Crypto fees, spreads, and friction

Definition

Crypto trading friction stacks four layers: explicit venue trading fees (maker/taker schedules, typically bps-scale and volume-tiered), the bid-ask spread and depth cost of the venue’s book, transfer costs (on-chain network fees plus venue withdrawal fees) whenever assets move between venues or into self-custody, and fiat on/off-ramp costs. The all-in figure is routinely a multiple of what an equity-habit friction model assumes, and it varies by venue, pair, size, and hour — friction here is MEASURED per route, never assumed (ms-slippage-friction).

How it works / structure

• Maker/taker fees: venues charge takers more than makers (maker rebates exist at tier extremes); tiers key on rolling volume. The maker/taker gap is wide enough that fee-aware execution (post-only entries — crypto-venue-order-types) changes strategy economics at bps scale.
• Spread + depth: major-pair spreads on deep venues compress to bps in calm hours; minor pairs and off-hours widen sharply (crypto-sessions-24-7). With no NBBO, the spread you pay is the venue you chose (crypto-spot-market-structure).
• Transfer friction: moving assets between venues costs a network fee (congestion-priced, spiking exactly in busy markets) plus venue withdrawal fees plus confirmation latency (crypto-transfer-settlement) — the friction bound that lets cross-venue price dispersion persist (Makarov-Schoar 2020).
• Fiat ramps: deposit/withdrawal rails (wire, ACH) carry their own fees and cutoffs — the only layer that still keeps banking hours (crypto-sessions-24-7 inverts here).
• Wrapper alternative: regulated wrappers replace this stack with equity-style friction — commission/spread on an ETP plus its expense ratio (crypto-etps), or futures fees plus roll (crypto-cme-futures) — often cheaper all-in for exposure, at the cost of hours and tracking differences.

When it applies

Every strategy expectancy calculation on crypto (friction is the first falsifier of high-turnover crypto strategies), venue and route selection, arbitrage/basis theses (the friction stack IS the arbitrage bound), and wrapper-vs-direct decisions where the friction comparison drives the structure choice.

Risk profile & failure modes

• Friction-blind backtests: strategies tuned on mid prices without fee tiers, realistic spreads, and transfer costs produce expectancy that vanishes live — the standard failure of ported equity habits (quant-backtest-hygiene).
• Congestion coupling: network fees and spreads widen together in stress — the friction model must be regime-aware, not constant (regime-volatility).
• Tier cliff assumptions: fee tiers assumed at backtest volume may not hold live; the schedule is a dated input.
• Hidden ramp costs: strategies that cycle fiat frequently pay ramp friction that never appears in exchange fee schedules.

Evidence & limits

The friction-bounds-arbitrage mechanism is peer-reviewed (Makarov-Schoar 2020). Specific fee schedules, spreads, and network-fee levels are venue- and time-specific operating facts that drift — this entry pins the structure of the stack, and the platform requires current measured values per route at thesis time, not stale citations.

Falsifiable-thesis examples

Illustrations only, not signals:

• “The strategy’s live all-in friction per round trip stays under 25 bps at executed size this quarter (friction-model thesis)” — falsified by the execution log.
• “All-in cost of one unit of bitcoin exposure via the spot ETP (spread + expense accrual) undercuts direct spot purchase plus custody for holding periods beyond 90 days (wrapper-economics thesis)” — falsified by the measured cost comparison.

Cross-references

• Friction fundamentals: ms-bid-ask-spread, ms-slippage-friction, ms-liquidity
• The layers: crypto-venue-order-types (fee-aware execution), crypto-transfer-settlement (network fees), crypto-spot-market-structure (venue dispersion)
• The wrapper alternative: crypto-etps, crypto-cme-futures