Pay Per Last N Shares

Unlike a simple division of one round, PPLNS tracks recent shares regardless of the previous discovery. At the next block, the same share may still be in the window and receive another allocation. Conversely, a share can leave the window without a discovery and receive nothing. This is not an automatic payment for every accepted share as in PPS. [Rosenfeld — PPLNS windows and scoring myths]

The simplest model uses the last N equally difficult shares of the entire pool. With variable difficulty, comparable work must be represented rather than blindly counting messages. Methods may use weighted units or multiples of network difficulty. N alone therefore specifies neither hours, worker count nor a fixed quantity of transferred data. [Rosenfeld — PPLNS windows and scoring myths]

In a simple full window of N=1000 equally weighted shares, a miner owns 100. If 3 BTC is available for distribution after the pool fee, the miner receives 100/1000 × 3 = 0.3 BTC. The example assumes this fund and accepted shares; it says neither that the subsidy is 3 BTC nor that this sum arrives every day. The next discovery may have a different window and different transaction fees. [Rosenfeld — PPLNS windows and scoring myths]

Increasing the window spreads the valuation of work over more future discoveries. It reduces the variance component associated with a short window but extends the time during which work can still participate in possible rewards. It does not eliminate the luck of the pool as a whole. A displayed expected return is therefore not already a fixed enforceable claim for all unrewarded shares. [Rosenfeld — PPLNS windows and scoring myths]

After joining, your part of the window builds up; after leaving, past work remains temporarily until newer work displaces it. In a correctly designed method, its aging does not depend on whether you keep mining. Intermittent operation can have greater variation but is not automatically penalized by losing each share’s expected value. [Rosenfeld — PPLNS windows and scoring myths]

PPLNS limits the advantage of joining at a round’s start because its window crosses round boundaries. Rosenfeld nevertheless distinguishes simple fixed N from unit-PPLNS: a naive version may not resist changes in difficulty or block value. Parameter changes should preserve existing expected entitlements. The name alone does not guarantee fairness in every implementation. [Rosenfeld — PPLNS windows and scoring myths]

The distributable fund depends on actual accepted blocks and the published handling of transaction fees. F2Pool explicitly describes PPLNS fluctuations from discoveries and fee revenue. FPPS instead prices work using an estimate; PPS+ may use PPLNS only for fees. A service deduction percentage is insufficient for comparison without knowing this basis. [F2Pool — Payout schemes] [F2Pool — PPLNS reward fluctuations]

Check block acceptance, allocation calculation, credit to the balance and actual withdrawal under the schedule and threshold. Switching methods may take effect only in the next accounting period; F2Pool, for example, publishes its own changeover timing. Compare work before and after the change, including the remaining window. PPLNS removes neither pool payment risk nor the miner’s operating costs. [F2Pool — PPLNS reward fluctuations] [F2Pool — Changing payout schemes]