I've been watching your posts here and there is a very simple approach to all of this...all pools will find the expected number of blocks (100%) over time (unless there is block withholding going on) as luck will always even out. The real difference between pools is transaction and pool fees. For example, antpoo does not give the transaction fees to the miners so you would lose big time there. Pools like KanoPool and Slush both give transaction fees back to the miners so the difference is 0.9% pool fees here vs 2.0% at slush. The math is painfully simple and test runs are not necessary as diff changes effects everyone's payouts ...if you want to get the most out of your hashrate, point all of your machines here.
I fully agree with this. The ONLY way you can accurately compare earnings between pools is to have a small pool of miners (say a dozen or more) running on one pool, and another set concurrently running on a 2nd pool. Even then, you would have to work on balancing the pool of miners so their total hashrates were below your error acceptance level. E.g. If your interested in a 0.1% difference in earning, the two pools of miners would need to be balance to within 0.05% worst case. 0.01% would be a lot better. Then you would need to run them for several hundred blocks worth, with whichever pool does the least blocks/month being the governer.
Just run the math... for instance, I have some 841s that only pull 13.3TH, others pull 13.6TH. If I just used one of each of those, one per pool, there would be a built in bias of over 2.25% in earnings between the two - independant of which pools I pointed them at.
If you take one miner and point at one pool for a few days or weeks, then use that same miner to point at another pool for a few days or week, your results are going to be corrupted by both the likely difficutly change that occured during the test period, and the dumb luck the pools had for that short period of time.
The only thing that holds true is what the quoted author states.