Do stale shares get paid?
No, on any pool and under any reward scheme. Why late work cannot pay, how stale differs from rejected, and how to keep your stale rate under 1 percent.
Are stale shares paid?
No. A stale share is not paid on any pool, under any reward scheme. No pool chose this as a policy.
The work was real, and your hardware did the arithmetic correctly. It simply arrived after its answer stopped being worth anything. No payout rule can change that, because there is no reward left to share out for it.
Why can late work not pay?
Every share answers one specific question: the block the network is building right now. When anyone in the world finds that block, the question is retired. Every miner on every pool is handed a new one.
If your answer is still crossing the internet at that moment, it lands too late. It is a correct answer to a puzzle nobody is solving any more.
So a stale share is not thrown out for being wrong, or to be strict. Its block already has a winner, so it can no longer help find one. A pool can only pay out of the blocks it finds.
How do stale, rejected and orphaned differ?
People use these three words as if they were one. They are different events, and only one of them is normal.
| Term | What actually happened | Paid? |
|---|---|---|
| Stale share | Valid work that reached the pool after the network had moved to the next block | No |
| Rejected share | Work that failed validation: wrong algorithm, below your assigned difficulty, malformed, or a duplicate | No |
| Orphaned block | A whole block the pool found that lost the race to another block at the same height | No, that block earns nothing |
The difference matters when you diagnose a problem. A trickle of stale shares needs no action.
Rejected shares in any quantity mean something in your setup is wrong and should be fixed. An orphaned block is luck, and nothing you do on a rig can influence it.
Does the reward scheme change the answer?
No. This catches people out, because PPS (pay per share) is often described as paying for every share you submit. The accurate wording is every share the pool accepts.
A stale share is not accepted, so it falls outside the deal. The same is true under PPLNS, proportional and solo mining.
No scheme can pay for work that cannot help find a block. That money would have to come from somewhere, and the only place available is the other miners.
What the scheme does change is how your accepted work is paid. It can be paid smoothly, in proportion to a round, or all at once when you find a block yourself.
That choice is worth understanding, and the reward schemes article in this Help Center covers it. It has no bearing on late shares.
A pool that says it pays for stale shares is paying you with other miners' money, or counting something else and calling it a share.
What stale rate is normal?
Under 1 percent is healthy on a well-chosen server region. Nothing reaches zero on any pool. Light through fibre is not instant, so some of your work is always in flight when the block changes.
Watch the trend, not the count. A steady low figure is the cost of being on the internet. A figure that climbs week after week means something changed at your end.
A hidden stale share costs you exactly what a reported one costs. A pool that folds late work into a padded hashrate figure has not made it pay. It has only made it invisible.
So comparing pools by dashboard hashrate rewards whoever rounds up the most. The comparison that holds up is what lands in your wallet over weeks.
How do you lower your stale rate?
- Mine the closest region. This is by far the biggest lever. The round trip between your rig and the pool sets how wide the window is in which work goes stale. Run the latency test instead of guessing from the map.
- Use a wired connection. Wi-Fi adds latency and, worse, jitter (delay that keeps changing). Jitter causes bursts of stale shares rather than a steady low rate.
- Back off an aggressive overclock. An unstable core or memory clock makes the miner stutter and re-submit. That shows up as stale shares long before it shows up as a crash.
- Find the one bad worker. A single rig on a failing cable or riser can drag your whole account average up. The per-worker view shows which one, so you fix one machine instead of tuning five.
- Check the path, not just the rig. A consumer router under heavy load and an ISP that reroutes at peak hours both add latency. They do it at the hours you are least likely to be watching.
What if every rig goes stale at once?
Sometimes the path between you and the region degrades for reasons neither end controls. Typical causes are an ISP reroute, congestion at an exchange point or a cable maintenance window.
The sign is a jump across all your rigs at once. One rig drifting away from the others is a different problem.
Re-run the latency test before you touch any hardware. If another region has become closer in practice, move to it, and move back later.