How FEEX models execution cost
FEEX is currently a working prototype. The comparison model uses documented fee and spread assumptions rather than live exchange market data.
Execution scope
Trade only — Trading fee + estimated spread / price impact.
Trade + withdraw — Trading fee + estimated spread / price impact + estimated exchange withdrawal/network fee.
Network selection applies only to Trade + withdraw.
Trading fee
Trading fees are modelled from the selected representative fee profile (below) and the liquidity role of the order. Market and Stop Market orders always execute as taker. Limit and Stop Limit orders default conservatively to marketable/taker; a limit order can still pay taker fees if it executes immediately.
Fee profiles
FEEX uses representative fee profiles because real qualification thresholds differ by exchange, account and jurisdiction. Profiles are illustrative tiers, not universal volume brackets.
Coinbase values are prototype account-tier assumptions only — Coinbase Advanced rates are account-specific and these figures are not official current rates.
| Exchange | Standard retail maker / taker | Active trader maker / taker | Higher-volume maker / taker |
|---|---|---|---|
| OKX | 0.10% / 0.20% | 0.080% / 0.13% | 0.055% / 0.065% |
| Binance | 0.10% / 0.10% | 0.090% / 0.10% | 0.040% / 0.060% |
| Bybit | 0.10% / 0.10% | 0.068% / 0.080% | 0.063% / 0.075% |
| Kraken | 0.40% / 0.80% | 0.22% / 0.38% | 0.12% / 0.25% |
| Coinbase | 0.40% / 0.60% | 0.25% / 0.40% | 0.15% / 0.25% |
| Crypto.com | 0.25% / 0.50% | 0.20% / 0.40% | 0.10% / 0.20% |
| Gemini | 0.60% / 1.20% | 0.25% / 0.50% | 0.075% / 0.15% |
Estimated spread / price impact
For taker execution, FEEX models the effective spread / price-impact rate as:
base exchange spread × asset liquidity factor × trade-size factor
Each exchange carries a static base spread assumption. Asset liquidity factors range from 0.75 (BTC) to 1.40 (ARB, OP); trade-size factors rise from 0.80 (≤ $1,000) to 2.00 (> $50,000). This is a prototype proxy for executable cost — it is not a live order-book simulation.
Maker execution
Post-only maker scenarios set the spread-crossing cost to zero, because a resting order does not cross the spread.
Fill risk not modelled. FEEX does not estimate fill probability, opportunity cost or adverse selection. Maker execution is not universally cheaper in practice — an unfilled order can cost more than the fees it saves.
Withdrawal / network fee
FEEX currently models one estimated exchange withdrawal/network fee when Trade + withdraw is selected. It does not add a separate gas charge to the same withdrawal flow, reducing the risk of double-counting network cost.
What 'cheapest' means
Cheapest means the lowest estimated total cost under the selected execution scope, fee profile and current prototype assumptions. It does not mean the exchange is universally best for price, liquidity, reliability or user suitability.
Asset liquidity factors
Trade-size factors: ×0.80 (≤ $1,000) · ×1.00 ($1k–$5k) · ×1.15 ($5k–$10k) · ×1.35 ($10k–$25k) · ×1.60 ($25k–$50k) · ×2.00 (> $50k).
Route availability
FEEX validates the exact exchange × asset × network combination before ranking a Trade + withdraw route. The current matrix is intentionally incomplete.
Official reference means support was represented in the cited public help documentation when reviewed. Jurisdiction restrictions may differ, including Singapore.
Prototype assumption means the route is explicitly modelled for product testing but is not asserted as current verified exchange support.
Unmodelled routes are excluded rather than assumed valid. A production implementation should query exchange asset/network metadata and operational status dynamically before every comparison.
Route-availability references · reviewed Oct 2026
Route support entries marked as official references were informed by the public help pages below. FEEX is not automatically synced to these sources.
Live order-book beta
Live order-book mode is currently implemented for OKX, Binance and Bybit spot USDT pairs only, and applies to Compare and FEEX Intelligence — Dashboard continues to use the deterministic model. FEEX Intelligence explains the normalised result produced by the same FEEX calculation layer; the LLM never receives raw exchange API payloads and never calculates or ranks costs itself.
Public REST snapshots are fetched server-side on demand. For a market buy, FEEX simulates spending the notional across visible asks from cheapest upward and computes a volume-weighted average price (VWAP). Effective book impact = (VWAP − mid) / mid.
A 5-second server cache reduces unnecessary public API requests. If fewer than two valid live venues are available, FEEX falls back to the deterministic model for the whole comparison — live and modelled venues are never mixed in one ranking.
Maker/post-only scenarios remain modelled only, because queue position, fill probability and adverse selection are not captured. Live market data does not make fee tiers or withdrawal fees live — those remain representative modelled assumptions.
Prototype limitations
- Static fee and spread assumptions
- Representative fee profiles, not account-specific tiers
- No live order-book or quote data
- Maker fill risk, opportunity cost and adverse selection not modelled
- No regional pricing variation
- No guaranteed future execution price
- Exchange / asset / network availability may differ in production
- Prototype does not execute trades
Reference fee schedules · reviewed Oct 2026
Prototype assumptions were informed by the public schedules below. FEEX is not automatically synced to these sources; figures may drift from current published rates.
FEEX is a decision-support prototype, not an execution venue or investment-advice product.
Compare a scenario →