Manual vs Automated Forex Signals: Human Analysts, Algorithms, Alerts and Auto-Execution

Manual forex signals are trade ideas written by a person, such as an analyst reading charts or news, while automated signals are produced by software following coded rules. Separately, any signal can be delivered as an alert that you choose to act on, or auto-executed in your account without you approving each trade, and that second distinction matters more for your risk and your legal protections than who or what wrote the signal.

In short: there are two questions, not one. Who generates the idea? A human, an algorithm, or a human using an algorithm. Who places the trade? You, after reading an alert, or a system acting on your behalf. Manual alerts give you control but add delay. Auto-execution removes delay but hands control of your account to someone else’s rules, and regulators treat that as a more serious service: in the US, directing trades in a client’s account is the kind of activity that removes common registration exemptions for advisers, and in the EU, ESMA has said that automatically turning third-party signals into orders requires authorisation, in particular as portfolio management.

The marketing version

Signal products tend to be sold in one of two flavours. One is personal: “Our head analyst has 15 years of experience and calls every trade himself.” The other is technological: “Our algorithm scans 28 pairs around the clock so you don’t have to,” often now with “AI” in the name. Both pitches imply the method is what makes the signals good. Neither is evidence. A human can be skilled or lucky or invented; an algorithm can be well tested or curve-fitted to the past. What changes between them is how the signal is produced, how quickly it arrives, and what can go wrong.

Who generates the signal

Human (discretionary) signals

A person looks at the market, using charts, news, or both, and decides on a trade. The style can be technical (based on price patterns and indicators) or fundamental (based on economic data and central bank policy), as discussed in technical vs fundamental signals.

Strengths: a person can take context into account, such as an upcoming central bank decision, a holiday with thin liquidity, or a headline the rules didn’t anticipate.
Weaknesses: consistency. The same person may trade differently on a good week and a bad one. Discretionary records are hard to test because the rules aren’t written down, so the only evidence is the live record.

Algorithmic signals

Software applies fixed rules to price data: for example, “buy when the 20-period moving average crosses above the 50-period average and the price is above its 200-period average.” When the conditions are met, the program produces a signal.

Strengths: consistency and speed. The rules can be backtested, monitored around the clock, and applied to many pairs at once.
Weaknesses: rules found by searching through historical data often describe the past rather than predict the future. This is overfitting, and it is the main reason backtested results look better than live ones. See backtests vs live results.

The hybrid most services actually use

Many “manual” services use screeners or indicators to generate candidates and then have a person pick which to send. Many “automated” services have a person decide which algorithm to run, when to switch it off, and which results to publicise. The label tells you little about the process.

Who places the trade

Alerts you act on

The signal arrives as a message, push notification, email or platform alert. You read it and decide whether to place the order, at what size, and with what stop.

You keep control. You can skip signals ahead of news, trade smaller, or stop entirely.
You add delay. Every second between the alert and your order is time for the price to move.

Auto-execution

The signal is converted into an order in your account automatically. This is done in a few ways:

Expert advisors (EAs) or scripts running on your trading platform, sometimes on a rented virtual private server so they run around the clock. See expert advisors explained.
Signal subscription or copy services built into a broker’s platform, which mirror a provider’s trades into your account.
Third-party bridges that read messages from a chat channel and place matching orders through your platform.
Broker-side arrangements. In the US, futures brokers have long accepted a customer’s letter of direction instructing the firm to place trades according to a named trading system’s signals. NFA guidance on third-party trading system developers describes this arrangement.

Auto-execution solves the speed problem and creates a control problem. A software fault, a misread message, or a provider having a bad day goes straight into your account, often at a size you configured once and forgot about.

The legal line: advice versus directing an account

Regulators draw a clear distinction between telling people what to trade and trading for them.

United States. The NFA describes a commodity trading advisor (CTA) as anyone who, for compensation, advises others on trading futures, options, swaps or retail forex. CFTC Rule 4.10(f) defines directing an account as being authorised to cause trades in it without the client’s specific authorisation for each one. A key CFTC exemption for advisers, Rule 4.14(a)(9), is not available to someone who directs client accounts. In practice, a service that only sends alerts may sit in a different regulatory position from one that places trades. The details are in are forex signal sellers regulated in the US.
European Union. In a 2012 Q&A, ESMA considered a website that let clients choose third-party signal providers and then turned those signals into orders once the client authorised it. ESMA concluded that this required authorisation under MiFID, in particular as portfolio management. Its 2023 supervisory briefing on copy trading says such services are likely to be either portfolio management or investment advice, depending on whether the client confirms each trade.
Canada. The CSA’s position is that anyone giving investment advice or selling investments in Canada must comply with securities or derivatives law. Check registration with the National Registration Search before giving anyone access to an account.

Worked example: what delay costs a manual follower

A provider sends: BUY EUR/USD, entry 1.0850, stop 1.0820, take profit 1.0910. That is 30 pips of risk for 60 pips of reward, a ratio of 2.0.

Suppose the market keeps rising while you unlock your phone, open the platform and enter the trade. You keep the provider’s stop and target, because that is what the signal says.

Your entry	Pips risked	Pips to target	Reward-to-risk
1.0850 (provider’s price)	30	60	2.00
1.0852 (2 pips late)	32	58	1.81
1.0855 (5 pips late)	35	55	1.57
1.0858 (8 pips late)	38	52	1.37

At 8 pips late, a trade designed to win twice what it risks now wins about 1.4 times. If the provider’s edge depended on that 2.0 ratio, being late can erase it entirely, even if you copy every signal faithfully. And because providers usually publish results at their own entry price, your results and theirs will diverge in a predictable direction: yours will be worse. The mechanics of this gap for copied trades, including slippage, are in slippage and execution on copied trades.

What “AI signals” usually are

Labelling a signal as AI-generated has become common. In a January 2024 customer advisory titled “AI Won’t Turn Trading Bots into Money Machines”, the CFTC warned that scammers claim huge returns from bots, “trade signal algorithms” and other AI-assisted technology. It cited the Mirror Trading International case: the CFTC alleged that the company and its CEO, Cornelius Steynberg, took in Bitcoin then worth more than $1.7 billion from at least 23,000 people in the US and thousands more worldwide, through a forex pool supposedly run by a proprietary bot that promised about 10% a month. Federal courts entered judgments against both in 2023. Most of the money was never traded.

Genuine machine learning is used in professional trading, but nothing about the label tells you whether a retail product is genuine, tested, or profitable.

Which is better?

Neither, by design. Each trades one problem for another.

	Manual alert	Auto-execution
Speed	Slow (seconds to minutes)	Fast (milliseconds to seconds)
Control	You approve each trade	Rules approve each trade
Typical failure	Late or missed entries	Faults, runaway trading, oversizing
Regulatory treatment	Often advice	Often directing an account or portfolio management
Evidence needed	Live record at realistic entry prices	Live record plus how the system behaves in fast markets

What can go wrong

Manual: you miss entries, skip the losing-looking signals (and some winners), and end up with a different record from the provider’s.
Automated: a bug opens duplicate trades, a connection drop leaves positions without stops, or a strategy that adds to losing positions runs unattended. See martingale and grid strategies explained.
Both: the provider’s record may be a backtest, a demo account, or selected screenshots, and the provider may be paid per trade you place rather than per profit you make.

The short version

Ask two separate questions about any signal service: who generates the trades, and who places them. Human and algorithmic signals both need a verified live record; neither label is evidence of skill. Manual alerts cost you speed, and the table above shows how fast a few pips of delay erode a signal’s reward-to-risk. Auto-execution costs you control and moves the service into territory regulators treat as managing your money.