Industry · 11 min read

Why we don't promise yield (and why every bot that does should worry you)

Pavel · Solo founder · 2026-05-11 · ← All posts

Spend 30 minutes searching for "crypto trading bot" and a pattern emerges. Most landing pages lead with a number. 5% monthly returns. 0.5% per day. 15-30% APY, hands-free. The number is usually printed in green, beside a screenshot of a backtest curve that goes up and to the right. Sometimes there's an asterisk leading to a footnote in 8-point grey type, sometimes not.

Quantor doesn't print a number. We sell software for running an automated trading workflow, not a return profile. We don't have a "projected yield" anywhere on the site, in any plan, in any blog post — and we never will. This post is about why that's a considered position, not an oversight.

It's also a buying-guide post: by the time you're done reading, you should be able to look at any other bot SaaS and tell within ten seconds whether their marketing math is plausible. Most of the time, it isn't.

Quantor sells discipline, not yield. The difference matters because one of them you can deliver every day, and the other you cannot deliver at all.

The math, in one paragraph

Compound returns are non-intuitive. Consider three claims you'll see commonly on bot landing pages:

"5% per month" — compounds to ~80% APY.
"0.5% per day" — compounds to ~520% APY.
"1% per week" — compounds to ~68% APY.

For reference: the historically best public discretionary trader of the last 40 years, Jim Simons of Renaissance Technologies' Medallion Fund, delivered around 40% annualised returns after fees over its multi-decade run. That number is considered an outlier in financial history — the most extreme sustained outperformance the public record contains.

A bot that promises 0.5% daily is implicitly claiming to outperform Medallion by 13×. A bot that promises 5% monthly is claiming to outperform Medallion by 2×. Neither of these is stated honestly anywhere on the product's page. The compounding math is hidden in plain sight.

Why retail users believe it anyway

The math being obviously false isn't enough to stop users from paying for these products. Three psychological mechanisms make the claim feel plausible:

Backtest overfitting. A strategy with enough parameters tuned over enough historical windows can be made to show any backtest curve you want. The user sees a chart, not a methodology, and "the curve goes up" reads as evidence even when the curve is statistical noise after 2000 iterations of parameter tuning.
Cherry-picked windows. The 2017-2018 BTC bull run, the 2020-2021 cycle, the early-2023 recovery — any of these reads as "consistent 5% monthly" if you crop your date axis correctly. Show the same strategy across a full cycle including 2022 and the curve looks different.
"Alpha drift" rationalisation. When the first month delivers 3% instead of 5%, the user is told "market conditions shifted, the algorithm is adapting". When the second month is -7%, "drawdowns are part of the strategy — long-term mean is still 5%". The user has already paid; the cognitive cost of admitting the claim was false is higher than the cost of waiting one more month.

The result is users paying through periods of negative return because the original promise was sticky. The bot operator gets the subscription revenue regardless of the trading outcome. That asymmetry is the actual product.

What the law says (where it applies)

The advertising-fixed-yield problem is well-understood by regulators in jurisdictions that have a securities authority worth the name.

EU. MiFID II requires that any communication relating to investment performance be "fair, clear and not misleading". Advertising a fixed forward yield on a speculative product is plainly misleading; the European Securities and Markets Authority has issued multiple warnings about crypto investment schemes promising fixed returns. MAR (Market Abuse Regulation) further prohibits dissemination of information likely to give false or misleading signals.
UK. FCA financial-promotion rules require that performance claims be evidenced. The crypto-asset financial-promotion regime (Oct 2023 onwards) explicitly forbids advertising fixed forward yield on unregulated crypto products.
US. SEC has sued multiple operators of crypto bots and "yield farms" for unregistered securities offering combined with misleading return claims. Bitconnect is the canonical example; many smaller ones have followed.

The catch: many bot SaaS operators are domiciled in jurisdictions where these rules don't apply or aren't enforced. The product is sold to users who DO live under one of the above regimes, but the company collects fees through a structure that the regulator can't reach. The legal risk falls on the user — not the company — because there's no clean enforcement path.

The corollary: the fact that a bot is allowed to keep advertising 5% monthly without legal consequence is not evidence that the claim is true. It's evidence that the regulator can't reach the company, which is itself a red flag.

How these products usually end

The historical pattern of "high-yield" automated trading products is consistent enough to call it a pattern. There are three terminal states:

The three ways the yield promise ends

Quiet capitulation. Eventually the curve stops going up. The operator switches the marketing focus to "long-term thinking", lowers the implied yield to something less specific ("attractive returns"), and slowly loses users without explicitly admitting the original number was wrong. Some users stay because the cognitive cost of leaving is higher than the cost of staying. This is the most common ending, and the least catastrophic.
Withdrawal freeze. The product was custodial — users deposited funds with the operator. When redemptions exceed inflows, the operator pauses withdrawals "temporarily" while they "rebalance". The pause is permanent and the funds are typically not recoverable. This is the 3Commas / Stoic / Bitconnect pattern: the product was built so it COULD stop withdrawals, and eventually the incentive structure required it to. A non-custodial bot can't do this — but most of the yield-promising ones are custodial.
Regulatory shutdown. The SEC / FCA / ESMA equivalent for the operator's country finally serves paper. The product stops accepting new users; existing users get a percentage of their funds back in liquidation, if any. This is the rarest ending — most operators are structured to avoid this — but when it happens it's the only one with a clear answer to "what should I do now?".

None of these three outcomes are surprises to anyone who's watched the bot industry for more than 18 months. They happen on a cycle. The current set of "5% monthly" products will follow at least one of the three within 3 years; some will follow more than one in sequence.

So what does Quantor promise?

We made a deliberate choice on day one: every claim on the product page must be something a user can independently verify with off-the-shelf tools. Not "marketing", not "directional". Verifiable. The constraint forces us away from yield language and toward specific technical and operational guarantees:

What Quantor commits to

Build identity. Every prod deploy is signed with Ed25519; any user can verify in 30 seconds with openssl that the code running on our server is the code we committed. See the post.
Self-custody. The API key has trade-only scope; we never touch user funds. If Quantor disappears tomorrow, your money sits in your exchange account unaffected. See the post.
Regime gate. When market regime is DANGEROUS, LIVE bot starts are refused regardless of plan or user permission. The classifier and thresholds are public. See the post.
14 risk gates. Before any live start, 14 named checks fire with public reason codes — symbol allowlist, kill switch, plan permissions, exchange-key verification, max notional. See the post.
Audit chain. Every payment, subscription mutation, and admin action is appended to a tamper-evident HMAC-chained log. Users can ask us later "what was the state of my account on day X" and the answer is cryptographically verifiable.
Honest plan limits. Every limit on every plan — max bots, allowed symbols, daily-loss percent, max notional — is pulled directly from the same code that enforces it. We can't lie about the limits because the page is generated from the source of truth.

Notice what's NOT on the list: a number. A target. An expected return. A backtest screenshot. We could put one up; we won't. The constraint that "every claim must be verifiable" eats all of those — return predictions about future markets are definitionally unverifiable.

The cost of this trade

This positioning costs us users. Specifically:

The cohort searching for "5% monthly yield trading bot" bounces off our landing page within seconds — we don't have the keyword they came for. That's a marketing trade-off; we accept the smaller funnel.
The cohort comparing us to bots that DO promise yield sees Quantor as "boring" or "doesn't show ROI". Some of those users go to the bots that promise yield, get burned, and come back. We don't optimise for catching them on the way in.
The cohort looking for "a tool to operate a discretionary strategy with proper risk gates" finds us through search terms like "non-custodial trading bot", "regime-aware crypto bot", "transparent risk policy". They tend to be slower to convert but better at staying — they bought the thing we actually sell, not a thing we don't.

The third cohort is the one we built the product for. The first two are noise. Trading is a tool — a good one — but it cannot deliver fixed yield, and pretending otherwise produces products that exist to extract subscription fees during the part of the curve where users are still hoping. We don't want to sell that. We don't think you should buy it.

A buying checklist (use it on any bot, including ours)

Five questions to ask any automated-trading product before paying. The right answers are specific and verifiable; the wrong answers are vague or absent.

"What's the worst day I can have with this running?" Right answer: a specific dollar number or percentage tied to plan limits. Wrong answer: "drawdowns are managed by our adaptive algorithm".
"Does my exchange API key need withdraw permission?" Right answer: no, trade-only, and here are the exact scopes to enable on Binance. Wrong answer: "yes, for rebalancing" or "we manage that for you".
"If your company disappeared tomorrow, what happens to my funds?" Right answer: they sit in your exchange account, untouched, exactly as they are now. Wrong answer: "we have insurance" or "contact support".
"Can I see the risk policy you apply before any live trade?" Right answer: a public document or endpoint with named conditions. Wrong answer: "it's proprietary".
"What's your projected return?" Right answer: we don't project return; we manage operational discipline. Wrong answer: any specific number.

If three or more of these get a wrong answer, the product is not built for retail users — it's built for the operator's subscription revenue. Walk away.

The promise we don't make is more important than the ones we do. If a competitor offers you a number, ask them to put it in a contract with a clawback. They won't. The number was decorative. Send criticism or counter-arguments to quantorsaas@gmail.com; Pavel reads everything personally.