A mining pool is a service that combines hashrate from many miners into one search effort and splits the proceeds. A machine mining alone with 200 terahashes per second against a network of roughly 930 exahashes per second would expect a block about once every 88 years. The same machine in a pool earns a small amount every day.
Nothing about pooling changes the total reward the network pays. Pools change its distribution over time, converting a high-variance jackpot into a wage. That service is what the pool fee buys.
How shares work
Miners prove their contribution using shares. The pool sets a target far easier than the network's, and every hash a miner finds below that easier target is submitted as a share. Shares have no value to the network; they are statistical evidence of how much work a miner performed.
Because hashing outcomes are independent and uniformly distributed, share counts are an unbiased estimator of contributed hashrate. A miner producing twice as many shares as another contributed roughly twice the work, and occasionally one of those shares also clears the network target, which is the block.
Payout schemes
The schemes differ in one respect: whether the pool or the miner absorbs the luck of when blocks arrive.
| Scheme | Miner is paid | Who carries the variance | Typical fee |
|---|---|---|---|
| PPS | For every share, at a fixed rate | The pool | Higher |
| FPPS | For every share, including an average fee component | The pool | Higher |
| PPLNS | From blocks found, over the last N shares | The miner | Lower |
| Solo via pool | Only when their own work finds a block | Entirely the miner | Lowest |
Pay-per-share pays a fixed amount per share whether or not the pool finds a block that day, so the pool bears the risk of a dry spell and charges for it. Full pay-per-share extends this to transaction fees, paying miners an average fee component rather than only the subsidy, which matters more when fee revenue is volatile.
Pay-per-last-N-shares pays out of blocks the pool actually finds, distributed across the most recent N shares. Fees are lower because the pool carries no risk, and payouts are lumpier. PPLNS also penalizes hopping between pools, since a miner who leaves forfeits their position in the share window.
Published fees on major pools sit broadly between zero and four percent depending on scheme, with pay-per-share variants at the higher end. A pool advertising a lower fee on PPLNS is not necessarily cheaper in expectation; it is selling a different risk profile.
Stratum, the protocol underneath
Stratum is the protocol connecting a miner to a pool. Under Stratum V1, the pool constructs the block template, choosing which transactions are included, and sends the miner a job to hash. The miner supplies hashrate and has no say in the block's contents.
That design gives the pool operator, rather than the thousands of machines behind it, control over transaction selection. Stratum V2 changes it by letting individual miners construct their own templates and negotiate them with the pool, alongside encryption and efficiency improvements.
Adoption moved in May 2026, when pools and entities representing roughly 75 percent of global hashrate joined a working group on the open standard for block template construction. A separate protocol operated by one pool has offered similar template choice since earlier. The change does not reduce how concentrated hashrate is; it separates the question of who supplies hashrate from the question of who chooses transactions.
Fees, thresholds and what actually reaches your wallet
The advertised fee is one of four things standing between a block reward and a miner's wallet. The second is the payout threshold, the minimum balance that must accumulate before the pool sends anything, which on a small operation can mean weeks of exposure to the pool holding your earnings. The third is the transaction cost of the payout itself, which some pools pass on. The fourth is stale shares, work submitted too late to count.
Stale rates are the one a miner can influence. A share submitted after the network has already moved on earns nothing, and the rate at which that happens depends largely on network latency between the machine and the pool server. Choosing a geographically close server usually matters more to net revenue than the difference between a one percent and a two percent fee.
Payout scheme also interacts with machine count in a way that is easy to miss. A single machine on PPLNS experiences noticeably lumpier income than fifty machines on the same scheme, because the pool pays from blocks found and a small contributor's share of any given block is tiny. Operations at small scale generally value the smoothing that pay-per-share provides more than the fee saving PPLNS offers.
Concentration
Pool concentration is the most persistent structural criticism of Bitcoin mining. Over the month to 8 September 2026, the largest pool accounted for roughly 24.6 percent of blocks found and the second for about 18.9 percent, giving the top two around 43.5 percent between them, according to mempool.space. The top five accounted for approximately 76.3 percent.
Those percentages describe where hashrate is directed, not who owns it. A pool with a quarter of blocks is coordinating machines owned by thousands of independent parties who can redirect them within minutes. That mobility is a real constraint on pool behavior and the reason concentration figures overstate the control any operator holds, though it depends on miners noticing and acting.
Choosing a pool
- Payout scheme. Whether you want the pool to absorb variance and pay for it, or to take the variance yourself at a lower fee.
- Fee, including what it covers. Whether transaction fees are shared, and how the pool treats them.
- Payout threshold and cadence. How much must accumulate before a payout, and how often payouts run.
- Server proximity. Latency affects stale share rates, which is a direct cost to revenue.
- Template control. Whether the pool supports miner-constructed block templates.
- Transparency. Whether the pool publishes blocks found and lets you reconcile your share of them.
The comparison worth making is expected revenue net of fees at your machine's hashrate, not the advertised fee alone. A one percent fee difference is small next to a persistent stale share rate caused by connecting to a distant server.