Contents
The numbers that decide everything
A signal service is judged on exactly three things: the percentage of trades that win, the average win divided by the average loss, and the maximum drawdown on the account. Everything else on the page is decoration. A service showing 87 percent accuracy is telling you almost nothing if the losing trades are catastrophic and the drawdown was 60 percent.
Expectancy is the number that combines them. Take the win rate, multiply it by the average win in points, subtract the loss rate times the average loss, and divide by the number of trades. A service with a 55 percent win rate, a 1.8 reward-to-risk ratio and 3.2 average loss has an expectancy of roughly 0.47 points per trade. That is a real edge. A service with 70 percent accuracy and a 0.3 reward-to-risk has an expectancy of roughly minus 0.17 and is losing you money while looking excellent.
The catch is that most services will not publish the second and third numbers. That absence is itself the most informative signal available.
What 'free signals' actually means
Free signals usually come in one of four arrangements. Genuinely free, where the provider gives a view and makes money from a subscription elsewhere, which is the honest version. Freemium, where the first few signals are free and quality is deliberately poor to demonstrate the paid tier, which is a legitimate sales tactic but frequently dishonest about the free tier. Delayed, where you receive the signal after the entry has already happened, which makes the published win rate meaningless. And bait, where the free tier exists purely to collect contact details for selling an unverifiable service.
The delayed category deserves particular attention because it is the most common and the least obvious. If the signal arrives after price has already moved in the predicted direction, the historical record is complete fiction. Any service that cannot show you the timestamp of the signal against the timestamp of the price is not showing you a track record.
There is a related issue with signal chasing on social platforms. A post showing a winning gold trade tells you the outcome, not the process, and the sample is selected to include the wins. Our guide to gold signal scams covers the specific patterns to avoid.
The four questions to ask any provider
First: can I see the full equity curve, including every losing trade and every period of inactivity? A track record with gaps is a track record that has been curated. Second: are the results net of realistic spreads and commission, using the spreads your broker actually gives you? Third: what was the maximum drawdown, and how long was the longest losing streak? Fourth: what happens if I miss a signal, or if I am in a different session when it fires?
If a provider answers all four clearly and shows the numbers, you have something worth testing. If they deflect or answer only the win rate, you do not.
The fourth question matters more than most traders realise. A signal that fires during Asian session and assumes you can act on it in the middle of your day job produces a completely different result from the published record. We cover this in copy trading and signal copying.
The cost side nobody mentions
Even a genuinely good signal loses its edge once you account for what it costs to act on it. If you are paying for a signal in pips, add the spread you cross, the commission your broker charges, and the slippage you get during fast markets. A scalp signal on gold is extremely sensitive to all three because the profit per trade is small.
There is also an unquantified cost in attention. A signal that requires you to be at the screen at a specific moment imposes a real constraint on your day, and if you take it in a state of distraction the execution will be worse than the signal's record implies.
This is the strongest argument for understanding order flow yourself rather than renting someone else's conclusion. When you can read the flow directly, a signal becomes a second opinion rather than a dependency, and the cost collapses. Our Order Flow Desk is built around that idea.
Testing a service before you commit money
Do not start with money. Take any service you are considering and treat it as a paper-trading provider for a minimum of fifty signals. Record the timestamp, the entry price your broker would have given you, the stated stop and target, and the outcome.
Fifty is enough to expose the difference between a service with a 70 percent win rate and one with a 45 percent win rate, and it is enough to see whether the free tier is being sabotaged. If you are not willing to spend fifty signals watching rather than trading, you are not evaluating it carefully, and you will be tempted to over-interpret the first few outcomes.
Do the same exercise with a free order flow terminal, so you have a direct read to compare the signals against. When the signal says sell and the flow says aggressive buying at a major level, that disagreement is where your real understanding comes from.