Has every item been tested first-hand?

No. Each page labels its evidence status. Published rules and calculations can be checked first, but KYC, deposits, orders, support and withdrawals require real account records before they are marked tested.

Why is there no star rating or recommendation ranking yet?

Scoring without comparable first-hand evidence creates false precision. An explainable conclusion will only be added after repeatable tests with timestamps and redacted evidence.

Binance, OKX and Bybit perpetual costs: a reproducible comparison workflow

Lock the comparison conditions first

Do not compare headline annualised numbers from three screens. Choose the same margin currency, underlying, directional convention and notional, and record the UTC timestamp. The worked example uses BTC USDT linear perpetuals and 10,000 USDT per leg. Its numbers demonstrate the calculation and are not live quotes for 2026-08-30.

Scope: tradable USDT linear perpetual contracts only
Account: record the actual VIP tier, maker and taker rates, and discounts
Window: use at least 30 days of realised funding and capture books within the same minute
Units: convert every cost to USDT and the same notional

Collect four data groups per venue

Collect contract rules, realised funding settlements, depth for the intended order size, and the fee rate that applies to the account. Contract rules include minimum quantity, quantity step, tick size, contract value and funding interval. Never substitute a predicted funding rate for a realised settlement, and never substitute a promotional minimum fee for the account rate.

Binance: exchangeInfo for trading and quantity rules, fundingRate for settlements, and depth for simulated fills
OKX: public/instruments for SWAP ctVal, lotSz and tickSz, then funding-rate-history and books
Bybit: instruments-info for LinearPerpetual, qtyStep and fundingInterval, then history-fund-rate and orderbook

Four spreadsheet-ready formulas

Trading fee = notional × fee rate. Count entry and exit separately; a two-venue hedge has four fills. Buy slippage = notional × (volume-weighted fill price ÷ best ask − 1). Sell slippage = notional × (1 − volume-weighted fill price ÷ best bid). Funding cash flow = notional × realised settlement rate; a positive rate normally moves from longs to shorts and a negative rate reverses the direction. Total cost = entry-and-exit fees + entry-and-exit slippage + net funding paid + other explicit costs.

Average daily funding = sum of realised funding cash flows in 30 days ÷ 30
Daily net carry = funding received − funding paid − dailyised trading and slippage cost
Break-even days = one-off entry-and-exit cost ÷ daily net carry; no usable break-even exists when daily net carry is zero or negative

Complete 10,000 USDT worked example

Assume a 10,000 USDT short on venue A and a 10,000 USDT long on venue B. Four taker fills at 0.05% cost 10,000 × 0.0005 × 4 = 20 USDT. Four fills with 0.03% book slippage cost 10,000 × 0.0003 × 4 = 12 USDT. If the realised funding difference is 0.01% every eight hours, the hedge receives 1 USDT per settlement, or 3 USDT per day.

The one-off cost is 32 USDT and the simplified break-even is 32 ÷ 3 = 10.7 days. This is not a return forecast. A smaller or reversed funding spread, greater order size, wider slippage or temporary divergence between the two legs can extend or eliminate break-even.

Execution checklist

Fifteen minutes before entry: save venue rules, account fees and 30-day funding snapshots
Five minutes before entry: capture three books one second apart and use the worst simulated slippage
Before orders: verify equal converted notionals and leave a maintenance-margin buffer
After fills: replace simulated VWAP with fill reports and record fees and timestamps
After every settlement: update cumulative carry; recalculate if actual carry is 25% below plan
Before exit: simulate both exit books again and do not count unrealised funding as received cash

Explicit stop conditions

Stop treating the spread as executable when conservative daily carry is not positive; break-even exceeds the intended holding period; either book cannot absorb the target size within the slippage limit; funding direction or cadence changes; the margin buffer cannot withstand temporary mark-price divergence; or the product is unavailable for the region or account.

What this method still excludes

Public books are snapshots, not guaranteed fills. Bybit states that RPI orders are excluded from its public book. OKX derivative book quantities can be contract counts and must be converted with ctVal. Funding cadence can change, so do not hard-code three settlements per day. Liquidation, ADL, insurance funds, transfer delays, API latency, outages and tax are outside the simplified formula.