Using a 60 minute average of profitability would be better than relying on either the instantaneous value when polled or the strange attempt to average the 24 hour and instantaneous data that NemosMiner and others are pursuing. However, MinerControl appears to have been abandoned several years ago while MrPlus has been working on a fork of NemosMiner* right now and so seems to be receptive to implementing good ideas provided by the community so I am giving it a shot now.
* -
https://bitcointalk.org/index.php?topic=2965976.0There is another factor that I think I have picked up that contributes to the loss of income with algo switching on zpool and it relates to the delay between when a new hash is solved and when it becomes "immature". If there are several "new" hashes and the miner switches way to a new algo you lose some of the income from those shares when they switch to immature. This issue is easily observable. simply mine a coin like x17 and when you see 4 or 5 new shares sitting in the pool stop your miner and see how many of those hashes you get paid for.