I have compared some software and some pools.
Between ahashpool, zpool and hashrefinery it looks like ahashpool is still a bit more profitable, but not by how much your algo-switching software will say. The estimated profits look kind of inflated, you should expect about 2/3 of what you see there.
I tried using various algo-switching software, and the powershell ones are all very similar (nemosminer, megaminer, sniffdogminer). They might have different optimisations built in, or they might get different miners for different algorithms but they should do the same thing, and you can always configure them the way you want them by editing the ps1 and json files.
I have 3 1070s and I am using MegaMiner (which benchmarks different miners for each algorithm), for neoscrypt it should settle on ccminerKlausT, which is pretty good, but I find that I get slightly better performance using ccminer-alexis78-ms2013-cuda7.5, so I just replaced it with this one, while also having set intensity to 16.63 (which is the limit for me to not get out of memory errors, but your millage may vary)
I also found out that it switches a lot, so what you can do is add a high PercentageToSwitch parameter (I use 30%), so your algorithm will not switch unless another algorithm is
30% more profitable, so you don't lose a lot during the ramping up, I set a small interval of 200s, so it checks pretty often if the current algorithm is still no less than 30% from the highest profitable algorithm.
That done, I found that it still switched a lot because of very short spikes in some algos/coins, so I decided to only use the more profitable more often or the more stable algos, so right now I only have blake2s, neoscrypt, lyra2v2, x17 and phi.

I'm still fiddling with these settings and I'm not sure how useful profit switching is myself.
Basically by the time you get some part of the coins for your hashrate from the pool, and by the time those actually get exchanged (if using autoexchange), their value may have changed a lot, and usualy not for the better.

You can also try to use ahashpool24h, which is the same pool but with a different api that returns profitability for the last 24 hours instead of current profitability. But I did saw a post from a user saying that he used 2 identical rigs, and got a bit more from the one using current profitability.

You can also do your own tests and let us know what you find out. Maybe use 2 cards to mine a single algorithm, and 2 cards to run profit switching, and then compare revenue (you'll have to be patient and probably use 2 different wallets)