you.

I believe he also has the bittrex-richie@gmail.com email address and has used that to scam people as well. He also changes his name to impersonate trusted people, and it seems to work surprisingly well for him.

When someone approaches you claiming to be me, please don't send them your money. They probably aren't me. Have them log in to the (1) account at Just-Dice if you have any doubt. And as far as I know the real bittrex-richie doesn't hang around the JD chat offering to promote your favourite altcoin on his exchange if you pay him some Bitcoins. Always use a trusted escrow when trading.

I think that's the wrong message to take out of it. Bitcoin is valuable. Scammers are going to try to steal it, like they will try to steal anything else of value.

The thing to learn here is to be careful who you trust. If someone isn't willing to prove that they are mining, and cannot demonstrate that they are solvent then it's really not a good idea to trust them or to trust that they aren't insolvent.

There was a guy in the JD chat yesterday who couldn't ever get a single connection to a CLAM node, although he had managed to sync part of the Bitcoin blockchain in the past.

After a few hours of back and forth with him, it turned out that he was on a college network and couldn't make outgoing connections on most ports. Port 80 was open, and so was port 8080. If I ran a CLAM node and told it to listen for incoming connections on port 8080 (port=8080 in clam.conf), and gave him the IP address. Using an "addnode' command he was able to get a connection and finish syncing the blockchain.

Edit: I should mention that http://portquiz.net:31174/ is a good tool for diagnosing such issues. Replace 31174 by various ports until you're able to get a connection. If you see:


it means you are able to connect out on that port.

As BAC said, it's to do with the orphaning.

CLAM divides time into blocks of 16 seconds. You only get to check for opportunities to stake each 16 seconds. Let's say there are 4 of these blocks per minute (it's close).

We aim to set the difficulty such that a block is found once per minute. That means that there's about a 25% chance of a block being found by someone in any particular 16 second time slot. If you (solo staking) and JD both find a block in the same time slot, you will start trying to build on top of your block, and JD will try to build on top of its block. The rest of the network will try to build on top of which ever one of the two they heard about first.

So when we both find blocks in the same time slot, if JD has 75% of the staking weight, it has a 75% chance of building on top of its block and orphaning yours. Of course there's a chance that the JD block gets orphaned too. If you solo-stake, find a block at the same time as JD and then stake again shortly after, or one of the peers who saw your block before he saw the JD block stakes next, he'll build on top of your block not JD's.

I've attempted to do some back-of-an-envelope calculations to determine what that means it real terms, and came up with a number like 14% orphan rate I think (*). Someone posted recently saying that they tried solo-staking and only got 2 orphans out of 50 or so blocks they staked, so maybe 14% is too high.

Basically if the orphan rate inflicted by JD is higher than the commission charged by JD, then you would make more staking at JD than solo staking. But we don't have a good number for that orphan rate.

Here's a list of the blocks that JD's staking wallet found recently which ended up being orphaned. Only one on 28th, 5 on 29th, 10 on 30th, 5 on 1st. So you can see there's a lot of variance:


Edit: I've not checked, but it's quite possible that most of these orphaned blocks were orphaned by JD's hot wallet. They run completely separately, and don't "go easy" on each other. They're competing with each other for stakes just like every other wallet is.

(*) Suppose JD has 75% of the stake weight, and that there are 4 16-second blocks per minute. Suppose you are solo staking. Suppose you just staked a block. There is a 1/4 probability that someone else staked in the same 16-second period (1 in the 4 windows per minute). And there is a 3/4 probability that it was JD. So there's a 3/16 probability that JD stakes at the same time as you when you stake. To orphan your block, JD would also have to stake the next block. That's also a 3/4 probability (JD has 3/4 of the stake weight). So for each block you stake, there's a (3/16)*(3/4) = 9/64 = 14.0625% that JD will orphan it. That was how I got the 14% figure, but I don't know whether my reasoning is valid or not. I'm not taking a bunch of things into account, like the chance of you orphaning a JD block, or the change of us both finding the next block in the same time period as well.

I can't wait until they start mining again. I'm sick of litecoin confirmations being so slow since s.cc's 850 GH/s went offline.

Hey! Welcome to the site. I guess you're partially responsible for the kicking investors have taken over the last few weeks:



Here's the all-time profit chart. We're getting very close to the expected 1% profit line now:

He tore your post to pieces with a well reasoned response. And all you can say back to him is that his post is too well written and that he is too smart? That makes no sense at all. How about taking him on point by point? Is he wrong in anything he said? If so, tell him how. If not, does it really matter that his writing skills put yours to shame or that he's smarter than you? I don't think so.

Attack the message, not the messenger.

You need to multiply by 100 to get percentage numbers. 1% of 2450 is 24.5, so the ~6 per day you're seeing is about a quarter of that, or about 0.25%.

Edit: I would encourage you try keep solo-staking for another month, and post your stats at the end of each week. I would be interested to see how much variance you experience, and whether you continue to outperform JD on average. I think it would be a close call, but from the analysis I've done I would expect the ~14% orphan rate you get solo staking to overwhelm the 10% commission that JD charges. It's possible that my calculation of the orphan rate is wrong. Are you saying you only saw 2 orphans in 43 staked blocks, one of which you attribute to your own activity? If so, that's much lower than I would expect.

Unfortunately, in order to validate new blocks you need to check that the transactions in that block are legally spending the coins they spend. And those coins could have been created in any previous block. So you need the whole blockchain (or at least a complete set of unspent outputs).

Also, while it isn't strictly necessary for you to include other people's transactions in your blocks, it's pretty rude not to. And so you need the blockchain to validate that the transactions you include are valid.

The Just-Dice hot wallet is currently using 1.51 Gb of virtual memory, with a resident set size of 491 Mb,
and the staking wallet is using 1.74 Gb of virtual memory, with a resident set size of 607 Mb.

I have a clamd running on my laptop which has been running for 24 hours with a new clean wallet. It has seen no transactions of its own.
It is using 1.41 Gb of virtual memory, with a resident set size of 245 Mb.
If I restart it and wait for it to load up the (tiny) wallet, it uses 1.04 Gb of virtual memory, with a resident set size of 401 Mb.

So yeah - it's kind of hungry. I'm sure there are things we can do to reduce its memory footprint if that seems worth doing to anyone.

We don't have the concept of "max bet". We have "max profit", and gamblers are definitely making bets which would win the maximum profit if they won. So I think their (hypothetical) argument is valid: "I've been betting 0.14 CLAM at 0.0001% over and over, trying to hit the max profit. I failed so far, and now you want to cut the max profit by 80%? What's wrong, am I just about to win it? Unfair!"

Don't forget that you can play with a 990,000x payout multiplier. You can bet just 0.14 CLAM at the lowest chance of winning and win the max profit. So it's not about gambling a high amount, it's about a player getting lucky. If they won the max profit they would have a hard time cashing out into a different currency without crashing the price of CLAM, but they could invest their winnings, or just stake it, and have a very nice regular income. You can think of it as if they are playing to win a large share of JD's future earnings.


Except for my 2nd point, about making it hard for them to maintain a sensible Kelly factor without constant babysitting. But you talk about that below I guess.


The optimal strategy for investors is to risk 1% of their investment on every bet, and so that's what some of them want to do. If I only let them risk 0.1%, they need to set their offsite amount to 9 times their onsite amount to compensate. That causes their onsite amount to chance 10 times more quickly than it would without the offsite setting, and means they need to regularly adjust their offsite amount to be 9 times their new onsite amount. I've not modelled this, but I'm sure if I did we would find that a "10x offsite, 0.1% risk" investment would diverge from the 'full Kelly' exponentially more quickly than a "2x offsite, 0.5% risk" investment does.

I've never looked into how mixers work, but it seems pretty obvious that you would only allow withdrawals of particular sizes. If you want to withdraw 2,224.35x BTC then maybe you have to make 2 withdrawals of 1000 BTC, 2 of 100 BTC, 1 of 10 BTC, 4 of 1 BTC, etc. Then your plan of "look at all outputs that add up to ..." fails immediately, since all the outputs are of generic sizes.

I'm happy to escrow for you.

Actually I'm currently working on a built-in escrow feature for Just-Dice which should streamline the process.

Your trust rating here isn't great - I've no idea who PyroTekNeks is/was but Tomatocage says you are he, and that that is a bad thing.

I guess it would help people believe if you signed a message saying "I want to buy 125 BTC worth of CLAM" with an address holding that amount.

LLTGC.

It is a bit like that.

If God existed and wanted to have people believe in Him, He could prove His existence to us. Write "Hey guys, I'm God and I exist" in the stars or something.

If scrypt.cc was mining and not simply paying investors with their own deposits and wanted to prove it, they could provide proof of solvency, and show us the blocks they are mining. Write "Hey guys, I'm Scrypy.cc and I'm mining" in all the coinbases they mine or something.

Neither chooses to do so. That doesn't prove that they aren't real, but it leaves room for doubts that could easily be avoided.

The article didn't really say what the exploit was:



The first quote makes it sound like the seed generator either generated the same seed for two different accounts, or was only called once for two different accounts.

The second quote makes it sound as if the "reveal old seed" behaviour could be made to happen without the "replace old seed by new seed" action having been done first. Maybe sending too many requests at once could trigger that. But it doesn't fit in with the first quote.

So I think we're really left guess about what really happened, and why.

I think it's unlikely the same exploit exists on other sites. I don't see how multiple accounts would share the same server seed accidentally, or why a server would reveal its server seed without first invalidating that seed. But there's no guarantee that other sites didn't make the same obscure mistake(s).