Almost every crypto trading bot you'll ever see advertises a track record. Almost none of them can actually be verified. The number on the landing page is a screenshot, the chart only goes up, and the word "verified" links to nothing. If you can't check it yourself, it isn't a track record — it's a marketing asset.
This is a practical guide to telling the two apart: the four tests a real, auditable record has to pass, a quick red-flag checklist, and exactly how to verify one with tools you already have.
Why a bot's "track record" is usually fiction
Four tricks do most of the damage, and they're so common they've become the industry default:
- Backtest dressed up as live. A curve fit to past data, presented as if the bot earned it in real time. A backtest can be tuned until it looks perfect; that tells you nothing about tomorrow.
- Cherry-picked windows. The screenshot shows the three good months and crops the six bad ones. Survivorship does the rest — the losing configs quietly disappear.
- Editable numbers. A dashboard figure with no signature can be changed at any time, retroactively. You have no way to know what it said last week.
- Costs left out. Results computed without fees, slippage, or funding look dramatically better than anything you could actually achieve.
None of these are verifiable. They ask for faith — and the trading-bot industry has spent a decade teaching people that faith is expensive.
The four tests of a verifiable track record
A record you can actually trust passes all four of these. Use them on any bot, including ours.
1. Signed and dated — so it can't be edited after the fact
The report should carry a cryptographic signature created with a private key the vendor keeps offline, plus the date it was signed. With the matching public key, anyone can confirm the file hasn't changed since. No signature means the number can be quietly rewritten whenever it's inconvenient.
2. Live, not backtest
The results should come from the strategy running on live market data in real time — even in a simulated (paper) portfolio — not from a curve fit to history. A backtest is a hypothesis; live behaviour is evidence.
3. Every trade, not a flattering window
The honest unit is the full trade set over a stated period — winners and losers, drawdowns and all — not a hand-picked range. If the methodology doesn't say "this is all of it," assume it isn't.
4. Re-checkable by you
You should be able to reproduce the verification with standard tools, not take a "✓ verified" badge on faith. If the only proof is the vendor's own word rendered in green, it isn't proof.
A bot you can't audit is a bot you have to trust on faith. The entire point of a verifiable record is that you never have to.
How Quantor publishes results you can verify
We built Quantor so you never have to take our word for the numbers. The mechanism is deliberately boring and checkable:
- Ed25519-signed monthly snapshots. Each production strategy runs as a PAPER portfolio on live market data. At month end the results are signed with an offline Ed25519 key and published on /performance — both the human page and a machine-readable file plus its signature.
- A public key you can check against. The
verification key is published, and the
signing mechanism is
documented end to end. The verification recipe uses
openssl— nothing proprietary. - An immutable trade record. Every closed trade is written to an append-only ledger and chained, so the snapshot is backed by a record that can't be silently rewritten.
- A per-decision journal. You can watch the bots think in real time on the live demo — every signal, gate, and action, as it happens.
The public key and the full method are live today. The first finalised, signed monthly snapshot publishes on 2026-07-01 — the first complete calendar month under the current configuration. We'd rather tell you that plainly than back-date a "track record" to look older than it is.
Red flags: when a track record can't be trusted
- Results shown as a screenshot or a bare dashboard number, with no signature and no public key.
- A "verified" badge that doesn't link to a method you can reproduce.
- A backtest presented as live performance — or no statement of which one it is.
- Only winning months on display; no drawdown, no losing periods, no full trade count.
- No mention of fees and slippage in how the numbers were computed.
- A headline yield or "X% per month." The compounding math makes sustained claims like that implausible — we wrote about why we don't promise yield.
Verifiable is not the same as profitable
Here's the part most marketing pages would never print: a verifiable track record does not mean a profitable one. Our own walk-forward testing across strategies, instruments, and timeframes — after realistic costs — does not show a durable edge over simply holding the asset. What it consistently shows is lower drawdown: a risk-reduction property, not a money-printing one.
So we don't promise returns. We promise that whatever the bots do — good month, bad month, a misconfiguration we had to correct — you will always be able to see it, dated and signed, and check it yourself. That's the whole product. When a bot underperforms, we publish that too.
See the method and (from 2026-07-01) the signed snapshots on /performance; the signing details are on /security. Watch the bots decide live at the demo. Crypto trading carries real risk of loss — Quantor does not guarantee returns. Found a flaw in any of this? bugs@quantorsaas.app.