Your Crypto Bot Break-Even Is a Number, Not a Feeling
The crypto bot break-even point is the average gross edge, per round trip, that your strategy must earn before it stops destroying capital. It is arithmetic, not opinion. Throughout this article break-even is calculated pre-tax, from exchange fees, spread and slippage only, with tax handled afterwards as a separate layer applied to net profit. You can calculate it in about ten minutes with your exchange’s fee schedule and a calculator, and most retail traders never do.
Here is the uncomfortable part. Once you calculate it, a large share of the strategies sold to Australian retail traders — scalpers, tight grid bots, high-frequency mean-reversion systems — turn out to require an edge that professional market makers would be pleased with. Not slightly better than break-even. Several multiples of what a retail strategy on a retail fee tier can plausibly produce.
This article works the arithmetic. It is not a survey of what algo trading costs. It is the specific question of what per-trade edge clears the friction at a given trade frequency, and why that friction compounds so brutally as frequency rises.
This article is general information only. It does not account for your objectives, financial situation or needs, and it is not financial, investment or tax advice. Automated crypto trading carries a real risk of losing some or all of your capital, and past or modelled results are not a guide to future performance. Consider seeking advice from a licensed financial adviser and a registered tax agent before acting.
The Three Costs That Set the Break-Even Line
Three costs set the break-even line: exchange fees, spread and slippage. All three are charged per trade, which is why frequency is the variable that matters most. Tax is deliberately excluded from the break-even figure, because it falls on net profit rather than on trades, and is applied as a separate layer once the pre-tax hurdle is known.
- Exchange fees. Charged per side. A round trip pays them twice.
- Spread. If you take liquidity you buy at the ask and sell at the bid, paying half the quoted spread on each side, so a complete round trip costs roughly one full quoted spread.
- Slippage. The gap between the price your bot saw and the price it got. It grows with order size relative to book depth, and it is worst exactly when your signal fires, because everyone else’s signal fired too.
- Tax, which is not part of break-even. Charged on net profit, not on trades, so it cannot raise the point at which losses stop. It is layered on afterwards, in its own section.
Some venues present a zero-commission interface and take their revenue in the spread instead. That is not cheaper. It is the same cost wearing a different label, and it is harder to model because the spread moves.
Fixed costs sit on top. A bot subscription, a VPS, a charting plan and a tax tool all carry monthly charges, and the total varies widely by vendor, so price your own stack rather than borrowing a figure. On an illustrative input only: if that stack cost fifty dollars a month on a $5,000 account, it would be a 12% annual hurdle before a single trade executes. That is a fixed drag rather than a per-trade one, but it stacks on everything below.
Working the Crypto Bot Break-Even Arithmetic
On favourable assumptions, total friction lands near 0.35% per round trip — roughly 0.20% in exchange fees, 0.10% in spread and 0.05% in slippage. That is the number the rest of this section builds on. Anything worse than favourable pushes it higher.
Take a deliberately favourable case. Assume a liquid AUD pair on an order-book exchange, a fee schedule around 0.1% per side, a quoted spread near 0.1% of which the round trip pays one full crossing, and modest slippage of about 0.05% across the round trip. These are illustrative inputs chosen to make the arithmetic legible, not quoted or observed rates, so substitute your own venue’s actual schedule before relying on any figure below.
| Component | Cost per round trip |
|---|---|
| Exchange fee, 0.10% on each side | 0.20% |
| Spread, 0.05% on each side (one full 0.10% quote) | 0.10% |
| Slippage | 0.05% |
| Total friction | 0.35% |
So this bot must earn 0.35% of gross edge on the average round trip just to stand still. Thirty-five basis points. Now multiply by frequency, because that is where the arithmetic turns ugly.
| Round trips | Per year | Friction vs capital deployed | Ending balance with zero gross edge |
|---|---|---|---|
| 2 per week | 104 | 36% | about 69% of starting capital |
| 10 per week | 520 | 182% | about 16% of starting capital |
| 5 per day | 1,825 | 639% | effectively nil |
Read the last column again. A strategy with genuinely zero edge — a coin flip, net of direction — that runs ten round trips a week ends the year having lost roughly 84% of the capital it cycled. Nobody had to be wrong about the market. The friction did all of it.
Flip the calculation to find the required edge. To finish the year up 10% while running 520 round trips, each trip must clear friction and add a little on top. The compounded arithmetic lands at roughly 0.37% of average gross edge per round trip, sustained across every trade for twelve months. Consistently capturing 37 basis points of gross move, after adverse selection, is not a retail outcome.
Cost as a Share of the Move You Are Chasing
The single most useful ratio is friction divided by target move. It tells you what proportion of your strategy’s theoretical profit is consumed before you see any of it. Below, the same 0.35% friction is measured against different profit targets.
| Target move per trade | Friction consumed | Verdict |
|---|---|---|
| 0.3% | Over 100% | Unprofitable before the signal is even correct |
| 0.5% | 70% | Requires a near-perfect win rate |
| 1.0% | 35% | Marginal; one bad month erases the year |
| 3.0% | 12% | Workable with a genuine edge |
| 10% | 3.5% | Friction is a rounding error |
Now consider a spread-priced platform instead. Venues that advertise zero commission and price their margin into the quote commonly cost a multiple of a competitive order-book taker schedule, though the multiple varies by venue, pair and time of day and is not something a general figure can pin down. Whatever the multiple, it scales every row of the table above proportionally — double the friction and you double the share of each target move consumed. Obtain a live two-way quote on your own pair and compare it against a published taker schedule.
This is the whole argument in one line. Friction is roughly constant per trade, so your break-even is set by how big a move you are reaching for, not by how clever the signal is.
Tax Does Not Move Break-Even, It Moves What Break-Even Buys You
Tax does not change your break-even point; it changes the gross return needed to reach a given after-tax outcome. You are not taxed on a losing trade, so tax cannot raise the point at which you stop losing money. Treating it as another per-trade fee double-counts it and produces a hurdle rate that is too high.
The gross-up is a simple formula: required pre-tax return equals your after-tax target divided by one minus your marginal rate. For many resident taxpayers the marginal impost is often modelled at about 32%, but actual tax depends on circumstances including Medicare levy settings and offsets. Using 32% as a modelling input, keeping 10% after tax means earning roughly 14.7% after all costs, so the bot has to be materially better than break-even rather than marginally better. Substitute your own marginal position before treating that gross-up as your hurdle.
Not every disposal is a CGT event. An investor holds crypto as a CGT asset, so disposals are CGT events, and under current law the CGT discount requires holding an asset for more than twelve months — a threshold almost no bot strategy reaches. Someone carrying on a business of trading holds the same coins as trading stock on revenue account, with proceeds assessable as ordinary income.
Trading frequently, in large volumes, or with sophisticated tools does not by itself make you a business. The ATO applies the general business indicia — commercial purpose and viability, organised business-like conduct, repetition and regularity — and classification follows from your actual circumstances rather than from how you label yourself. Revenue treatment is not simply worse: it removes any prospect of the CGT discount, but losses may be deductible against other income rather than quarantined against capital gains. Confirm your treatment with the ATO’s guidance or a registered tax agent.
Two timing traps are worth flagging. Capital losses for an investor offset capital gains rather than salary income, so a bot that wins in March and loses in May does not simply net out against your wages. And an unrealised loss sitting on the books at 30 June does nothing to shelter realised gains banked earlier in the year.
From 1 July 2027 the 50% CGT discount no longer applies to crypto assets held by individuals, trusts or partnerships. Under the Treasury Laws Amendment (Tax Reform No. 1) Act 2026, which received Royal Assent on 26 June 2026, it is replaced by cost base indexation — the cost base is adjusted for inflation, so only the real gain is taxed — together with a 30% minimum tax rate on that real gain. This is a general CGT change applying to shares and property in the same way, not a crypto-specific measure, and transitional rules preserve the existing 50% discount for gains accrued up to 30 June 2027.
Separately, digital asset businesses providing financial services must take a qualifying step by 30 September 2026 — lodging an AFS licence application or variation, or putting an authorised representative arrangement in place — to fall within ASIC’s no-action position, a policy decision rather than a legal opinion, which ASIC may withdraw or revise. Neither changes the break-even arithmetic; both change the planning environment around it. ATO guidance on the technical application of the new CGT rules is still being finalised, so confirm the detail with a registered tax agent.
Why High-Frequency Retail Bots Usually Lose
High-frequency retail bots lose because they sit on the wrong side of every term in the equation at once. Retail fee tiers are the highest available. Retail order flow is the least informed. Retail infrastructure is the slowest to the book.
Consider what a high-frequency strategy is actually claiming. It says the market misprices an asset by a fraction of a percent, repeatedly, and that a bot polling a public API from a suburban VPS will detect and capture that mispricing before firms with colocated servers and rebate-tier fee schedules do. Stated plainly, the claim is implausible.
Then there is adverse selection. Your limit orders fill fastest when the market is moving against you and go unfilled when it moves your way. Backtests rarely model this, which is why a backtest assuming mid-price fills and ignoring queue position produces results live trading never reproduces.
Rate limits close the loop. Exchange APIs generally impose request and order-rate limits, and where a bot’s polling or order churn exceeds them it can be throttled — a throttled high-frequency strategy is just a slow strategy paying high-frequency costs. Limits vary by venue and endpoint and are revised without much notice, so read your exchange’s current API documentation rather than assuming headroom. If your edge depends on speed, you have entered a competition you cannot win.
None of this means automation is pointless. It means the profitable region sits at lower frequency and wider targets, where friction is a small share of the move and your structural disadvantages matter less.
How to Calculate Your Own Crypto Bot Break-Even Before Deploying
Calculate your crypto bot break-even in seven steps, before you fund an account: find your true per-side fee, measure the spread, estimate slippage, sum the three, divide by your target move, multiply by annual trade count, then add fixed costs and gross up for tax. It takes ten minutes and it has prevented more bad deployments than any backtest ever has.
- Find your true per-side fee. Use the taker rate unless your bot exclusively posts and you have verified the fill rate. Assumed maker fills that never happen are a common source of phantom edge.
- Measure the spread on your actual pair, at your actual trading hours. Australian overnight liquidity is thinner than the daytime screenshot suggests.
- Estimate slippage from real fills. Compare intended price against executed price across at least fifty live trades, because paper trading understates this badly.
- Add the three together. That total is your friction per round trip.
- Divide friction by your average target move. Above roughly one third, the ratio implies a fragile strategy. Above one half, more than half the theoretical profit is consumed before the signal is even correct.
- Multiply friction by expected annual trade count. Compare that total against a return you would call excellent — 10% a year is the benchmark used throughout this article. If annual friction exceeds it, the arithmetic has already answered the question.
- Add fixed costs as a percentage of account size, then gross up for tax. Steps one to six give you the pre-tax break-even; this last step converts it into the after-tax target you actually need.
If a strategy only clears that hurdle under optimistic assumptions, it does not clear it. Rerun the numbers with fees one tier worse, spread doubled and slippage tripled. A strategy surviving that treatment has a real margin of safety. One that does not was always relying on conditions that will not hold.
The arithmetic points somewhere unfashionable. Because friction is roughly fixed per round trip, fewer trades against larger targets on liquid AUD pairs leave a smaller share of each move consumed by cost than the same signal run at ten times the frequency. That is an implication of the cost structure, not a promise about returns — no trade frequency or target size makes a strategy profitable if the underlying edge is not there.
Frequently Asked Questions
What is a realistic break-even cost per round trip for an Australian retail bot?
No general figure is reliable, because fee schedules, spreads and depth differ by venue, pair and time of day and change without notice. The worked example above uses roughly a third of a percent per round trip as an illustrative input for a liquid AUD pair on an order-book venue, and spread-priced platforms commonly run a multiple of that. Measure your own from a live quote and your published fee tier, because the gap between venues compounds enormously at volume.
Does a maker-only strategy solve the break-even problem?
It helps, but less than the fee table implies. Posting rather than taking reduces the fee and can earn you the spread instead of paying it, which is a genuine improvement. The catch is adverse selection, since the orders that fill are disproportionately the ones you would rather had not.
How many trades per year is too many for a retail bot?
There is no fixed threshold, but the same ratio used in the worked example gives you one. Once annual friction — cost per round trip multiplied by trade count — exceeds a return you would describe as excellent, the strategy is fighting its own cost structure. Take 10% a year as that excellent-return benchmark and 0.35% as friction per round trip: 100 round trips consumes about 35% of deployed capital, and the point where friction alone equals the whole 10% target is 10 divided by 0.35, or roughly 29 round trips a year. So on those illustrative inputs the boundary sits near thirty round trips annually, and it moves in direct proportion to whatever friction figure your own venue produces.
Should tax be included in the break-even calculation itself?
No, and including it produces a misleading number. Break-even is the point where gross edge equals costs, and no tax is payable there. Model tax as a separate gross-up applied to your target return once the pre-tax hurdle is cleared.
Why do backtests show profits that live bots never produce?
Most backtests fill at the mid-price, assume unlimited depth, ignore queue position and apply a single optimistic fee. Each of those assumptions quietly removes part of the friction described above. Re-run any backtest with realistic taker fees, a full spread crossing and pessimistic slippage before believing the equity curve.
Does a bigger account improve the break-even arithmetic?
It improves fixed-cost drag substantially and fee tiers modestly. Subscriptions and hosting become trivial as a percentage of a larger balance. Slippage moves the other way, because larger orders eat deeper into the book, so size helps less than traders expect.
