Choose a crypto trading workflow before comparing brands

Last Updated: 2026-10-27

Risk warning: Trading, leverage and third-party custody can cause losses. This is a requirements and cost worksheet, not financial advice or a tested platform ranking.

A useful comparison starts with one task: buying an asset with local currency, placing a particular order, reviewing charts, following a strategy or running a rule. A long list of pairs, a low headline fee and an affiliate bonus cannot answer all five. Record the exact product, interface, account mode and legal entity being compared.

1. Eliminate products you cannot use

For each candidate, save the official eligibility page and date, your residence, the entity serving it, required verification and the specific product. Access to spot trading does not establish access to derivatives, staking or copying. An account opening successfully does not prove every product is permitted. Do not use a VPN, a different identity or a misleading payment description to bypass restrictions.

Next verify the asset and funding route: exact asset/network, fiat currency, deposit and withdrawal method, minimum amounts, limits and receiving-bank requirements. A ticker appearing somewhere on a venue does not prove the required pair, network or withdrawal route exists. An internal account transfer is not the same as an on-chain withdrawal.

Keep the candidate marked unresolved when one of these checks has no current source. A low fee cannot compensate for an unavailable product or unusable exit route.

2. Separate analysis, execution and custody

Scroll sideways to see every column. Keyboard: focus the table region and use the arrow keys.

An embedded chart library may meet a simple analysis need without a separate paid subscription. Conversely, a venue's built-in charts may lack a particular alert or data workflow. Write the missing feature down before purchasing it. See the charting-versus-execution guide for a worked requirements example and TradingView's support index for current feature and integration documentation.

Do not call a connected charting or automation account financially harmless. If an alert, webhook or API can trigger trades, compromised instructions can affect an execution account even without withdrawal permission. Keep the permission boundary in the comparison.

3. Put costs on one turnover basis

Define filled turnover as the sum of the notional value of each executed fill. A $1,000 entry and $1,000 exit contribute $2,000. If a table already includes both, do not double it again. Use actual maker/taker proportions; a limit order is not necessarily a maker fill.

commission = maker turnover × maker rate + taker turnover × taker rate

Add separately: observed spread/slippage, subscription, funding or borrowing, conversion, withdrawal and other applicable fees. Do not subtract spread twice if actual entry/exit prices already include it. A provider's commission to this website is not a customer's cost advantage.

Synthetic example: $10,000 filled turnover at 0.1% costs $10. A $29.99 monthly subscription cannot beat that $10 on trading commission alone if its eligible turnover is capped at $10,000, even if the eligible commission becomes zero. Other genuine features might justify it, but they need a separate reason.

At $100,000 filled turnover, reducing 0.10% to 0.04% saves $60, not $1,200. Twenty $1,000 fills a week for 52 weeks produce $1,040,000 turnover. Twenty round trips with two such fills produce twice that. Always state which meaning applies.

Provider schedules can separate basic, advanced, API and subscription products. For example, Kraken's fee schedule, checked October 25, 2026, distinguishes instant/recurring transactions from Pro trading and describes subscription exclusions. This is why copying one brand-level fee into every row is unreliable. The fee worksheet records current inputs rather than declaring a universal cheapest exchange.

4. Decide whether a subscription has a job

Use break-even eligible turnover = subscription cost ÷ rate reduction, only when the rates cover the same transactions and no cap or other cost intervenes.

A hypothetical $9.99 subscription eliminating a 0.05% fee saves $1 at $2,000 turnover, so it costs $8.99 more. At $20,000 it avoids $10, leaving only $0.01 before any other differences. A zero-fee promotion may change the comparison again. These examples do not claim that any provider currently offers those terms.

Token-based discounts also require a separate exposure decision. Record the token amount and acquisition cost, price risk, holding conditions and duration. Do not describe holding a volatile asset as a free reduction in fees. Do not manufacture extra trades to reach a tier without including their costs and checking the rolling eligibility window.

5. Compare the actual instrument and margin mode

For derivatives, record contract type, settlement/quote currency, linear or inverse payoff, margin asset, cross or isolated mode, collateral haircuts, maintenance requirements and funding/borrowing rules. An options product is not interchangeable with a perpetual simply because both are called derivatives.

Shared collateral may let assets support multiple positions, but it can also transmit losses between them. A per-leader allocation is not automatically a hard maximum loss when account modes share funds. Compare the same position set under each candidate's documented mode before calling one capital-efficient.

For multiple positions, portfolio leverage is total relevant notional divided by equity, with direction and correlation recorded separately. Three equally funded 10× allocations do not automatically create 30× leverage. They can still create a concentrated 10× exposure.

6. Measure depth for your order, not the coin count

Choose a proposed order size and record executable bid/ask depth at a timestamp for the exact pair. Estimate a volume-weighted fill through the displayed levels and include fees. A published venue throughput figure does not measure the latency or fill quality of your account; placing a few trades does not validate the matching engine's maximum capacity.

Keep the distinction between an observation and a promise. A small sample with no outage or missed fill cannot establish “zero latency,” “no slippage” or future availability. A token's listing does not establish adequate depth for an exit during stress.

Preserve the observation record

For a matched comparison, retain the instrument, side, size, timestamp, quoted depth, reference price, actual fills and costs. Do not compare one venue during a quiet minute with another during a release and attribute the difference entirely to the venue. A paper fill estimate is not a measured live fill.

Synthetic example: a buy of two units has a reference mid-price of $100. Fills of one unit at $100.10 and one at $100.30 give a $100.20 average, or 20 basis points above that reference. At a 0.05% commission, the $200.40 filled turnover adds $0.1002. Keep commission separate from the $0.40 difference from the mid-price. The reference must use a recorded timestamp; otherwise the apparent slippage may include an unrelated price move.

For withdrawals, distinguish request, approval, broadcast, first confirmation and recipient availability. For support, distinguish an automated acknowledgement from a substantive response and resolution. Four successful withdrawals or a fast response in one sample do not prove future availability, safety or solvency. A testnet's order book also does not establish production execution quality.

A hypothetical $69 quarterly subscription eliminating a 0.1% spot commission breaks even at $69,000 eligible filled turnover before caps and other differences. At $25,000 it avoids $25, leaving a $44 net cost. Count any separate derivatives rebate against its own eligible turnover; do not combine unlike rates into an unexplained winner.

7. Native tools versus external automation

A built-in tool may avoid an additional third-party integration, but the venue and strategy risks remain. External software may support a specialized rule or multiple venues while adding another service, permission boundary and subscription. Verify the exact supported exchange, product, order types and plan limits; do not infer them from a generic integrations logo.

Before a paper trial, specify order identifiers, duplicate handling, partial-fill reconciliation, stale-data expiry, restart behavior and a halt that survives a process restart. A response timeout must not trigger a duplicate order without checking its state. A withdrawal-disabled trading key can still lose funds through execution.

For an API workflow, record endpoint-specific request weights, order-rate limits, retry instructions and supported order types from the provider documentation. A native bot does not answer those API questions. An options comparison needs options contract and fee inputs; perpetual fees are not a substitute. Splitting activity across venues also adds transfers, fragmented collateral and possibly lower fee tiers. Record those costs before describing a multi-venue setup as an improvement.

For strategy selection, use the copy-versus-bot control comparison and the DCA order-ladder worksheet. Neither supplies a return forecast.

Write a decision that can be revised

Scroll sideways to see every column. Keyboard: focus the table region and use the arrow keys.

Finish with a conditional choice such as: “A meets the documented withdrawal requirement; B is unresolved. Recheck A's fees before using it.” Do not turn unknowns into a scored winner. If neither meets the requirement, the useful result is neither established, not a forced recommendation.

Affiliate disclosure: AITR may earn commissions from marked referral links elsewhere. This worksheet does not rank providers by commission, assert hands-on testing or supply an affiliate signup recommendation.