A crypto trading bot's total cost includes software access, hosting or data services, exchange fees, execution price differences and any instrument-specific carrying or transfer charges. Build separate cash and economic-cost budgets, then test them against expected trading volume. Trading capital belongs beside the budget as money at risk, rather than being counted as a software expense.
The useful question is how much your intended operation will consume in a quiet month, an active month and its first month. A subscription quote cannot answer that alone. This worksheet builds those estimates without assuming that more trading creates more profit.
Define what you are buying
A hosted service, downloaded software and a custom-built bot leave different responsibilities with you. Before comparing prices, write down the exchange, instrument, number of accounts, operating hours and functions required. Include who maintains the connection, stores records and responds when orders become uncertain.
For hosted software, confirm the billing period, renewal terms, usage allowances and what happens when a pass or subscription expires. For software you operate yourself, include deployment, monitoring and maintenance. For custom work, obtain a written scope covering acceptance checks, documentation, ownership and ongoing changes. Development quotations without that scope are difficult to compare.
Keep the software decision separate from the strategy decision. Our guide to evaluating an AI crypto trading bot covers selection evidence. The companion explanation of whether crypto trading bots work distinguishes reliable automation from evidence of a trading advantage. Neither question is answered by a low price.
Use two totals, with capital shown separately
The cash budget records invoices and explicit charges. The economic-cost estimate adds measured or modeled execution differences and, if useful, a stated value for your time. These totals answer different questions: how much money must be available for charges, and how much implementation may consume relative to a benchmark.
| Worksheet line | How to estimate it | Evidence to retain |
|---|---|---|
| Software access | Billing-period price plus required add-ons | Dated quote, allowance and renewal terms |
| Hosting and operations | Compute, storage, logs, backups and monitoring | Configuration and itemized estimate |
| Data or usage services | Included allowance plus plausible extra usage | Request, message or data-retention units |
| Exchange trading fees | Each executed value multiplied by its applicable rate | Account schedule and actual fill charges |
| Execution price difference | Executed quantity versus a declared reference price | Timestamped benchmark and fill records |
| Carrying and transfer charges | Relevant funding, borrowing or transfer events | Contract rules and settlement records |
| Setup and exit work | One-time migration, configuration or export costs | Separate deliverables and quotation |
| Your operating time | Hours multiplied by a disclosed personal hourly value | Time log, shown separately from invoices |
Put allocated trading capital and a loss limit below this worksheet. A $5,000 account allocation is exposure that may change in value; it is not a $5,000 operating charge. Margin requirements, locked balances and cash needed to pay fees may constrain available funds even though they are not subscription expenses.
Also separate a contingency reserve from expenditure. Reserving $100 for an incident does not mean an incident cost $100. Report actual spending when it occurs, with the reserve remaining visible as a funding requirement.
Calculate trading fees from fills and account rules
For a simple percentage charge, the fee is executed notional multiplied by the rate. A purchase and later sale usually create two sets of executed value. Counting only entry activity understates the budget when both sides are chargeable.
Do not assume every limit order receives maker pricing. Coinbase Advanced's fee documentation says an immediately filled limit order can incur taker fees. An order's immediate and resting portions may receive different treatment, and its order-time tier determines the applicable fees. These are documented Coinbase product rules; use your own venue's current account and pair schedule.
For planning, weight maker and taker rates by executed value rather than order count. Ten tiny maker fills do not outweigh one large taker fill simply because there are more of them. Reconcile actual charges afterward, preserving any fee token, rebate and reporting-currency conversion.
An execution estimate also needs a clear reference price. If it measures fills against a decision-time midpoint, the price difference already incorporates spread crossing and other movement over that interval. Adding a separate spread allowance can double-count part of the same effect. The transaction cost analysis guide explains that reconciliation in detail.
Worked monthly budget: the same bot at three activity levels
The following is entirely hypothetical. All amounts are USD, the market is unleveraged spot, and fees are paid in USD. None of the software prices, rates or execution assumptions is a current provider quotation.
Assume $5,000 of allocated trading capital and $50 of recurring service charges: $35 for software, $10 for hosting and $5 for data. Assume setup assistance costs another $120 once. There are no funding, borrowing or transfer charges in this example.
The invented fee schedule is 0.06% maker and 0.14% taker. If 75% of executed value is maker and 25% is taker, the blended fee assumption is:
0.75 × 0.06% + 0.25 × 0.14% = 0.08%, or 8 basis points.
Add an assumed adverse execution difference of 0.04%, or 4 basis points, against the declared midpoint benchmark. The combined variable economic-cost assumption is therefore 0.12% of gross executed value. Actual price differences can be favorable or adverse; this positive allowance is a planning input.
Here, gross turnover multiple means total buy-plus-sell executed value divided by the $5,000 allocation. A complete $5,000 purchase and $5,000 sale creates $10,000 of gross value, or 2 times the allocation, if values are unchanged. This explicit definition avoids confusing one-way turnover with both sides of trading.
| Monthly input or result | Lower activity | Middle activity | Higher activity |
|---|---|---|---|
| Gross turnover multiple | 2× | 10× | 30× |
| Buy-plus-sell executed value | $10,000 | $50,000 | $150,000 |
| Explicit fees at 0.08% | $8 | $40 | $120 |
| Recurring service charges | $50 | $50 | $50 |
| Recurring cash charges | $58 | $90 | $170 |
| Modeled execution difference at 0.04% | $4 | $20 | $60 |
| Modeled recurring economic cost | $62 | $110 | $230 |
| Economic cost divided by allocated capital | 1.24% | 2.20% | 4.60% |
The middle column's cash charges are $50 + $40 = $90. Adding its modeled price difference gives $110. First-month cash charges become $90 + $120 = $210; first-month modeled economic cost becomes $230.
If two hours of monthly oversight are valued at an invented $25 per hour, show another $50 as time cost. The middle column then has $160 of recurring economic cost including time. Unpaid oversight changes the economic comparison even though no extra hosting invoice appears.
These percentages describe costs relative to a fixed allocation. They are neither return forecasts nor instructions to seek a matching monthly return. Trading losses and changes in inventory value remain outside this operating-cost illustration.
Turn the estimate into an operating ceiling
Suppose the same hypothetical operation has a $150 monthly ceiling for recurring services, fees and modeled execution differences, excluding personal time. With $50 committed to services and a 0.12% variable assumption, the remaining allowance supports:
($150 − $50) ÷ 0.0012 = $83,333.33 of gross executed value, approximately 16.67 times the $5,000 allocation.
That is a planning threshold under stated assumptions, not a safe trading limit. Partial fills, changing prices and other charges can alter the result. Use actual cumulative charges and a separately stated estimate for pending work when monitoring a budget.
Now stress one assumption. If every fill pays the invented 0.14% taker rate, the middle scenario's fees rise from $40 to $70. Holding the $20 execution allowance unchanged, its recurring economic cost rises from $110 to $140. The activity did not change; the execution mix did.
Place the ceiling, review interval and action on breach in a written crypto trading plan. A useful rule names who reviews the operation and when new activity pauses. A spreadsheet ceiling has no protective effect unless somebody or a configured control acts on it.
Check the costs a headline price can miss
An included cloud allowance can expire, and compute alone may not cover the deployment. For example, AWS's EC2 On-Demand pricing page treats compute usage alongside separate billing considerations for data transfer, public IP addresses, storage and monitoring. That is an AWS example, not evidence that every bot needs those services. Price your actual architecture and applicable taxes.
For on-chain automation, record the network and relevant transactions. Ethereum's gas documentation explains that included execution can consume gas even when it fails. A rejected submission that never executes is a different case. Budgeting only successful swaps can therefore omit real charges; use receipt evidence and the applicable chain's rules.
Other common gaps are exchange conversion costs, historical-data purchases, export work and a second service needed to recover after an outage. A derivative strategy also needs its actual funding or borrowing assumptions. Enter zero only when the item is inapplicable or verified, and mark an unknown amount as unknown.
Validate the first bill before expanding activity
Start with a bounded observation period. Preserve the dated service quote, usage counters, distinct fills, fee amounts and benchmark observations. Compare actual fixed charges with the worksheet, then compare fees per dollar of executed value with the assumed blend. Separate a pricing change from a shift in activity or maker/taker mix.
Investigate discrepancies at their source. A duplicated fill inflates volume; a missing fee token understates charges; a fee-tier change makes last month's rate stale. A quiet bot can still incur subscriptions and storage costs. A busy bot can cross usage allowances even if its trading allocation stays constant.
Before renewing or increasing activity, update the three scenarios with reconciled inputs and record unresolved items. Keep the cash total, execution estimate, time value and capital at risk visible as separate lines. The next decision should follow from that completed worksheet: which cost needs measurement, which service is necessary, and what activity the stated budget can support.