Make money while you sleep. No experience needed. The bot trades, you collect. It is the most seductive pitch in trading — and it has cost real people staggering amounts of money. Before you hand your account to a bot, look at what actually happened to the people who did.
Start small. On the MQL5 forum, a trader reported losing more than $10,000 testing one system after another for a year, with the largest drawdowns always arriving at the end. On WarriorForum, a trader with ten years of experience wrote that over two years he tested more than 30 forex robots on small live accounts — and nearly all of them ended up blowing the account. These are not beginners making beginner mistakes. These are people who did the homework and still watched bot after bot go to zero.
Then the scale jumps. In 2020, the U.S. Commodity Futures Trading Commission (CFTC) charged two Florida companies and their founder with raising $1.6 million from hundreds of customers for a forex algorithm called ART that was marketed as ready for release, with a high success rate of predicting trades. It was not.
In 2024, the CFTC brought a case against a Louisiana man, Michael B. DePetrillo, and his firms Meteor LLC and NOLA FX Capital Management. The court found they fraudulently solicited and misappropriated more than $3 million from investors, issued false account statements, and ran what resembled a Ponzi scheme. DePetrillo pleaded guilty, was sentenced to nine years in prison in July 2025, was ordered to pay $6,748,412 in restitution, and was hit with a $9.2 million civil penalty.
EminiFX was bigger. The CFTC case against Eddy Alexandre and EminiFX describes tens of thousands of investors sending in contributions exceeding $262 million, lured by reported returns of 5% to 9.99% per week. The accounting told a different story: net losses exceeding $49 million, at least $15 million misappropriated by Alexandre himself, and new investors' money used to pay earlier ones. In August 2025, a federal judge ordered $288 million and wrote, plainly: EminiFX was a Ponzi scheme.
And then there is the largest one. Mirror Trading International told pool participants an "Advanced Intelligence Software" bot was trading forex for them. According to the CFTC case, the scheme misappropriated 29,421 Bitcoin — valued at over $1.73 billion — from participants, including at least 23,000 victims in the United States. In April 2023, a federal court in Texas ordered founder Cornelius Johannes Steynberg to pay $1,733,838,372 in restitution and the same amount again as a civil penalty.
It did not stop there. In September 2026, the U.S. Securities and Exchange Commission (SEC) charged two groups that took at least $15 million from hundreds of retail investors by renting out bots "programmed with artificial intelligence" and promising guaranteed profits. The SEC alleged there were no AI trading bots at all. In a separate May 2026 case, the SEC alleged another bot operation took at least $6.2 million for personal expenses and used about $5.5 million for Ponzi-like payments, while calming investors with fake account statements. The FBI's Internet Crime Complaint Center reported that crypto investment fraud was the largest source of losses it received in 2025, at $7.2 billion.
Not every blown account is a fraud. Most bots fail for honest, structural reasons — and understanding them is what separates a careful trader from the next victim.
Martingale and grid math. The most popular bot strategies double position size after every loss (martingale) or stack orders at fixed intervals (grid). Risk grows exponentially while rewards grow linearly. A $5,000 account running the same strategy as a $500 account does not get ten times safer — it gets a wipeout ten times bigger. Martingale systems do not show a pattern of wipeouts you can plan around; they can go years and then wipe an account six times in a row. As one veteran reviewer put it: if you want to trade a martingale, you would have better odds betting your whole account on black at roulette.
Backtests lie. A backtest is a story about the past, optimized on the past. Bots are routinely curve-fitted: parameters tuned until the historical equity curve looks beautiful. Live markets do not repeat the exact past. Add latency and slippage — the gap between the price the bot wanted and the price it actually got — and a strategy that looked profitable on paper quietly bleeds in real trading. A few pips of slippage per trade can amount to thousands of dollars a year.
Regimes change. A bot built for a trending market dies in a ranging market, and vice versa. The "set and forget" promise is the core lie: market conditions change, and a bot that worked for a year can stop working in a week. The trader who cannot tell which regime they are in cannot tell when their bot has stopped working — until the account tells them.
Here is the honest answer: not all bots are scams. Real automated systems exist, and they tend to share the same boring traits: modest returns, a developer with a history that predates the product, a verified live track record of six months or more on a regulated broker, real risk controls in the settings, no martingale, no grid, and real user reviews on independent forums. The rule of thumb used by experienced reviewers is simple: three red flags means do not run it, and a fail on verified track record or on martingale means do not run it regardless of how good everything else looks.
And one more rule: demo it for thirty days before you fund it. A bot that refuses a trial is telling you everything you need to know.
A bot does not remove the need to understand the market — it increases it. If you cannot explain why your bot wins, you cannot tell when it stops winning, and you will discover it the same way thousands of traders already have: from the account statement. Trust is not something a bot's marketing page gives you. It is something a verified track record, a transparent strategy, and your own understanding of market structure earn over time. Everything else is a pitch — and the pitch has already cost people $1.73 billion.
No. Real automated systems exist, but they are quiet, modest, and verifiable: a live track record of six months or more, no martingale, no grid, real risk controls, and a developer with a history. The problem is not bots as a category — it is the marketing playbook scam vendors use.
No legitimate bot promises that. Bots that advertise guaranteed weekly returns — like the 5% to 9.99% per week EminiFX promised — have consistently turned out to be frauds. Realistic automated trading aims for modest, consistent returns with controlled risk.
Demand a verified live track record (Myfxbook or FX Blue, six months or more, on a regulated broker), confirm the strategy is not martingale or grid, check independent reviews, never hand over your broker login, and demo it for thirty days before funding.
Because backtests are optimized on historical data (curve-fitting), and live trading adds slippage, latency, and changing market regimes that the backtest never faced. A beautiful backtest is the easiest thing in trading to manufacture.
Want to understand what a bot is supposed to be reading in the first place? Download the free Trading Hub 3.0 handbook — it teaches market structure from the ground up.