I think a test page should show a losing result in the same place and the same size as a winning one. The DinexSpace Test Lab page for its crypto bot does that: it publishes a demo run that is negative and says so. I read that page (dinexspace.com/en/protocol/lab) on 10 November 2026, and every number below comes from it.
What the page says
The Test Lab page covers two crypto products still in testing, a TradingView indicator called Futures C2 and a bot called Bot C2 that places orders on Bybit and OKX. It describes itself as publishing "the backtest results for the test period and the current testing status, exactly as they are." The product is labelled "In testing, not for sale yet".
For the bot, the page reports a demo run on practice accounts, with virtual money, started on 6 October 2026. The data shown is as of 6 November 2026, after 31 days. Both demo accounts began with a base capital of 3,000. OKX Demo stood at 2,854.7 (−4.8%) and Bybit Demo at 2,850.6 (−5.0%). The page lists the bot orders sent, 91 and 88, and notes that all open positions were on one side, short.
The same page lists a stretch the demo is missing. From 7 to 9 October the bot stopped because of a network proxy problem, since removed.
How the page frames it
Under the numbers is a short block called "Reading notes". It says 31 days is not enough to conclude anything, that demo accounts differ from real money in fills, slippage and funding, that the numbers are negative and are shown as they are, and that all open positions sit on one side, so risk is concentrated. The page ends the section with "No conclusion about performance yet" and a line that no profit is promised.
I like three things about this. The negative figure sits next to the backtest instead of being hidden on another page. The gap caused by the outage is written down, not smoothed over. And the conclusion is left empty because the data does not support one.
Why I work this way
When I deliver automation or a bot, I add a list of what was checked and what was not. A weak result belongs on that list for the same reason a strong one does: whoever reads it has to decide something. If I only show the good runs, the reader learns nothing about how much weight the good runs can carry.
A backtest is also a weak kind of evidence. The same page notes that backtests are historical test data, not live trading results, and the demo run uses virtual money, so it is not a real-money result either. Labelling each number with where it came from is more useful to a reader than a polished chart.
This post covers only what the public page says. I did not check the strategy itself, and I make no claim about whether Bot C2 will ever be profitable.
If you have a process or a bot and want to know what a fair test report would look like for it, send me how it works today and what you would count as a pass.