A backtest does not show the future.
It is how we get better.
We say this deliberately: no test run on past data tells you what happens tomorrow. But the backtest is the most valuable data we have — every parameter, every threshold and every protection layer in the engine reached its present form by passing through thousands of simulations.
Yes, it looks like we have built an engine that performs very well. Perhaps we have. But we never forget for a moment how unforgiving the market is — and on this page we show you both the numbers and the assumptions they came out of, exactly as they are.
Every number on this page is is simulation output. It is not realised customer return and cannot be taken as an indication of future performance.
What these numbers mean, and what they do not
- The engine was calibrated in earnestEvery threshold was tested on its own and together; not a single number was picked at random.
- Costs includedWith commission, slippage, funding, partial liquidation and cross-margin effect all accounted for.
- It held up under stressAcross runs repeated with random start dates, behaviour stayed consistent.
- This is what you will earnA simulation result is not a promise. Your result will be different.
- No chance of lossA market move never seen before lies outside every model.
- The market has resolvedNot solved. Calibration is continuous work, never finished work.
10 slots · 20 days · 50 simulations
Fifty independent runs with random start dates. The box below shows, exactly as it is, the assumptions the engine ran under.
The share of loss caused by liquidation in this run %0.00 came out. The reason is not luck: two-layer protection steps in twice before the exchange's liquidation level is reached.
Same engine, more aggressive capital
Raising order size lifts the peak, but it lifts the liquidation count too. We are not hiding this table — aggressiveness is not free.
| # | START CAP. | PEAK D. | PEAK $ | PEAK % | FINAL $ | BUST |
|---|---|---|---|---|---|---|
| 1 | 36.5 | 56.5 | $28,396 | 2739.6% | $28,340 | no |
| 2 | 47.3 | 67.3 | $24,174 | 2317.4% | $23,915 | no |
| 3 | 36.1 | 56.1 | $23,334 | 2233.4% | $23,276 | no |
| 4 | 35.9 | 55.9 | $22,321 | 2132.1% | $22,320 | no |
| 5 | 9.3 | 29.3 | $21,829 | 2082.9% | $21,754 | no |
| 6 | 7.9 | 26.9 | $21,739 | 2073.9% | $20,468 | no |
This table best runs; the average run is far below this, and the median and p10 values are stated plainly in the boxes above. Do not judge a system by its best day — we don't either.
Six accounts, six different stories
We ran the engine separately with different capital, different baskets and different market days. Six of those runs are shown below exactly as they came out — including the two that closed at a loss. Showing only the ones that ended well would be easy; the instructive ones are the others.
These accounts were tested on different market days and with different coin baskets; they are not repeats of one another. These are our own test runs, not customer accounts and cannot be taken as an indication of future performance.
Sometimes it will look as though nothing is happening
You will look at the panel, the slots will be idle, and not a single trade will open for hours. Don't worry — that is not a fault, it is a decision.
Not losing is how you survive in this business. And surviving is winning. What matters is not that the engine opens a trade in every condition, but that when conditions are unsuitable it not opening is expected.
Let us be plain: we are extremely fussy about this. At the slightest uncertainty we pull you aside. A missed opportunity can be made up; lost capital cannot always be.
Every idle slot shows as "awaiting signal" in the panel, and the reason it is waiting is written in the SYS.LOG stream. We never go quiet.
Every figure on this page comes from simulations run on historical data; it is not realised customer return and cannot be taken as an indication of future performance. Leveraged derivative trading can result in the loss of all capital. XTrad does not provide investment advice, does not manage portfolios and makes no promise of profit. Do not use money you cannot afford to lose.
You have seen the numbers. Now try it in real conditions.
One-week Starter trial · your own account, your own server.