Algo-trading and black-box bots

Somewhere in your feed there is an advertisement for a trading bot. It promises to buy and sell for you automatically and make you money while you sleep, for a small monthly fee — ₹2,000, maybe ₹5,000. It shows a chart climbing steadily upward. It uses words like "algorithm," "AI," and "backtested."

This module asks one plain question, and the whole thing turns on it: if this machine genuinely prints money, why would anyone sell it to you for ₹2,000 a month? Not "is the technology real" — automated trading is real and widely used by big institutions. The question is about the incentive. A person who truly owned a reliable money-making machine has a far better option than renting it to strangers. Understanding why they don't take that better option tells you almost everything you need to know.

The master question of this shelf does the work again here: who profits if you believe this — the seller from your trades, or the seller from your subscription? Once you see which one is the real business, the shiny chart in the advertisement stops mattering.

First, some plain words. Algo-trading simply means buying and selling done automatically by a computer program following pre-set rules, instead of a human clicking each order. That part is genuine and ordinary; large funds run most of their trades this way. A trading bot is just a consumer version of this — software sold to individuals that promises to trade on its own. And a real edge is a genuine, repeatable advantage that makes money on average, after costs, across many trades. An edge is the rare and precious thing. A bot is just a wrapper around one — if there is one inside.

Now the crucial fact about a real edge: it is worth an enormous amount to the person who has it, and almost nothing once they share it widely. Suppose a strategy honestly earned even 3% a month. Left to compound on your own capital, ₹10 lakh would turn into several crores within a decade [illustrative] — and anyone with a track record like that can raise near-unlimited money from investors and take a share of the profits. The rational owner of a real edge does not chase ₹2,000 subscriptions. They trade it quietly and get rich from the trades.

There is a second reason they guard it. Most edges have limited capacity — they only work while few people use them. The moment a crowd runs the same strategy, their buying and selling moves the price against the very signal the strategy relies on, and the edge is arbitraged away: competed out of existence by everyone piling into it. So broadcasting a genuine edge is self-destroying. Selling it to thousands of subscribers would kill the thing being sold.

Put the two together and the logic is airtight. If the edge is real, selling it is irrational — you'd trade it yourself and you'd protect it. So a bot marketed hard to the retail public is, by its own incentives, telling you the reliable money is the subscription, not the trades. The customers are the revenue. The trades are the advertisement.

This is the objection that traps the most people, so it deserves a careful answer. The advertisement shows a beautiful chart: "Backtested +300% over five years." Surely that is proof?

No — and here is exactly why. A backtest means testing a strategy on past data to see how it would have done. The problem is that the past has already happened, so you can keep adjusting the strategy's rules until it looks perfect on that specific history. This is called overfitting: tuning a strategy so tightly to old data that it fits every past wiggle and then fails completely on new data it has never seen. An overfit backtest is not a prediction. It is a drawing traced over a photograph — flawless on the photo, useless on tomorrow.

Think of it like this. Give me last year's cricket scores and I can invent a "system" that would have won every bet — bet on this team on Tuesdays, that one after a loss, and so on. It will show a spectacular "backtested" record, because I built it knowing the answers. It will win you nothing next season, because next season is data it was never really tested against. A trading backtest can be exactly this trick, dressed in finance.

The honest test is completely different: real money, traded live, going forward, over a long enough time to include bad periods — and shown transparently, with the losing months left in. That is a live track record. Notice which one the advertisements almost always show, and which they almost never do.

So when the pitch leans on a backtest and shows no audited live record, read it plainly: the dazzling number is a sales tool, not a result. A backtest tells you the seller is good at making a chart look good. It tells you nothing about what your money will do next month.

Take a typical pitch and read the incentive underneath it. illustrative

An account called @QuantEdgeBot runs an ad: "Our AI algo delivered +312% backtested returns. Fully automated. Just ₹2,499/month. Beginners welcome — no skill needed. Limited seats!" A slick equity curve — a line showing an account's value over time — rises from bottom-left to top-right. A countdown timer adds urgency. Testimonials scroll past. illustrative

Now apply the two questions from this module. First: if it works, why sell it? At +312% the owner would compound a modest stake into a fortune and could raise serious money on that record — renting ₹2,499 seats would be beneath the effort. The only world where selling seats is the smart move is the one where the seats are the income. Second: is the 312% a backtest or a live record? The word "backtested" is right there. It is a chart fitted to the past, not money earned forward. No audited live statement is offered.

You have your answer without ever opening the software. The real product is the ₹2,499 monthly fee, multiplied by "limited seats" that are rarely as limited as the timer suggests. The bot's trades are the story that sells the subscription; whether they win or lose, the seller is paid on the first of the month either way. This is the same disconnection you met with finfluencers — the seller's income does not depend on your result — just wearing the costume of technology.

The incentive argument is strong, but a few honest caveats keep it from becoming a blunt instrument.

It does not mean all automation is a scam. Plenty of legitimate tools simply execute your own decisions automatically — placing a stop-loss, rebalancing on a schedule, splitting a large order. These sell a convenience, not a secret money-machine, and they don't promise returns. The red flags in this module are specifically about products that promise profits from a hidden strategy. Automation that just does what you told it to, transparently, is a different animal.

It cannot verify a track record for you. Even a claimed "live" record can be cherry-picked or fabricated. Real verification means audited statements over years, including the bad stretches — something almost no retail bot provides. The absence of that is itself the finding.

And it is worth knowing where the regulator stands, because the rules are tightening. SEBI — the Securities and Exchange Board of India, the market regulator — has moved to bring retail algos under a formal framework: algos offered to individual investors are being routed through brokers and exchanges with registration and approval, and unregulated "guaranteed return" algo products and misleading performance claims are squarely in its sights. That regulation is a reason to distrust anonymous ₹2,000-a-month bots even more, not a reason to assume any given one is safe. Rules reduce the risk over time; they do not vet the specific ad in your feed today.

The same reasoning slips catch buyer after buyer. Named, they are easy to hold onto.

1. Taking a backtest as a result. A chart fitted to the past is a drawing, not a forecast. Without a live, audited record, "+300%" is a design choice.

2. Reading "guaranteed" as reassurance. Real returns vary and include losses; a guarantee of steady monthly profit is a promise no honest market product can keep. The word is a warning, not a comfort.

3. Believing "no skill needed." If the machine truly needed no skill and always won, its owner would never share it. That phrase is aimed at beginners precisely because beginners can't yet see why it can't be true.

4. Mistaking technology for an edge. "AI" and "algorithm" describe how orders are placed, not whether there is a real advantage inside. Fancy plumbing does not create water.

5. Forgetting to ask why it's for sale. The one question that dissolves most of these pitches: a genuine edge is traded and guarded, not rented to strangers by the month. If it's for sale to everyone, the sale is the business.