Home · Learn · Evaluating an indicator
There is no "best" one, and any page telling you otherwise is selling something — this one included. What there is: eight questions that separate a usable tool from a marketed one, and a test you can run yourself in ten minutes.
Read this knowing who wrote it. We sell a futures indicator. That is a reason to discount our opinion, not to trust it. So the criteria below are written to be applied by you, to anything — and the last section applies them to our own product, including the three it fails. If that section were absent you should have stopped reading at the first paragraph.
Indicators do not have a ranking. They have a fit: to an instrument, a session, a holding period, and to how much decision-making you want left to yourself. An indicator that suits someone taking four trades a day on the Nasdaq open is a poor fit for someone taking one position a week, and neither is better.
What can be assessed is whether a tool is honest and inspectable. That is a real question with real answers, and it eliminates most of the market before preference enters into it.
The most important question by a distance. A repainting indicator changes its historical signals as new bars arrive, so the chart shows a past that never happened. Every backtest, screenshot and win-rate claim built on it is meaningless.
Vendors rarely lie about this outright. They just do not mention it. Assume nothing and test it — the method is below.
Not a results table. Not screenshots. The actual signals on your own chart, on your own platform, scrollable back as far as the data goes. If the evidence only exists in the vendor's marketing, the evidence does not exist.
An arrow tells you a direction. It leaves entry price, position size, stop and target to you, in the moment, with money moving — which is where most damage occurs. Neither approach is wrong, but be clear which you are buying, because a tool that answers one question out of five is not "a system".
Any real strategy has losing weeks and losing months. A vendor showing none is either curating or has too little history to have met one yet, and both should change your view. Look specifically for the worst month, the longest losing streak, and the maximum drawdown. If those numbers are not published, that absence is the finding.
A win rate on its own carries no information. Ninety per cent wins with a stop five times the target loses money; forty per cent wins with a target four times the stop makes it. Any figure quoted without the loss size beside it is marketing, and the number chosen is the flattering half of a pair.
The related question: how many consecutive losses does your account survive? That is what a wide stop actually costs.
A trial with full access is the cheapest honesty signal available. A vendor confident in their record has no reason to withhold it. Be more sceptical of a card-required trial that auto-converts than of no trial at all — the first is designed around forgetting to cancel.
A tool whose logic is hidden cannot be reasoned about. You cannot tell whether it is curve-fitted, whether it depends on a market regime that has ended, or whether it will behave in conditions it has not seen. You are trusting the vendor's testing rather than your own judgement.
Vendors have a legitimate commercial reason for hiding logic — a fully disclosed rule set can be reproduced by anyone in an afternoon. That reason is real and it is still your cost, not theirs. Weigh it consciously rather than accepting "proprietary" as though it settles the matter.
Backtests are cheap and forward performance is not. A vendor publishing their own execution — including the trades they skipped and the ones they got wrong — is making a claim that can be falsified. Ask whether the record is live-money or simulated, and treat those as different things, because they are.
Ten minutes, no special tooling, on any platform with a bar-replay feature.
Anything other than an exact match means the historical chart is not what you would have traded. It does not necessarily mean fraud — some indicators are honestly documented as confirming on close, which is fine as long as you know. It does mean every performance figure needs re-reading with that discrepancy in mind.
Red flags worth walking away from: a performance claim in the product name; a win rate with no stop or drawdown beside it; screenshots as the only evidence; no losing month anywhere in the record; countdown timers and closing-soon pricing; "reviews" on sites that earn a commission on the sale; and any suggestion that results are typical, guaranteed, or that the tool removes risk.
Ours is a futures indicator, so it should be held to the same eight questions. Here is how it does, including where it does badly.
| # | Question | How Falcon does |
|---|---|---|
| 1 | Repaints? | Passes — signals confirm and then stay put. Don't take our word for it; run the replay test above during the trial. |
| 2 | Full history inspectable? | Passes — the historical strikes sit on your own TradingView chart. |
| 3 | Defines the whole trade? | Passes — direction, entry, size, target and stop on every ticket. |
| 4 | Losses shown? | Passes — the simulator includes the losing months, and the forward test log publishes every session. |
| 5 | Win rate quoted with its stop? | Passes — the site states plainly that one full stop can give back several winners. |
| 6 | Evaluate before paying? | Passes — one week, no card. |
| 7 | Logic knowable? | Fails. It is a deliberate black box. The settings panel exposes colours and table visibility, nothing else. You cannot audit the reasoning. |
| 8 | Vendor trades it publicly? | Partially fails. Yes, and published daily — but on a funded prop account executing in a simulated environment, not live money. And as of writing that log is days old. |
Two more that do not fit the list and matter anyway. It trades one instrument — Micro E-mini Nasdaq, New York session. If you trade anything else it is useless to you, and no amount of quality changes that. And the 19-month backtest was run by us, which is exactly the arrangement question 8 exists to be sceptical of. The trial is offered because those two facts are not resolvable by argument.
If question 7 is the one you care most about — and it is a reasonable thing to care most about — then Falcon is the wrong tool for you, and an open-source or fully-documented system is a better fit even if its results are worse. That is a real trade-off and we would rather you made it deliberately than discovered it after paying.
The honest summary: we can prove questions 1 to 6 and we cannot prove 7. The trial exists so you can run these tests against a live chart rather than accepting a table written by the vendor.
Run the tests yourself — free week, no cardThis page is published by the vendor of one of the products it discusses and should be read with that conflict of interest in mind. Nothing here is financial advice or a recommendation to buy any tool or trade any instrument. Futures trading involves substantial risk of loss and is not suitable for every investor.