Check the result, with its units

Crypto Futures Formulas and Exchange Scores

Use the equation for the product you are trading. Linear position costs, inverse payoffs and account-wide margin are different models. The examples below run through the same calculation functions as our tools; a displayed estimate is not an exchange fill or account guarantee.

Margin, notional and the quantity you can actually open

For a linear contract, notional N equals underlying quantity q multiplied by price P. Initial margin M is N divided by leverage L, before additional exchange requirements. With $10,000 exposure at 5×, the simple margin is $2,000.00. That amount is neither the maximum loss nor the cash required after opening fees and other reserves.

ExposureN = q × P

Simple initial marginM = N ÷ L

Risk-sized exposureRisk budget ÷ (stop distance fraction + modeled cost fraction)

The position-size calculator limits that risk-sized amount by available capital. It reserves the supplied opening fee alongside initial margin, then rounds quantity down to the supplied lot step. A request for more leverage cannot override an exchange’s stricter initial-margin rate, contract ceiling or position limit.

Worked example: $1,000 capital, 1% risk budget, entry $100, stop $98, 5× leverage, 10 bps of modeled costs and a 0.1-unit lot step. The $10 risk budget divided by 2.1% permits about $476.19 of exposure. Rounding leaves 4.7 units, $470.00 notional and $94.00 simple margin. Modeled stop loss including those costs is $9.87; a gap or worse fill can exceed it.

A position cap is a dollar amount, while leverage is a multiple. The engine checks capacity and resident eligibility separately; it does not take the minimum of a dollar cap, a leverage multiple and a yes/no eligibility flag. Use the position-size calculator for a specific stop and budget.

Trading fees and the exact break-even price by side

A fee is executed notional times the fee percentage divided by 100. A 0.05% fee on $10,000 is $5.00, not $500. The closing fee applies to the closing notional, which changes with price even if the underlying quantity stays constant. Charging both legs against the starting notional is useful for a quick cost estimate, but not an exact break-even exit.

For the equations below, E is entry price, q is quantity, f is the exit-fee fraction and C contains all costs except the exit commission. C includes the opening fee, modeled slippage, funding and other entered costs. A funding receipt is negative. Convert 0.05% to 0.0005 before putting it into f.

Long: solve q × (exit − E) = C + q × exit × fExit = (E + C ÷ q) ÷ (1 − f)Short: solve q × (E − exit) = C + q × exit × fExit = (E − C ÷ q) ÷ (1 + f)

Our example has $10,000 starting notional at $50,000, so q = 0.2. Entry and exit fees are both 0.05%; the entered total slippage allowance is 4 bps, and funding paid is $2. C is $5 + $4 + $2 = $11. These inputs are hypothetical, not the fee schedule of a recommended exchange.

Same inputs, different closing-price solution
Position	Unrounded exit	Conservative $0.10 tick
Long	$50,080.040020	$50,080.10
Short	$49,920.039980	$49,920.00

The starting-notional cost estimate is $16.00, or 0.160%. The exact price calculation above accounts for the changing exit commission. For a long, round the required exit up to an allowed tick; for a short, round down. Formatting both prices to two decimals is not a substitute for a market’s order-price step.

Spread, book impact and an all-in slippage allowance can overlap. Add only costs the chosen input has not already captured. The break-even calculator also distinguishes a reachable exit from a result already covered by net receipts or an unreachable modeled short target. These are linear-position equations; inverse coin P&L requires a different denominator.

Funding fees: count settlements before annualizing anything

For each funding event, payment equals the position value at that event times the signed rate. Under the convention used here, positive payment is money paid: a positive rate costs the long and credits the short. Borrowing is a separate charge and is not reversed merely because the position is short.

Variable observationsTotal payment = sum of (event notional × event rate fraction × side sign)Long sign +1; short sign −1. Our projection tool holds notional fixed unless you calculate separate scenarios.

At a constant 0.01% per eight hours on $10,000, each event costs a long $1. A ten-hour position with the next event in two hours encounters events at hours 2 and 10: $2.00 in the scheduled model. A pro-rata or continuous model produces $1.25 for the same duration. They are different settlement assumptions, not alternative rounding choices.

The two-phase example starts at +0.01%, changes to −0.02% at hour 10, and ends at hour 18. The new rule owns the boundary event: payments are +$1 at hour 2, −$2 at hour 10 and −$2 at hour 18. Net payment is $-3.00: the long receives $3. The engine does not charge both the old and new rates at the transition.

At zero holding time the projection charges no event. At a positive holding period ending exactly on the modeled settlement instant it includes that event; actual exchange timing and position eligibility control real charges. The annualized equivalent multiplies a fixed interval rate by intervals per day and 365. It neither compounds nor forecasts the next rate. Use the funding calculator to choose the appropriate schedule.

VWAP, price impact and the portion of the order observed

Volume-weighted average price is total executed quote value divided by total executed underlying quantity. Do not average displayed prices without weighting the quantities. Our order-book model consumes the observed asks for a buy and bids for a sell; contract multipliers are applied where the venue reports quantity in contracts instead of tokens.

Buy the visible example book1 unit at $100 + 2 units at $102 = $304 for 3 unitsVWAP = 304 ÷ 3 = $101.333333With best bid $99 and ask $100, midpoint = $99.50.

Buy impact from midpoint is (VWAP ÷ midpoint − 1) × 10,000, giving 184.2546 bps. Against the best ask it is 133.3333 bps. Those references answer different questions. Midpoint impact already contains the cost of crossing to the ask; adding the full spread again would overstate it.

If the requested size exceeds the captured levels, only the observed portion has a calculable VWAP. Missing depth is not proof that the rest of the exchange has no liquidity. An exit simulation is another sweep of a captured book, not a trade we executed. Pool capacity is reported separately and cannot be converted into an invented order-book fill percentage. The slippage estimator explains the coverage of its selected mode.

Liquidation equations only work inside their supported model

For the generic linear isolated model, let m be maintenance-margin fraction, f the closing-fee fraction, B additional margin in dollars and D the maintenance deduction in dollars. Quantity stays fixed. The equation values maintenance at the estimated closing mark; it is more specific than subtracting 1/leverage from the entry price.

Generic linear isolated longPrice = E × (1 − 1/L − B/N − D/N) ÷ (1 − m − f)Generic linear isolated shortPrice = E × (1 + 1/L + B/N + D/N) ÷ (1 + m + f)

At $50,000 entry, 10× leverage and 0.5% maintenance, with no fee, extra margin or deduction, the generic long estimate is $45,226.13. This is a worked model result, not a claim that every exchange liquidates a BTC long there. Mark prices, additional balances and the applicable maintenance ladder matter.

Model selection in the liquidation tool
Model	Distinct treatment
Generic linear isolated	Closing fee enters the maintenance-side denominator; tier selection is checked again at the estimated liquidation notional.
Bybit UTA isolated USDT	Uses the documented UTA fee treatment, including its adjustment to extra margin; not the generic denominator.
Hyperliquid isolated	Uses margin available, position size and maintenance rate without a separate liquidation clearance-fee input.
Loss-threshold products	Uses the documented loss threshold and accrued costs; this is not a maintenance-margin ladder.
Cross / portfolio or inverse	No universal full-account reproduction. The supported mode and missing inputs must remain explicit.

A short’s dollar notional grows when price rises. It may therefore cross a maintenance boundary before liquidation. The tier solver must converge on the bracket containing its own estimated notional; an incomplete or inconsistent ladder returns no solved result. A value exactly on a Bybit cap uses that provider’s inclusive convention. Other normalized catalogs can use the next tier at a shared boundary.

The liquidation calculator exposes the selected model. It is not a promise that a stop will fill first, and it does not simulate an entire unified account from one position’s inputs.

How the exchange score is calculated

For each of the five product criteria, divide the grade by 5 and multiply by its percentage weight. Add the contributions and round once. The arithmetic is reproducible; the grades are editorial judgements supported by the reasons and sources in the rating record.

Binance: the current calculation

One exchange rating used throughout the site
Criterion	Grade / 5	Weight	Contribution
Trading costs	4.5	20%	18.00
Execution and capacity	4.5	25%	22.50
Trading and risk controls	5.0	25%	25.00
Collateral and moving funds	4.5	15%	13.50
Operational safeguards	4.0	15%	12.00

Total: 91/100. Model: exchange-quality-2026.10.10.2. Recommendations and scenario costs are explained separately.

Finder orders compatibility categories first. Within each category, a complete cost estimate precedes an unknown one, and the lower estimated cost comes first. Conditions break remaining ties. Missing charges are never treated as a verified zero cost.

How we rate each exchange

We rate the trading product, including costs, execution, controls, moving funds and operational safeguards. Each component gets an editorial grade from 0 to 5; the weighted total becomes a score out of 100. The sources support the facts behind our judgement, not the grade itself.

costs (20%): Regular trading charges, holding costs and fee exceptions for the named product. A temporary promotion or a private VIP tier is not the default.
execution (25%): Practical pricing and trade capacity. Order books, pool quotes and synthetic pricing are assessed through their own execution rules; an absent connector or collection earns no penalty.
controls (25%): Useful entry, exit, margin and position-management controls. More leverage and more token listings do not automatically mean a better product.
funding (15%): How well the product supports collateral, deposits, transfers and withdrawals for its intended users. A DEX does not lose points for lacking bank deposits.
resilience (15%): Mechanisms for liquidations, abnormal prices, failed transfers and access to funds, including relevant incidents and unresolved limitations. This is not a solvency or smart-contract security certificate.

A missing API or an uncollected order book does not lower the score. We assess pool-based exchanges on how their own pricing and trading work. A wallet-based platform does not need bank deposits to score well. Affiliate payments and advertised maximum leverage earn no points.

Scores are editorial comparisons, not safety guarantees or funded-account test results. When a critical product question remains unresolved, we withhold the overall rating. Source record dates are listed separately from the rating review.

Download the grades, reasons and sources

Binance: 91/100

Global USDⓈ-M and COIN-M futures. A strong all-round futures platform: competitive regular fees, substantial major-market capacity and extensive position controls.