Is the algorithm set up to not switch until a block has been mined? Until it reaches a certain % over the usual? How does it work?
I noticed a lot of rounds that found 0 blocks, if its switching before a block is finished often, that's just wasted hash rate.
My assumption is that this was just some hiccups, but can you verify?
Can I get a response to these questions?
I see an awful lot of rounds that are highlighted in green and/or have no payout listed. That combined with a very low payout so far have me second guessing.
We check all coins profitability every few minutes, we have a base minimum time per round that we hit no matter what (because there is a small cost of lost mining time every switch), after that minimum time there are various rules to determine whether we should switch. Generally speaking, one coin has to be a decent amount more profitable than the current, to switch within X amount of time.
From time to time, we get unlucky and get no blocks but that's just the card we were dealt, so to speak
We can always increase the "base minimum time per round" so if that's something that will increase payouts we can do it.
At the absolute minimum you should never stop mining until you hit a block (exceptions at the end).
Think of it this way: Coin A takes an average of 10 minutes per block and at current trading rates is 200% more profitable than some set coin, such as litecoin.
Coin B takes an average of 20 minutes per block and is currently trading at rates of 150% that same set coin.
So you start mining A. 5 minutes in, coin B becomes more profitable.
I propose to continue mining until you get a block of coin A. If the average is 10 minutes and the minimum is 0 minutes, just about 100% of the blocks will fall between 10 and 20 minutes (yes I know the math is more complicated than that). 99.7% of the data will fall within 3 standard deviations above/below the average, so we can infer the standard deviation is at most 3.33 minutes. This also means that 95% of the blocks mined will fall between 3.33 and 16.66 minutes. Realistically, the normal distribution will be much closer to the average.
Worse case scenario, is that it takes 20 minutes to mine a block. If it takes 20 minutes, we spend 15 minutes mining a 'less profitable' coin. How profitable does the new 'more profitable' coin have to be to outweigh this difference? Well, assuming that we will mine a block of the more profitable coin at the average amount of time (20 minutes), we have the following equation:
200% * (15 minutes / 15 minutes for a block) = B% * (15 minutes / 20 minutes for a block)
B% = 266.667%. B must be 266.667% more profitable than our coin setpoint.
However, 95% of the time, this block will mine in 16.66 minutes or less. The equation then works out to be B% = 344.82%. In this case, B needs to be almost twice as profitable as A to expect rewards from switching.
This is assuming a huge amount of variation in block size, which is unlikely. For blocks that take an average of 10 minutes, my guess is that something like 95% will fall between 6 and 14 minutes.
If you set something like litecoin or bitcoin as a setpoint, your profitability calculations to determine when to switch and when to stay should be as follows:
(current time spent mining A + average time for block of B / average time required for block A) * profitability vs. setpoint
> average time for block B * profitability vs. setpoint.
The moment that > becomes < is when you should switch, even if time has been spent on coin A.
This is because on average, we should be looking at expected blocks of A vs. profitability of A as compared to B, given a switch, assuming average expected blocks of B. In fact, when we do this, we find the profitability requirements to switch are much, much higher.
Of course, once a block is mined, it would be time to switch to another coin.
I don't think that fruitless rounds are a significant problem at the moment but I think you raise an interesting general suggestion that is worth considering.
Make the check for profitability, and change if appropriate, upon the discovery of a block or after a longer period of time.
I believe this pool has a great framework with unlocked potential profit that can hopefully be found by tinkering with the rules and carrying out some structured testing.
Show Posts
