What Is Actually Inside a Hands Free Trading Robot

A hands free trading robot is a program that runs inside a trading platform and executes a set of rules written in advance. It does not think about the market, and it does not take the account off your hands. It applies the same instructions every time the conditions in those instructions are met, whether you are watching or not.

The word robot does most of the damage here. It makes people picture something that reasons the way a trader reasons. What is being sold is closer to a machine on a production line: it reads a few inputs, compares them to conditions someone wrote months ago, and sends an order when they match. Everything the software can do was decided before it reached your account.

That is not an insult to the category. A rule engine that never gets bored, never widens a stop out of hope and never doubles a position out of frustration is a real advantage in an evaluation, where one emotional decision can end the account. But it changes what you should inspect before you pay. You are not judging intelligence, you are judging a specification.

Hands free describes who clicks the buttons, and nothing more. The human still chooses the software and the settings, keeps the machine it runs on alive, reads the reports, updates the configuration when the firm changes a rule, and decides when to stop. That is less work than trading by hand and it is a long way from no work, and the way those jobs fall across an evaluation is set out in our companion piece on how a hands free setup works during a prop firm challenge.

Every trading robot, whatever its price or its marketing, is assembled from the same five components.

The order of importance is the reverse of the order in which vendors present them. Entry signals sell the product because they sound like insight. Risk locks and position sizing decide whether the account is still alive at the end of the month, so ask about the last three first, which is the approach we take in our guide to choosing a trading bot for a prop firm account.

A robot missing one of the five is not simpler, it is incomplete. A system with no risk lock still has a behaviour during a losing streak, and it is exactly the behaviour a funded account cannot absorb.

A trading robot has no existence outside the platform that hosts it. It is not a cloud service you subscribe to. In the MetaTrader world it is a file attached to a chart, and the code only executes while that terminal is open, logged in to the broker account and connected. The platform is the engine and the robot is a passenger.

That has a consequence people discover at the worst possible moment. If the laptop sleeps or the connection dies while a position is open, the robot is not paused, it is absent. The position stays on the broker server with whatever stop was already attached, and any exit logic that lives in the code, such as a trailing stop or a time based close, does not fire until the terminal comes back. This is the entire reason traders rent a virtual private server for a prop firm robot instead of running one at home.

Execution is the other half of the environment. Your software sends an instruction and the broker server decides the fill, so spread, slippage and requotes belong to that connection rather than to the code. The same robot on two accounts with different brokers can produce visibly different results from identical signals.

A configuration file is a saved list of parameter values that tells one robot how to behave. In MetaTrader it is usually a set file, loaded separately from the program itself. The same executable running two different configuration files is, in practice, two different systems: different risk per trade, different instruments, different session hours, different caps.

This is the most misunderstood part of buying automation. Traders compare products when the meaningful comparison is often between settings, since a file built for a small evaluation with a tight daily loss limit will not behave like one built for a larger account with more room. Our explainer on what a set file changes and what it cannot goes through the parameters that move the outcome.

Two practical tests follow. A vendor who ships one universal file for every firm, account size and instrument is guessing on your behalf, because rule sets are commonly structured but differ by firm, account type and platform. And when a firm changes a limit, the fix is usually a parameter change, so ask who issues the new file and how fast.

The honest way to evaluate a robot before you buy it is to read what it does with orders rather than what it claims about the market. Order behaviour is the part of an automated system a copywriter cannot write, and a detailed report answers in five minutes what sales material dodges.

The most revealing of these is what happens to size after a loss. A system that increases its position to recover a losing trade is applying a recovery pattern, and why that pattern is treated as a specific risk in evaluations is explained in our piece on martingale and grid robots under prop firm rules. A configuration commonly considered conservative does the opposite: one position at a time, a stop attached from the moment the order exists, and a size that never grows to chase a loss.

Read the worst stretch, not the best one. Any report can show a good week. What tells you about a system is the deepest run of losses and whether the worst single day would have breached a daily loss limit on the account size you intend to run.

A deterministic robot produces the same decision every time it is shown the same inputs. An adaptive system changes its own parameters or rules in response to new data, so the version trading on Friday is not the version that traded on Monday. Almost every product sold as an AI trading robot is deterministic, with a human retuning the parameters between releases.

That is not a scandal. Determinism is a feature during an evaluation, because a fixed rule set can be audited against a firm rulebook, tested on history and predicted within known bounds. A system that rewrites its own behaviour halfway through a challenge cannot be checked in advance, since the behaviour being checked no longer exists. The gap between marketing language and mechanism is unpacked in our article on whether any trading bot is really AI.

The useful question is not whether a product is intelligent. It is whether the behaviour you are buying is written down where you can read it, and whether the parts that protect the account are fixed rather than discretionary.

A short list of specific questions separates a vendor who built something from one who resold it. None of them are about returns, and all have factual answers.

Disclosure: we publish this blog and we sell trading software. PraxAI is a one time purchase at $997 by card or $797 by crypto, against a normal price of $2,997, with lifetime access, unlimited accounts, a 7 day guarantee and rule updates within 48 hours. Its risk limits are applied in code by PraxAI GUARD, which is not artificial intelligence but your own limits enforced by the software, and its gold validated configuration takes one position at a time with no martingale and no grid. Apply the questions above to it as you would to anything else on our comparison of AI trading bots for prop firm accounts.

Understanding the anatomy is the cheap part. The expensive part arrives later, since only 1 to 3 percent of funded traders keep the account long term, and the same five components decide that outcome too. What it takes to keep a funded account is a story about risk locks and sizing rather than entry signals, which is where this article started.