Methodology

1. The data path

No page on this site talks to an external API directly. Data is fetched by a scheduled job, validated, stored in the database, and served from there.

Source → Provider → Validation → Database → Cache → Signal engine → Display

The consequence is that if every API failed at once, the site would still show the last valid value with an accurate timestamp. A third-party outage never empties or breaks a page.

2. Being honest about Iranian data

For much of what an Iranian reader actually needs — the free-market dollar rate, domestic gold and coin prices, the stock index, Tehran housing prices and the Statistical Centre's CPI — there is no free, licensed, stable public API. The official bodies publish HTML pages and PDFs, not APIs, and the aggregators that do serve JSON generally have neither clear terms of use nor any stability guarantee.

So those indicators are entered by hand or via a CSV we control, with their source stated explicitly. It is slower, and it is correct. Scraping an unlicensed site and presenting the result as authoritative is precisely what would destroy this project's credibility.

Global series — the World Bank, ECB reference rates, commodity closes — are free, documented and keyless, and are fetched automatically.

3. Frequency is what it is

Every indicator declares its real frequency. Inflation is published monthly, so it is never labelled "live" — even if it was read from the source a minute ago. Each card's freshness language comes from the data's real period, not from when we requested it.

4. Signals

Signals are the output of a few simple statistical rules: threshold crossings, moving-average crossings, sustained direction, acceleration, volatility jumps, z-scores and divergence from related indicators. None of them are "AI", and every one can be reproduced in a spreadsheet.

Every signal displays the rule that produced it and the number that triggered it. There is nowhere in this system's schema to store a guaranteed prediction.

5. Relationships are written by hand

The connection map and the "If this → then what?" explorer are built on relationships a person wrote and explained. None were mined from statistical correlation.

The reason is simple: a map of coincidental correlations looks impressive and means nothing. Correlation is not causation, and false authority is the worst thing an analytical platform can manufacture.

6. Fact, interpretation, scenario

These three are kept apart in the database, not just in the prose. "Data" is one column and "my take" is another, and they are rendered differently on the page. Every piece states its confidence level.

Instead of "the dollar will rise", this site says "these indicators are currently creating upward pressure". The difference between those two sentences is the entire difference between this project and a signal-selling channel.

7. If we don't know, we say so

If we have no data, no number is shown. If a source is weak, we say so. If an analysis is only a hypothesis, it is called one. If several readings are possible, all of them are shown.