Patrike,
Anyway we can do some brainstorming together on a way to improve the Profit Switching mode of the software?
If i understand NemoMiner, basically its logging the 24hr est, 24hr actual, and Current est, then its doing a calculation to figure out a running offset.. then applying it to smooth out the outlier spikes in profit that result in poor actual pay.
Example:
x17:
Current: 0.01523 *which is based on current sharerate and exchange price
24hr Est: 0.01853 *which is a projection of the next 24hrs based on the current exchange rate
24hr Actual: 0.01544 *which is based on actual payouts from past 24hrs of exchanged coins
Trust:
(3/26/2018 0500EST 24hr Est:) - (3/27/2018 0500EST 24hr actual) = Trust 24hr Offset
Example:
(0.01853)-(0.01544)=0.00309 Difference of 16.7%
So on the profit switching list,
it will compare the current price with a running offset to determine if its worth switching to, because if the current est... greater than the 16.7% current offset, than its more likely it will actually payout more.
Then the next day lets say the new next 24hr trust offset is 17.0%, it does an averaging
16.7 and 17.0 for an average running offset of 16.85% and so far as time of the software collects data.
One thing i noticed on nemo's is if the offset places the coin below the 24hr actual, it defaults to 24hr actual price on the list for profit switching.
Basically, the trust offset in nemo miner is just saying hey, this is the expected amount of drop in value on this algo after all the peeps switch over to catch the spikes in high current payout and its actually exchanged for payout to peeps. The part i dont know how it does it is how it is taking into account current prices to this trust offset to mix those values into the calculations, thus why i suggest some brainstorming by the masses to figure out a good way of doing this.
Thanks for your feedback. I'm very much open for a discussion on this topic.
Awesome Miner already have access to these numbers, so making some adjustments to the profit switcher should be possible. Please give me a little more time to investigate before I get back with a longer response.
What I had in mind since before was that the Awesome Miner cloud could help out here to build this statistics trend, so you don't have to run Awesome Miner for a longer period of time to build the trust-level. It could be provided for you.
I've tested and used for several days Nemos Miner 3 and I do think that:
1) Nemos Miner has a lot more miners (10 ccminer forks for various miners and cuda versions to maximize hashrate+ high prfitability RVN x16r mining) and to select the best miner for each algo it will benchmark them all and use the best one with highest hash-rate (for 300 seconds and bench-marking is done in in mining mode, so earnings are nt lost, and important is that mining begins after benchmark is complete, and this is very useful when compared to AM that doesn't have this feature).
2) Nemos Miner has a better algorithm switching implementation using BrainPlus and EarningTracker with past statistics
Patrike, I recommend you to check their sourcecode and statistics files from /statistics folder to improve performance.
I started on the same pool both Awesome MIner and Nemos Miner and after 48 hours Nemos miner had a better real BTC payout than AM (never trust inflated statistics, always compare real BTC payments).