Choosing a mining pool is not like picking a mobile carrier. It's closer to structuring a financial instrument - one where the wrong architecture costs mid-to-large operations six figures annually while showing up nowhere on a fee comparison table.
The fee percentage is the number every pool advertises. It's also the least predictive variable in your actual earnings. Operators who understand what drives real yield - payout model risk, infrastructure latency, transaction fee capture, hashrate distribution - consistently outperform their theoretical revenue benchmarks. Those who optimize for the fee line alone don't.
Here's what the due diligence actually looks like.
**The Fee Illusion: Why 0% Can Cost You More Than 2%
**
Stale shares are shares your miners submit that the pool rejects because they arrived after a new block was already found. They represent real computational work that earns nothing. And unlike pool fees, they don't appear in the pricing table.
Run 100 TH/s at 0% fees with a 2% stale rate. Compare that to 100 TH/s at 2% fees with a 0.1% stale rate. The math:
- Pool A (0% fee, 2% stale): 100 TH/s × 98% effective = 98 TH/s working
- Pool B (2% fee, 0.1% stale): 100 TH/s × 99.9% effective × 98% net = 97.9 TH/s working
Effectively identical at 100 TH/s. But stale rates compound with scale. At 1 PH/s, a 1.5% stale rate differential - entirely plausible between a well-optimized global pool and a poorly routed regional one - costs $40,000–$80,000 per year at current network economics. At 10 PH/s, that's a material P&L line.
Stale rate is driven by two infrastructure variables: server geographic proximity to your farm and job latency - how fast the pool pushes new block templates after the network finds a block. Pools running globally distributed infrastructure with tight job propagation consistently outperform single-datacenter setups on this metric. MARA Pool and Luxor have both built North American latency infrastructure specifically to address this for large US-based fleets. It's an operational investment that doesn't show up in any fee comparison but shows up clearly in 90-day BTC-per-terahash performance.
**Payout Architecture: Who's Holding the Variance?
**
The payout model determines whether you or the pool absorbs network luck. This matters more than the fee differential for most operations.
PPLNS - deployed by Braiins Pool and Luxor - ties your payout to the pool's recent block-finding luck. Good luck streak: you earn above theoretical. Dry spell: you absorb it. For fleets running continuous, stable hashrate, PPLNS is workable. For anyone running maintenance windows, firmware migrations, or variable power schedules, PPLNS is structurally punishing. Every disconnection burns your score window. You come back online and earn sub-proportional payouts until your shares rebuild.
PPS gives you a fixed rate per valid share regardless of block luck. The pool carries the variance; you pay a premium for certainty. Predictable treasury cashflow, higher fee.
FPPS - used by Foundry USA, AntPool, ViaBTC, Binance Pool, and WhitePool - extends PPS to include transaction fees in the base payout, not just block subsidy. This is the detail most operators miss until they model it.
Miners on FPPS pools captured that premium automatically in their per-share rate. Miners on pure PPS pools didn't - their rate was calculated against block subsidy only. The spread during that period ran 8–14% in favor of FPPS for multiple weeks. Post-halving, when block subsidy dropped and fees became a larger proportion of total block reward, the same dynamic repeated. FPPS isn't just a structural preference - it's a material revenue difference during the network events that are becoming more frequent, not less.
WhitePool's FPPS implementation includes daily BTC settlement with a 0.001 BTC minimum threshold, which matters for treasury operations managing cash flow against fiat-denominated power contracts. Settlement timing and threshold structure are rarely discussed in pool comparisons but directly affect working capital for operations running on thin margins.
**
**Hashrate Concentration: What Your Pool Affiliation Actually Signals
**
**AntPool and Foundry USA collectively represent 50–55% of global network hashrate most weeks. Both run excellent infrastructure. That's not the concern. The concern is what that concentration means for the network's censorship resistance, regulatory exposure, and - increasingly - for institutional operators whose legal counsel is asking questions about where their hashrate goes.
Pools like Braiins, Luxor, F2Pool, and WhitePool collectively hold smaller but structurally important shares of global hashrate. For compliance-aware operations - particularly those with institutional capital structures or operating in jurisdictions with evolving mining regulation - pool selection is part of the risk disclosure conversation, not just an operations decision.
**How to Actually Compare Pools
**
The table above tells you the structure. It doesn't tell you the outcome. The only honest comparison method is BTC earned per TH/s over 30-90 days in real operating conditions. Hashrate Index publishes pool-level payout benchmarks that let you track this dynamically - it's one of the few public datasets where real revenue per terahash diverges visibly from theoretical fee-adjusted expected values.
If you're seriously evaluating a pool switch, run a parallel hash split for 30 days. Same ASIC models, same firmware, same power settings, 50/50 split between pools. The data will tell you more than any fee comparison.
**Before You Point Another Terahash at a Pool
**
When you evaluate a pool, you're making decisions across five dimensions simultaneously:
1. Payout model - who carries variance, and does the structure match your uptime profile
2. Transaction fee capture **- does the pool pay you FPPS or leave fees on the table
**3. Infrastructure quality - latency, stale rate, uptime track record in your geography
4. Settlement mechanics - payout frequency, minimum threshold, currency options
5. Hashrate distribution - your regulatory exposure and censorship-resistance position
Fee percentage is one input inside dimension three. It's not a summary of the decision.
The operators consistently beating their theoretical revenue aren't the ones who found the 0% pool. They're the ones who ran the 90-day data comparison, matched their payout model to their operational profile, and selected infrastructure that performs in their geography. That's not sophisticated - it's just rigorous. The industry's growing up. Pool selection should too.
Pool fee and payout data current as of publication. Mining revenue projections vary with network difficulty, BTC price, mempool conditions, and fleet-specific hardware configuration. Historical fee event data sourced from public mempool analytics.
_Disclaimer: This is not financial or investment advice. Do your own research before making any decisions. Use at your own risk.
_
