I have some bad news for MrPlus... the profitability averaging function which attempts to blend 24h and instantaneous profitability does not appear to confer any advantage over NiceHash and, indeed, performs about 25% worse than it.
The predicted per-day earnings displayed right before the list of algos is quite accurate after reaching 100% trust level - it says I will make about 0.666 mBTC per day and that is, indeed, what I have been making the last 3 days (more or less). If I take the average per-day earnings of the top 6 algos, however, I should be making 0.865 mBTC, which is similar to what NiceHash makes per day on this rig.
Here's a recent capture of the relevant data as an example:
Trust Level 100% [ 00 days 12:31 ]
Average BTC/H BTC = 0.00002775 | mBTC = 0.028
Average BTC/D BTC = 0.00066595 | mBTC = 0.666
Estimated Pay Date 2/27/2018 12:00:32 AM > 0.0025 BTC
+++++
Type: NVIDIA
Miner Algorithm Speed mBTC/Day BTC/Day USD/Day BTC/GH/Day Pool
----- --------- ----- -------- ------- ------- ---------- ----
ccminerpolytimos X17 35.66 MH/s 1.044 0.00104 10.062 0.02928 zpoolplus-
CcminerAlexis78cuda9 Xevan 12.94 MH/s 0.982 0.00098 9.465 0.07590 zpoolplus-
PalginHSR NeoScrypt 3.86 MH/s 0.927 0.00093 8.936 0.24025 zpoolplus-
ccminerpolytimos Lyra2RE2 149.93 MH/s 0.795 0.00079 7.658 0.00530 zpoolplus-
ccminerpolytimos Nist5 174.32 MH/s 0.738 0.00074 7.115 0.00424 zpoolplus-
DSTM Equihash 1.78 KH/s 0.704 0.00070 6.789 395.58398 zpoolplus-
Again, this has been consistent for several days now, using either 1.3.1 or 1.3.2 (which differ only in the Neoscrypt miner, AFAIK). I should also note that NemosMiner 2.4.2 - which only looks at instantaneous profitability - did outperform NiceHash by about 10%, but I am going to test that again just to make sure.
Earning Tracker has memory and reads the past. So is accurate if your mining capability are close to constant.
The estimates in the algo list however, have no memory. These are instant estimations provided by the pool based on CURRENT total pool hashrate and CURRENT coin value. These cannot be accurate in anyway because 1) the total pool hashrate is not constant and 2) the coin value is not constant and the one we should use is what it will be 10 to 12 hours later which off course... no one could really guess.
We use estimations to take decisions TRYING TO predict the future.
We use past data to tell how many BTC/D we've been making.
The estimation bit does not apply to NiceHash as they pay the instant price depending on what bidders are willing to pay. Meaning that on NH, there is no 10 to 12 hours problem to solve. NiceHash do not provide estimate, they tell you how you've been doing and project to tell you when you get paid the same way the Earning Tracker does.
I have not tried NH for a while now. So cannot tell if NH gives higher profits than other pools right now. However, I can tell even if it is now. It won't in a couple days.
Lastly. Yes, you will find find short periods of time where following current_estimate will give better results than Plus. But in the long term... Plus shows better results. (At least for my rigs.)
The Plus logic is not perfect and i'm happy to ear these constructive feedbacks.
Keep it coming. They might help find the magic logic one day