Back to all posts

Trading Bots: Are They Worth It?

By Abdelhak K. on May 19, 2025

Updated 31 August 2026. A bot is a script that sends orders when conditions in code are true. It will not get scared. It will also not notice that this week’s range is not last year’s trend. Prototraders does not sell expert advisors and does not run a public bot. Abdelhak’s view: automate only a rule you already follow by hand and can turn off.

What a bot is good at

Firing the stop and size you already defined. Working while you sleep — which is also how it can stack correlated losers at 3 a.m. Backtests against history, if you understand overfitting: a curve-fit EA looks holy until the next regime. Speed versus a human click is real in some microstructure; it is irrelevant if the logic is “buy every hammer.”

What it is bad at

Changing its mind. News gaps. Broker disconnects. Spreads that were 0.2 pips in the test and 4 pips live. Vendors with equity curves that never show the max drawdown or the skipped years. “Guaranteed monthly” is a scam tell in crypto and a lie in FX.

If you did not write the rules, you cannot know when they expired. Renting a black-box EA is paying for someone else’s undisclosed risk. If you do write the rules, you still need the 1% cap inside the code, not as a hope.

Who should even consider it

Not a week-one beginner. First: demo the manual loop (invalidation, calculator, journal). If that is stable, you might encode your loop. Monitor it. Have a kill switch. Internet and VPS failures are your problem. This is not a recommendation to automate, and not a claim that discretionary is “better” — it is a claim that unattended leverage is how small accounts vanish quietly.

Backtests, forward tests, and the equity curve that never happened

A historical test will find parameters that fit that history. That is not a crime. Treating the fit as a forecast is. Hypothetical teaching case, not a vendor review: an EA optimized on 2017–2019 trend data looks smooth. 2020 volatility and 2023 ranges chew it. The vendor’s sales page still shows the 2017 curve. If you cannot see max drawdown, average trade, number of trades, and the years excluded, you are looking at marketing. Walk-forward and a demo forward test do not make the bot “proven.” They only show whether the logic still fires in a new sample. Proven is a word we do not use for robots or for discretionary desks.

Spread, slippage, and rejected orders belong in the test if you want the test to mean anything. A 0.1-pip commission assumption on a news spike is fan fiction. If the bot scalps, those details dominate. If the bot swings, a weekend gap can still retire the account while you sleep. Unattended leverage is the theme. The leverage article still applies when the clicker is a script.

Infrastructure is part of the strategy

A home PC sleeps, updates Windows, and loses Wi-Fi. A VPS can still die. The broker can reject, requote, or disconnect. Your kill switch — disable autotrading, flatten, or pull the EA — has to be a habit, not a wish. If you cannot explain every condition in the code, you cannot know when to pull it. Renting a black box means you are the last to know when the vendor changed a filter. That is not a partnership. That is undisclosed risk you paid for monthly.

“Guaranteed monthly” is a scam tell. Crypto grid bots that need you to deposit into a third-party dashboard with withdrawal delays are often just a pool. A bot on your own broker account, running rules you wrote, is a different object. We still do not sell one. We still do not run a public one.

The only automation we respect

After the manual loop is boring on demo — invalidation, calculator, stop in the ticket, journal — you might encode that loop: same 1% cap from the risk process, same no-add-to-losers, same flatten-into-red-folder if that is your rule. Then you watch it. Then you turn it off when the regime is not the one you coded. That is still your judgment. The bot did not replace it. Beginners who buy an EA in week one skip the only part that teaches. This is not a recommendation to automate and not a claim that discretionary is superior. It is a claim that unattended size is how small accounts vanish without a story. Educational only. Not financial advice. Hypothetical tests here are teaching, not results.

Disclaimer: This article is for informational and educational purposes only. It is not financial advice. Trading Forex and cryptocurrencies involves substantial risk of loss.