Numbers you can audit
Crypto Backtesting Platform
A backtest is only as honest as its assumptions. QANTERION publishes what its simulations model and what they leave out — then gives you paper forward-testing as the out-of-sample exam.
A crypto backtesting platform replays a trading strategy against historical market data to estimate how it would have behaved. Its output is only as credible as the frictions it models: omit fees, slippage, or realistic fills and the result is not a conservative estimate but a systematically flattering one. QANTERION publishes which frictions its simulations model and which they deliberately leave out, defines the formula behind every reported metric, and treats the backtest as a rejection tool rather than a forecast.
Backtesting replays a strategy against historical data to estimate how it would have behaved. It is the fastest way to reject a bad idea — and, done dishonestly, the fastest way to manufacture confidence in one. Look-ahead bias, overfit parameters, zero fees, and perfect fills can make almost any rule set look profitable on the data it was tuned on.
The fix is not a more impressive equity curve. It is stated assumptions: which frictions are modeled, which are not, and who owns each number. That is the standard this page describes, and the standard the methodology page documents line by line.
Each friction, and what happens when a backtest omits it
Every omission moves the result in the same direction — flattering. That is why an unfrictioned backtest is not imprecise but systematically wrong.
| Friction | Modelled here | If it is omitted |
|---|---|---|
| Slippage | Priced against volatility and order size rather than assumed away | Tight-stop and high-frequency systems look profitable and are not |
| Maker / taker fees | Charged on every simulated fill at the tier actually paid | Any strategy with a thin per-trade edge passes a test it should fail |
| Order book depth | Depth-aware matching decides whether size fills at one level or walks several | The test assumes infinite liquidity at the top of the book |
| Metric definitions | Published formula for Sharpe, max drawdown, and exposure | Two people read the same run and see different numbers |
What the simulation models
Slippage models
Market orders do not fill at the last traded price. High-precision slippage modeling prices that gap instead of assuming it away.
Maker / taker fees
Every simulated fill pays the fee structure it would have paid live. Strategies whose edge is thinner than their fees fail here first.
Depth-aware matching
Order book depth decides whether a size fills at one level or walks several — modeled, so a backtest does not assume infinite liquidity at the top of the book.
Defined metrics
Sharpe ratio, max drawdown, exposure: every reported metric has a published formula, so two people reading the same run see the same number.
What is deliberately not modeled
Every backtest engine leaves things out. The difference is between leaving them out silently and stating the omission next to the result.
- Market impact of your own order flow — a simulation cannot know how the book would have reacted to you.
- Exchange outages, wick anomalies, and data gaps in the historical record.
- Anything the strategy execution core does not itself model — the boundary of each assumption is documented, not smoothed over.
- The future. History is the dataset the strategy may have been tuned on; passing it is necessary and nowhere near sufficient.
How to read a backtest without fooling yourself
- 1 Check what was assumed Fees, slippage, funding, fill logic. An unstated assumption is a favourable one.
- 2 Read drawdown before return The depth and the duration. A curve is only holdable if you would have held it.
- 3 Check for in-sample tuning Were the parameters chosen on the same data the test ran on? If so, the result is a description, not evidence.
- 4 Forward-test on paper The strategy sees only the present, and every fill is out of sample. That is the exam a backtest cannot give.
Related tools and reading
QANTERION does not guarantee investment returns. Backtests and simulated results describe behaviour under stated assumptions and do not predict live outcomes. Live execution and real-money funding are disabled by default and require explicit enablement.
Questions worth asking first
The honest version, not the landing-page version.
Does a good backtest mean a strategy will be profitable?
No. A backtest estimates behavior on historical data under stated assumptions. Parameters tuned on that same history can make a weak strategy look strong. QANTERION treats backtests as a rejection tool, and pairs them with paper forward-testing on unseen live data.
What assumptions does QANTERION publish?
Slippage modeling, maker/taker fee handling, depth-aware fill matching, and the formula behind every reported metric. The methodology page lists each assumption and marks which ones belong to the strategy execution core rather than to QANTERION itself.
Which metrics should I look at first?
Max drawdown before return, always. Then Sharpe ratio for risk-adjusted consistency, exposure for how much capital was actually at risk, and fee drag for whether the edge survives costs. Return is the last number to trust, not the first.
Can I verify a number myself?
Yes. The Sharpe ratio and max drawdown calculator is free and unauthenticated — paste your own series and check the math. Every metric formula used in the terminal is published in the methodology.
What does access cost?
Terminal access follows Paper Free ($0), Starter ($29 / month), and Pro ($79 / month). A subscription purchases terminal access — not investment returns, managed trading, or a share of strategy performance.
Reject bad strategies faster
Published assumptions, defined metrics, paper forward-testing — and no backtest presented as a promise.