How to choose a mining pool in 2026: payout models, fees and the numbers that matter

Here is a number that reframes most mining pool advice: over the week to 26 August 2026, transaction fees made up 0.78% of the total Bitcoin block reward. Over the last 4,032 blocks, 0.74%.

That matters because the argument people spend most time on — FPPS versus PPS+, which differ only in how they treat transaction fees — is currently worth less than one percent of your revenue. Meanwhile the gap between a 0% pool and a 2.5% pool is 2.5%, roughly three times larger and entirely predictable.

This guide covers what actually moves the needle, with real fee schedules for Bitcoin, Kaspa, Litecoin/Dogecoin and Ethereum Classic — the four algorithms we sell hardware for.

Payout models, briefly and precisely

For a single hosted machine, FPPS or PPS+ is almost always right — you want smooth daily income, not lottery variance. PPLNS makes sense at farm scale where the law of large numbers does the smoothing for you.

The Bitcoin pool landscape right now

Share of blocks found over the seven days to 26 August 2026:

Two things worth noticing. SpiderPool and SECPOOL are now larger than Luxor and Braiins — most "top mining pools" listicles still on page one of Google are years stale. And the top two pools control 44.8% of blocks over 90 days, the top three 59.2%. If pool centralisation concerns you, that is an argument for pointing a small miner somewhere other than Foundry or AntPool.

Bitcoin fee schedules

Foundry's 0% looks unbeatable, and for large operators it is — but onboarding is institutional and not generally open to a single-machine customer. For retail, ViaBTC's 0.0001 BTC threshold is often worth more than a lower headline fee, for reasons the next section explains.

One honesty note: several of these pools render their fee pages in JavaScript, and secondary sources disagree between 2.5% and 4% for ViaBTC, F2Pool and AntPool. Check the live figure before committing.

What actually costs you money — ranked

Take a machine earning $8/day, or $2,920 a year.

1. Headline fee. Each 1% is $29.20/year. Foundry at 0% versus F2Pool at 2.5% is $73/year per machine — $730 across ten.
2. Stale and reject rate. A 1% stale rate costs exactly what a 1% fee costs, and it never appears on your dashboard as a charge. No pool publishes per-pool stale rates, so this is the invisible one.
3. Payout threshold. This is the one people underestimate. At $8/day you mine about 0.000101 BTC daily. A 0.005 BTC threshold means 49 days — roughly $395 — permanently in transit. At 0.001 BTC it is 10 days. At ViaBTC's 0.0001 BTC, about one day. If you ever stop mining below the threshold, that balance is stranded: forfeiting $395 wipes out 13.5 years of savings from a 1% lower fee.
4. Stratum latency. Use the EU endpoint. Sub-40ms is normal in Europe; pointing a European machine at a US-only stratum raises your stale rate.
5. Whether the pool pays transaction fees. Worth about 0.78% currently — real, but the smallest item on this list.

The ranking is the point: threshold and stales usually matter more than the fee percentage everyone shops on.

Kaspa pools — and why the variance argument disappears

Kaspa's block structure changes the maths completely. Post-Crescendo the network produces roughly 10 blocks per second — about 864,000 a day, against Bitcoin's 144. A 12 TH/s IceRiver KS5L is about 1/28,000th of the network, and still expects around 31 blocks a day even mining solo. GhostDAG means no orphans.

The consequence: PPLNS on Kaspa is already smooth, so paying a PPS premium buys you almost nothing. Fee percentage and payout threshold are the entire decision. Note also that ACC-Pool showed zero miners and zero hashrate when we checked — treat it as dormant.

Litecoin and Dogecoin: the merged mining detail people miss

Scrypt miners submit one hash to both the Litecoin and Dogecoin chains via AuxPoW, and you are paid in both coins separately. That much is well known. What is not is the split.

Structurally: 6.25 LTC per 2.5-minute Litecoin block, against 10,000 DOGE per 1-minute Dogecoin block — 4,000 DOGE for every 1 LTC. At LTC $50.27 and DOGE $0.0868 on 26 August 2026, that works out to roughly:

• Dogecoin: ~87% of gross Scrypt revenue
• Litecoin: ~13%

ViaBTC's own worked example puts it at 89/11. Either way the conclusion is the same, and it is genuinely counterintuitive: an "LTC miner" is really a Dogecoin miner with a Litecoin kicker. A pool that does not merge-mine DOGE costs you about 87% of your revenue.

litecoinpool.org runs pure PPS on both coins at a 2% nominal fee with fee-free automatic payouts. ViaBTC, F2Pool and AntPool also merge-mine — but verify their Scrypt-specific rates, which differ from their Bitcoin rates. If you are running an Antminer L9 or an ElphaPex DG-series machine, this is the single most important setting you will choose.

Ethereum Classic pools

2Miners is the practical pick for European miners — 2-hourly payouts, the pool covers payout fees, and it publishes EU ping times (Stockholm 24ms, Helsinki 30ms, Warsaw 31ms, Berlin 34ms).

One piece of news that invalidates most ETC content: Ethereum Classic's block reward dropped 20%, from 2.048 to 1.6384 ETC, at block 25,000,000 in late July 2026. Any ETC profitability figure written before August 2026 is overstated. We also could not confirm that Ethermine still operates an ETC pool — its site is JavaScript-only — so we have left it off rather than list a pool that may not exist.

Pool choice when your miner is hosted

This is worth stating clearly because it is a common misconception: a hosted machine points at your pool account and your payout address. Rewards never touch the host's balance sheet.

Three consequences:

• Pool choice is completely independent of your hosting provider. You can switch pools without touching the hosting contract.
• Your pool dashboard is an independent uptime oracle. A 24-hour average below spec is downtime you can raise with the host regardless of what their status page says — see the uptime and SLA guide for how to use that.
• Set worker-down alerts pool-side. 2Miners has a Telegram bot, Braiins a mobile app, Kryptex and Woolypooly both have monitoring bots. All free, all vendor-neutral.

For hosted customers specifically we would recommend a low payout threshold, so that a machine going out for RMA does not strand a balance you cannot withdraw.

Solo mining: the honest odds

With the network around 870 EH/s, an Antminer S21 Pro at 234 TH/s represents about 0.000027% of global hashrate.

• Expected: 0.0000387 blocks per day — one block roughly every 70.7 years on average, median about 49 years.
• Probability of finding a block in a full year: 1.40%.
• The prize: about 3.1497 BTC, roughly $249,000.
• Ten machines gets you to a 13.1% chance in a year.

It is a lottery ticket with roughly fair expected value and brutal variance. Never a cash-flow plan. If you want to try it, solo.ckpool.org charges 2%, needs no registration, mines to your own address and has run since 2014 — its EU endpoint is eusolo.ckpool.org. ckpool itself warns that below 100 GH/s it is unrealistic. In the last 90 days Solo CK found four blocks, Public-Pool one and Braiins Solo one.

A practical default

If you want one answer rather than a framework:

• Bitcoin, one to ten machines: ViaBTC on PPS+ for the 0.0001 BTC threshold, or Braiins if you want Lightning payouts with no minimum.
• Kaspa: Kryptex at 1% with hourly payouts and a 10 KAS minimum.
• Litecoin/Doge: litecoinpool.org or ViaBTC — but confirm DOGE merged mining is enabled.
• Ethereum Classic: 2Miners, for the 2-hourly payouts and published EU latency.

Read next: hosted miner uptime and SLAs, the profitability ranking, or why ASIC prices fell in 2026. Pool distribution from mempool.space, fee schedules from pool operators' own documentation, prices from Coinpaprika — all 26 August 2026. Fees and thresholds change; verify before pointing hardware.