Hey, all exchanges for now under maintenance due to upcoming wallet upgrade. Team will try to finish it as soon as possible, stay turned.

for what algo?

Yes, you are not alone 
Your 'stealth' rig contain nvidia cards and mining eth?

Does simpleminer support miners with both AMD and Nvidia cards yet? I wasted $20 on this a long time ago and it was nothing but headaches.

I guess it depends on what your mining habit/goal is and whether you are more into micro-managing/risk taking. If all is just for simplicity and just like NiceHash, where all your coins are converted to a few currency of your choice, then I believe the above setting would be a good starting point, using 24h Actual will provide perhaps most stable coin-flow of course, the statistics from the pool gets even more accurate especially the pool itself has found many blocks across the coins it supports (but whether that's good for decentralization, is another matter lol).

However, if you are a coin collector/speculative miner and you don't mind the initial setup process + a bit more of ongoing micromanagement, then definitely use Yiimp pools other than the ones in the predefined online services in conjunction with @soothaa's plugin (https://bitcointalk.org/index.php?topic=2979494.0) and setup your own pool groups for profit switching. I have detailed reason to stay away from pre-defined pools in posts in the previous page if you wish to go down this track, I do not know if that's going to be fixed or it is remain to be working as intended.

Just remember if you are the latter type, you need to make sure either you run nodes/wallets yourself or you are able to find exchanges that can accept direct mining to exchange wallet from pool. and a few points to look out, most importantly, if PPLNS/PPLNT pools, you need to check the pool's average block extraction time vs PPLNS/PPLNT window, if the window is > time to find block, no problems, you will get credited for most if not all the shares you submitted. Otherwise, you'd be better mining on pools with PROP, RBPPS type of payout scheme even if fees are 1 percent or 2 higher. Note, most Yiimp pools are PROP, Mining-Dutch is PROP (if you setup miner there, watch out for the port, use vardiff normal diff port, not the high diff port), cryptoknight.cc is PROP (but most of your GPU won't need to mine cryptonight coins for extended time as they yield less than AMD counterparts). Minerpool gives lots of ethash PoW coins and uses PPSW (dunno what that is, I assume a form of PPS), good to setup there since your cards won't mine ETH/ETC with much profit and the 2 3GB cards are troublesome too with current epoch > 170...even with Claymore's -eres workaround. However, other coins like UBIQ, PIRL, ELLA can still be profitable at times. Also minerpool have good hashrate so finding blocks is regular thing. I haven't found good PROP Equihash pool...so staying with AikaPool, they are PROP but hashrate is low on many coins (though much better than Pool.Sexy) so it could be a few days to find block on unpopular coins there, same goes for Equipool.1ds.us, which is also PROP.

Regardless what you use, just remember to use AM's rule to avoid getting caught by sudden spikes in profitability in all cases so minimal time are wasted mining temporary super profitable coins that aren't profitable at all.

That's complicated lol, luckily, Patrike already is laying the foundation by collecting hourly profit data from the Yiimp pools, current form probably only eases out spikes, but surely it can be expanded based on the same dataset (not sure needed more regular collection interval though) and implement something similar to Nemos or the likes. Give it time, I'm sure Patrike can comeup with something surprisingly good (already, there's almost no doubt he's working on it! IMHO)

The difficult thing to achieve in your case is that you wants to ride the difficulty dips, but the profit reported by the pools have a majority factor based on already achieved luck + difficulty which gets oversimplified before it gets fed into AM as just a simple profit/hashpower figure. so agreed with you that a workaround like you suggested could possibly be in place until something awesome could be implemented