Author

Topic: Want to check if I understood PPLNS correctly. (Read 1072 times)

full member
Activity: 231
Merit: 100
No, this is not correct.

With PPLNS (and most other reward methods), the expected value per share is always:
Code:
block reward * (1 - pool fee) / difficulty

PPLNS works roughly like this:

Whenever a block is found, your proportion of shares in the last N shares (usually in the last M shifts) is calculated. If you submitted 1% of the shares, you'll get 1% of the block reward (minus pool fees). The shares are not reset when a block is found. If the pool finds a new block after a couple of seconds, you'll get once again 1% of the block reward.

Say mining pool B has twice the hash rate of mining pool A. At pool B, you'll get only half of what you'd get in pool A when a block is found. However, pool B will find twice as many blocks as pool A. If both pools have the same fees, you'll earn the same at both given enough time.

The only aspect where "size matters" is variance. A big pool will find blocks faster. So even if a block took 6 * difficulty shares to find, it will last only a couple of hours. Small pools can go days or even weeks without finding a block. The expected earnings over time will still be the same, but the earnings in a big pool will be more steady.
full member
Activity: 224
Merit: 100
The way I understood everything with PPLNS provided your pool is able to find at least one block per N-interval, it does not matter which pool you mine on as far as results go because the value of a block is set.

Or said otherwise: The higher your hashrate is the more preferable become smaller pools because that increases the value of the shares generated by you in relation to the shares generated by the pool in total.

That said amongst the PPLNS pools the one with the lowest fee is best as long as N is big enough to guarantee at least one block find per round.

The second most improtant decision criterium then would be how many faster miners are in a pool, competing with you for the leadership in shares created per round. The less the better.

Does that seem about right?
Jump to: