Are Forex Signal Services Worth Paying For? How to Tell Real From Fake

Forex signals are a mature market: Telegram groups, subscription sites, broker-supplied signal feeds, "mentors" inside prop firm communities. Pricing runs from $30 to $300 a month. The question is not whether signals can work — it is how you would know that a particular one does.

Start with the break-even

Suppose the subscription is $100 a month and the account is $2,000. For the subscription merely to pay for itself, the signal must return 5% a month — before your own trading costs.

The same $100 on a $20,000 account is 0.5% a month, a very different burden. The subscription as a share of account equity determines whether a signal is usable for you at all, and most people skip this calculation entirely.

Three fee structures with three different incentives

Price is not the issue; where the revenue comes from is. Three common structures:

Subscription (fixed monthly). Revenue is independent of how much you trade, which is theoretically neutral. The incentive is retention, which favours showing a flattering equity curve.

Profit share. Revenue is tied to your results — the best-aligned structure. The cost is a 20%–40% cut, and usually a requirement to use a specified broker.

Rebate (broker commission). Free or even subsidised on the surface, funded by your trading volume. Here it pays the provider to have you trade more, and high-frequency, high-leverage signals get recommended hardest.

Once you see this layer, the motive behind "free" and "very cheap" signals becomes fairly clear.

Three ways a "verified" account misleads

Most sellers show you a Myfxbook or FXBlue "verified" link. What that verification guarantees is that the account data has not been hand-edited. It does not guarantee that the account represents real trading.

• Demo account: the verified link points at a demo, so the curve is clean but contains no real slippage or execution delay
• Selective disclosure: only the profitable account is published; the losing ones stay private
• Hidden holding times: losing positions are carried until they come back, so the curve shows no drawdown while the real risk is extreme

There is a subtler variant: rebate capture. Some providers earn from you opening an account at a specified broker. Their income comes from your volume, not your profit — so "trade more" helps them and may not help you.

Survivorship bias: you only see the survivors

Imagine 1,000 people each start a signal group and trade at random. After a year, probability alone leaves several dozen with beautiful curves. Those are precisely the sellers you will be shown.

This does not mean every seller is a fraud. It means a good historical curve cannot be your evidence, because it cannot separate skill from luck.

Seven questions worth asking

• Is there any out-of-sample record — live results from after the subscription started, not a backtest?
• Does the record span more than 12 months, and at least one high-volatility episode?
• What are the average payoff ratio and maximum drawdown, not just the return?
• What is the risk per trade as a share of the account? Anything above 2% can be dismissed immediately.
• Are all signals published, including the losers?
• Does revenue come from subscriptions or from broker rebates?
• Is there a stated invalidation condition — a point at which you should stop following it?

Minimum sample size

To judge whether a signal has positive expectancy, a rough rule of thumb is at least 100 completed trades, spanning different volatility regimes. Below 30 trades, noise dominates signal.

A more practical method is shadow tracking: risk nothing, record two months of simulated fills at real spreads, then compare against your own manual trading. That alone filters out most of the "looks great" services.

A six-field log is enough

Shadow tracking needs no tooling. Six columns:

• Signal time and the time you actually saw it — that gap is your latency
• Quoted entry and the price you could realistically get — that gap is your slippage
• Stop and target — to check the risk-reward is at least 1:2
• Outcome in R, not in money
• Session (Asia / Europe / New York overlap)
• Whether data was scheduled, flagged so it can be analysed separately

After two months, split the results by session and by data/no-data. If the signal performs noticeably worse around releases, its execution assumptions do not match your account — far more useful than the headline return.

Signals versus strategy

A signal gives you an outcome (buy or sell). A strategy gives you rules (why buy, when not to, how much to lose before stopping). The first you can only believe; the second you can test.

If you must use signals, use them as an entry trigger only: size, stop and exit stay under your own rules. Then even a mediocre signal leaves your risk controlled.

Three rules that cage a signal

One: fix the risk, not the lots. The signal gives direction; the 1% rule gives size. The same signal then cannot do different damage on different accounts.

Two: give the signal a trial stop. Ten consecutive losses, or an 8% drawdown, ends the subscription. Decide this in advance to avoid the sunk-cost trap of "give it a bit more time".

Three: log the original wording. Screenshot what the seller actually said, with a timestamp. In review, "what he said" and what you remember often differ — this step separates a bad signal from a misread one.

FAQ

Are free signals usable? The problem is rarely quality but incentive structure: a service with no subscription revenue is usually funnelling you toward a broker or a course.

Do copy-trading communities count as signals? Closer, because there is a real fill history and equity curve — but the same survivorship bias applies, since only the top of the leaderboard is shown.