Intellectia.AI Review: 200% Backtested, Zero Risk Disclosure

1. The four ways historical simulations flatter a strategy

Backtesting is a legitimate research tool. It becomes marketing when the output is presented as proof. Four well-documented effects push simulated returns above what the same rules deliver in the future, and all four are invisible in a headline number.

• Overfitting. Run enough parameter combinations against history and some will look spectacular by chance. The more variations tested, the more certain it becomes that the winner is fitted to noise rather than signal.
• Survivorship bias. Historical stock universes often exclude companies that went bankrupt or were delisted. A strategy backtested on today's surviving companies has been quietly told which firms not to worry about.
• Look-ahead bias. If the simulation uses data that was not actually available at the moment of the simulated trade — restated earnings, later-revised figures, index membership known only in hindsight — it is trading with information from the future.
• Costs and slippage. Commissions, spreads, and the price impact of actually buying are frequently omitted. On a weekly-rebalanced ten-stock portfolio, which is what the AI Stock Picker describes, those costs are not a rounding error.

A 200% annualised return would place a strategy far above the long-run record of essentially every professional investor who has ever operated. The reasonable inference is not that the model is extraordinary. It is that the simulation has one or more of the problems above, and no information on the page lets a reader check which.

2. The missing sentence, and why it is standard

Past performance is not indicative of future results appears on fund factsheets, brokerage advertising, newsletters and trading platforms across the industry. It is there because regulators, litigation and hard experience have all established the same thing: consumers systematically read historical returns as forecasts. Where the performance shown is hypothetical or backtested rather than actually achieved, the convention is stronger still — such figures are normally labelled hypothetical and accompanied by an explanation of their limitations.

We searched for all of it. Past performance: zero. Hypothetical: zero. Risk of loss, lose money, loss of principal: zero. Not investment advice: zero. Informational purposes and educational purposes, the usual framing for research tools that wish to stay clear of advice: zero. Those searches covered a terms of service of over 26,000 characters and a privacy policy of over 32,000, so this is not a case of a short site with little text.

We also checked, with word boundaries to avoid false matches, for SEC, RIA, FINRA, registered investment and broker-dealer. All zero. We make no claim about whether registration is legally required here — that turns on how the service is structured and on facts we cannot see. What we can say is that a platform issuing trade signals and executing orders through connected brokerage accounts tells the reader nothing at all about its regulatory position.

3. When the model is wrong and the account is connected

There is a meaningful difference between software that tells you what it thinks and software that acts. Intellectia describes both: a one-minute setup to link Webull, Alpaca and others, execution of market and limit orders from inside the research dashboard, and a copy-trade product that syncs real-time AI signals automatically so that, in its words, no manual analysis is required.

It also claims the system pauses trading during high-risk conditions to protect your capital. That is a substantial promise about behaviour in exactly the conditions where models fail most often — regime changes, gaps, liquidity droughts. No detail is offered about how high-risk conditions are detected or what the pause does to open positions.

Add the language-model dimension and the risk compounds. Systems built on large language models produce fluent, confident text whether the underlying number is right or invented; a hallucinated earnings figure or ratio reads exactly like a correct one. Anything an AI asserts about a company should be verified against a filing before money moves — which erodes the time-saving that is the product's main practical selling point.

4. The one connection to this site, and it is a cautionary one

Readers of a GLP-1 site have watched a live demonstration of why thematic momentum trading is hard. The obesity drug story has been one of the defining market narratives of the decade, producing enormous gains for the companies involved — and then severe drawdowns when trial readouts disappointed and competitive dynamics shifted. Investors who arrived late, on the strength of a trend that had already run, learned what that costs.

That is exactly the pattern a momentum-driven weekly picker is built to chase. A model selecting the top ten stocks each Monday on recent strength will load into precisely the themes that have already moved, and will be holding them when the readout misses. If you have followed the GLP-1 sector as a patient, you have front-row evidence for why the 200% number deserves scepticism rather than enthusiasm.

Why is a GLP-1 site reviewing a stock trading app?

It came through our affiliate queue and there is no health angle. We are reviewing it as a consumer purchase because it is the kind of product where the marketing and the risk are furthest apart, and because there is one genuinely apt connection: GLP-1 drugs have been among the most dramatic market stories of the decade, producing enormous gains and then severe drawdowns when trial results disappointed. If you want an illustration of why an AI momentum picker on a hot theme is dangerous, this audience has lived through one.

What is the 200% claim?

The homepage states, twice, that the AI Stock Picker strategy has proven, via backtesting, to achieve an over 200% annualised return. Backtesting is a simulation of how a strategy would have performed on historical data. It cannot prove anything about future returns, and the phrase proven via backtesting is close to a contradiction in terms. Backtests systematically flatter strategies through overfitting to past data, survivorship bias in the stock universe, look-ahead bias, and the omission of trading costs and slippage.

What disclaimers does it carry?

None that we could find. We searched the homepage, the pricing page, a 26,000-character terms of service and a 32,000-character privacy policy for the phrases past performance, not investment advice, investment advice, risk of loss, lose money, loss of principal, hypothetical, informational purposes and educational purposes. Every one returned zero matches on every page. Past performance is not indicative of future results is the most standard sentence in all of consumer finance, and it does not appear.

Is it a registered investment adviser?

We found no claim either way. Searching all four pages with word boundaries for SEC, RIA, FINRA, registered investment and broker-dealer returned zero matches. We are not asserting that registration is required — that depends on facts we cannot assess from outside. We are reporting that a service generating trade signals, offering copy trading and executing orders through linked brokerage accounts says nothing anywhere about its regulatory status.

Does it actually place trades?

Yes, if you connect it. The site describes linking brokers including Webull and Alpaca, executing market or limit orders without leaving the dashboard, and a SwingMax Copy Trade feature that syncs and copies real-time AI signals automatically. That is materially different from research software. Once signals reach a live brokerage account, the gap between a bad model and real money disappears.

What does it cost?

There is a genuinely free tier and a $1 trial, which we credit. Paid products run from $29.99 a month for the Backtest Playground, through $199.95 for six months or $319.95 a year for the whales tracker, up to the SwingMax Portfolio at $499.99 a year, discounted from a stated $799.99. Tiers are labelled Free, Basic, Pro, Max and Expert. The SwingMax description says it is ideal for investors with at least $10,000 in capital.

What about the testimonial claiming 22.6% a month?

Take it as a warning rather than a selling point. A homepage testimonial states a return of 22.6% per month, up from 7.9% before. Compounded, 22.6% monthly is well over a thousand per cent a year — a rate that would make the individual concerned one of the most successful investors in recorded history within a few years. Testimonials are not evidence, unverified individual returns least of all, and displaying that figure without any accompanying risk statement is the clearest signal on the page.

Can an AI model just make things up about a stock?

Yes, and it is a real risk with any language-model-driven analysis. These systems generate fluent text whether or not the underlying figure is correct, and a confidently stated but fabricated earnings number or ratio reads exactly like an accurate one. Any AI-generated financial claim should be checked against a filing or a primary data source before you act on it — which somewhat undercuts the time-saving proposition.