Quote from: philipma1957 on March 31, 2016, 03:43:31 PM

Quote from: tlhIlwI on March 31, 2016, 03:40:17 PM

It will be interesting to see how this works out since we have different strategies.

I'm now 99.2% (0.8% loss) on the rentals with 53.52% of rental shifts paid 5+ rewards.

Now, that said, my rentals did turn off 11 hours ago due to high prices (right about when the luck began here, actually). They are gradually falling off the N range. Time will tell which side of the variance this gap falls on, since it is the future luck that will determine if it is good or bad to not be making up for the missing rental hash right now (it is equally possible for me to be lucky to not be renting during those shifts if we have a series of long blocks as it is possible for me to be unlucky and missing out on a continued block party).

yes and this is why your method is failing. the pool gets hot and lots of people jump in bump you out of the hot streak.

I pay a slight premium on purpose for this pool only when I rent hash for mining here.

The "hot streak" (positive variance) pays the shares I mined in the past. That is why I recovered a 20% loss in one day from the shares submitted by rentals when we weren't in a "hot streak". I can not guarantee that will continue, so I have an equal chance of coming out ahead as I do falling behind by not renting when the prices are high. Unless the quality of hash is influencing pool luck (I doubt it is in any significant way) then randomness is randomness and we should have an equal number of profitable and unprofitable shifts in the long run. The only difference should be in how much was paid to obtain those rewards.

I'm open to being proven wrong, but if what you say is provably true, then the shifts I choose to rent in are guaranteed to bring a loss and the best strategy (even better than tracking the N range) would be to rent the exact opposite shifts only since they are the ones not guaranteed to result in a loss (in other words, only rent when the price is high). Variance doesn't (or shouldn't) work that way.

The only caveat is this-- due to diff changes rewards cost more to obtain in the future than they cost today. If my method gets positive variance then that is to my advantage, but the negative variance is harder to recover from.

That said, I would agree that this pool has an excellent record of 107% in the long run. That can't be guaranteed to continue, but it is reasonably low risk to put in orders up to 0% (and maybe even +1%) instead of limiting to -2% as I have. Above that seems very risky though. That doesn't change the fact that my method should work just as well but with more variance in the rewards (and at less expense).

....and another block just came in so all of the numbers I typed are invalid already...