TradeLens — Forex & Gold Strategy Analyzer

A port of a popular Pine Script built on smart-money ideas. Unlike the weighted engine, nothing votes: six filters must all pass — price on the right side of the 200-period average, the fast average on the right side of the slow one, volume above its average, a strong candle, RSI on the right side of 50, and a cooldown since the last signal. Only then can one of four triggers fire: a reclaimed order block after a break of structure, an EMA crossover, a breakout from a tight box, or a stop-loss hunt where price sweeps a swing level and closes back. It sets its own stop at the signal bar's extreme and its target as a multiple of that. The top half of the original script was not available, so the conditions it defined are rebuilt from their conventional meanings — each is a named, adjustable setting, and the signal panel shows exactly which ones passed on every bar.

Not directly — there is no Pine interpreter in the page, and Pine's execution model differs from a browser backtest in ways that change results, such as when a bar's values are final. What you can do is rebuild the rules with the weighted engine's settings, or ask for a script to be ported the way Deep Smart Money was. Either way the backtest applies the same next-bar fills, costs and held-back period to it, so the comparison with your TradingView numbers is fair — and where the two disagree, the disagreement is usually informative.

Neither. This is not financial advice, it is not personal to you, and it is not a recommendation to buy or sell anything — it is a laboratory for testing ideas. Every buy or sell here is the arithmetic output of rules you can read and change, applied to past prices — it knows nothing about you, your circumstances or the future, and it can be wrong. Treat a signal as a prompt to look at the chart yourself, not as an instruction.

It is how strongly the rules you switched on agree with each other, as a percentage of the total weight you gave them. A score of 72 means rules worth 72% of the weight point the same way. It is not a probability of the trade winning, and nothing in the tool could calculate that honestly — which is why it is never described that way.

Yes, and it is the most accurate option. MetaTrader publishes a free Python package that reads history out of a running MT5 terminal. Because that connection is local to your machine, no web page can call it directly, so the Gold desk tab gives you a small bridge script to run next to MetaTrader. It serves your broker's own XAUUSD candles — with their contract size, digits and live spread — to this page on 127.0.0.1 and nowhere else. The script is read-only: it cannot place an order and never touches your balance or positions. There is also a CSV mode if you would rather not run a server at all.

It makes the test honest, which is not the same thing. Your broker's real bars, spread and contract size mean a backtest is measuring the strategy rather than an approximation of the market. A rule set that loses money on accurate gold data is telling you the truth sooner. No data source, and no combination of indicators, can make a strategy reliable enough to trust blindly.

Four sources, all free. Your own MetaTrader 5 terminal through the local bridge; a public exchange endpoint that needs no key, whose gold instruments are tokens backed one-to-one by physical gold and therefore a proxy rather than XAU/USD; the European Central Bank's daily reference rates for forex; and optionally Twelve Data or Alpha Vantage using a free key you obtain yourself. Your key is stored in your browser and sent straight to that provider — never to DO101, and no key is built into this page.

Signals are read at a bar's close and filled at the next bar's open, never at the price that produced them. When a single bar contains both the stop and the target, the stop is taken, because the order they happened in is unknowable. Gaps fill at the open rather than at the level. Every trade pays the spread and any commission you set. Trading is blocked until the indicators have warmed up. And if your data has closing prices only, the engine says so and refuses to pretend it can test an intrabar stop.

Any search over enough parameter combinations will find something that worked on the data it searched — that is what searching does. So the tool splits your history: parameters are chosen on the first part, and then reported on a later part the search never looked at. It also shows the rank agreement between the two halves and whether the winning settings sit on a plateau or an isolated spike. If the held-back result collapses, the settings described the past rather than the market.

That is not a question this tool will answer, because the honest answer depends on your account, your broker's margin rules and a tolerance for loss a web page cannot know. Leverage is treated here as a ceiling you set: it never increases what a stop costs you, but it caps how large a position you may hold, and the risk panel shows the margin it ties up and how far the market would have to move to wipe the account out. Higher leverage lets you take positions that a small adverse move can end.

No. There is no broker connection and no order endpoint in this application, and the MetaTrader bridge deliberately calls only the three read-only functions it needs for price history. The paper account is a number and a list in your browser's storage. Nothing here can move real money.

Date, open, high, low, close and optionally volume, with or without a header row — MetaTrader's date-and-time export, TradingView, Dukascopy and ordinary spreadsheet saves all work. It detects the delimiter, works out which way round ambiguous dates are and tells you what it assumed, infers the timeframe from the gaps between bars, collapses duplicate timestamps and widens any bar whose high or low does not contain its own open and close. A close-only file is accepted and clearly flagged.

No. CSV files are read in the page, settings and the paper account live in this browser's local storage, and every calculation — indicators, backtests, the parameter search — runs on your own device. The only network requests are the price fetches themselves, which go straight from your browser to whichever source you picked.

Because the rules disagree, or because a filter you set has cut in: agreement below your minimum score, an ADX reading saying the market is directionless, volatility at double its recent normal, or a higher-timeframe trend pointing the other way while you have asked for agreement. The cautions listed under the verdict always say which of these it was.