How the numbers are made
Every number here you could work out by hand — there's no secret score. Here's exactly how each one is built.
How the signal is built
The signal is a single number between −100 and +100. It blends five things we measure on the daily price chart, and nothing else — no news, no opinions, no hidden model. The weights below are read straight from the running server, so this page can never drift from what the app actually does.
The five ingredients
Each ingredient gets its own score from −100 (strongly down) to +100 (strongly up). We multiply each by its weight and add them up. These five add up to 0.9 rather than 1, because there's a sixth ingredient — trend strength — that multiplies the others instead of adding to them. So in practice the score runs to about −90 to +90, and in calm markets it stays far smaller than that.
| Trend (EMA) | 30% |
| Momentum (MACD) | 20% |
| Overbought / oversold (RSI) | 15% |
| Price band position | 15% |
| Trading volume | 10% |
The sideways-market guard
Trend ingredients sound most confident exactly when they are least reliable — in a market going nowhere. So we measure trend strength (ADX) and use it as a dimmer switch on the two trend ingredients: at 25 and above they count fully, and they fade toward nothing as trend strength drops. Below 20 we stop trusting the read at all and hold the signal at neutral, whatever the score says.
When it commits to a direction
The score has to clear 15 in either direction before we call it leaning up or leaning down. Anything in between is honestly just noise, so we say "no clear direction" rather than invent one.
Where the weights came from
They began as an educated guess, not a discovery. We try a few alternative weightings on older data, pick whichever did best there, then measure it on data it has never seen with trading costs subtracted. That out-of-sample record sits on every asset page — including the assets where the signal loses to simply buying and holding, which on Iranian gold and coins it clearly does.
Why costs decide everything here
Every round trip pays a spread. We assume 0.02% per side on currencies and 2.5% on Iranian gold coins. A signal that trades often has to be right far more than half the time just to break even, which is why a strategy can look clever on a chart and still lose money in real life.
How risk is calculated
Likely bad day (VaR 95%, 1-day)
We line up the last 250 daily moves and take the 5th-worst percent. Read it as: on about 5 days out of 100, a one-day loss bigger than this is expected. Its blind spot — it says nothing about how bad the worst days get.
Worst drop (1 year)
The deepest fall from a peak to the following low over the last 365 days. This is the number that actually shows the pain, and it's why we never show the "likely bad day" on its own.
Volatility right now
We take the last 30 days of ups and downs and rank them against the past year: calm below the 25th percentile, choppy above the 75th, wild above the 95th.
Risk score (1–10)
An equal blend of four things, each turned into a 0–1 riskiness: daily swing (capped at 8%), the volatility ranking, the likely bad day (capped at 10%), and the worst drop (capped at 60%). Score = 1 + 9 × the average.
Position sizing
Suggested size = (account × risk %) ÷ (daily swing × your stop multiple). For illustration only, and it needs numbers you type in — we never assume or store them.