How the AI scores its confidence

The bias comes from the top, and it overrides everything

The reading is top down, in three steps, and the order is not negotiable.

• The highest timeframe visible in the image sets the macro bias. This is the most important output of the whole analysis.
• The intermediate timeframe confirms the current dealing range and the structure.
• The lowest timeframe only provides the entry signal.

The rule that follows is blunt: a setup that contradicts the higher timeframe bias is never rated highly. If the higher timeframe is bearish and the lower one shows a clean bullish setup, the analysis rates it C grade or lower, or names it a counter-trend trap. It does not split the difference.

The confidence scale, and why it is not always 70

A score that is always between 70 and 75 tells you nothing. The scale is therefore explicit, and the analysis is instructed not to default to the middle.

• 85 to 100. Alignment across timeframes, a completed sweep and reversal, entry on a zone that has not been touched yet, and a trap risk under 20.
• 65 to 84. Three or more agreeing signals, a clear higher timeframe bias, a visible zone to enter on, minor ambiguities.
• 45 to 64. A setup exists, but one or two signals contradict it. Moderate trap risk, marginal risk to reward.
• 25 to 44. Low conviction. The higher timeframe is unclear or the structure contradicts the setup.
• 0 to 24. No viable setup. The analysis has to state exactly why, and the grade becomes No Setup.

Why a poor image can never score high

The first thing the analysis does is judge the image itself, before reading anything from it. A chart with many legible candles, readable axes and visible indicators gets the full treatment. A chart with few candles, unreadable axes and no indicators gets a short, modest answer.

The rule is written as a hard limit: a sparse or low detail chart can never earn high confidence. This matters more than it sounds. The easiest way to make an analysis look impressive is to write confidently about an image that does not justify it, and that is precisely what is forbidden here.

A brief honest reading of a simple chart is treated as a success, not a failure.

What it refuses to do

Three refusals are written above every framework, and they override all of them.

• It never invents a price level, an indicator value, a timeframe, a session, a zone or a pattern that is not directly readable in the image.
• If a value is not legible, it says so in plain language instead of guessing it.
• It never pads. No manufactured institutional story that the pixels do not support.

This is why two analyses of the same asset can differ: they are readings of two images, not of a market.

What the verdict is not

It is not a prediction, and it is not an instruction. It is a statement about what one image supports, with a number attached that says how strongly.

The app has no market data feed and no access to the web. It reads the chart you give it, and nothing else. Everything in the analysis has to be visible in that image, which is both its limit and the reason its numbers mean something.

Questions people ask

Where does the direction come from?

From the highest timeframe visible in the image. That bias is the most important output of the analysis and it overrides every signal from a lower timeframe. A lower timeframe setup that contradicts it is rated C grade or lower, or named a counter-trend trap.

Why is the confidence sometimes very low?

Because the image does not support more. Confidence measures how strongly that one chart supports the read, not how confident the writing sounds. A sparse chart, unreadable axes or too few candles all cap it, by rule.

Can the analysis say there is no trade?

Yes, and it is expected to. Below 24 the grade becomes No Setup, and the analysis has to state exactly why no setup exists rather than manufacture one.

Does it use live market data?

No. There is no market data feed and no web access. The analysis reads the image you provide, which is why it never quotes a level that is not visible on it.

Reading a chart without scoring it by hand

Trading AI reads a photo or a screenshot of any chart and returns the direction, the confidence out of 100, the market regime and the full structure behind them, with a trade plan and the risk that goes with it.

Market structure is where the bias comes from: What is market structure?

The trap risk is rated before any plan is written: What is liquidity in trading?

All of these ideas belong to one framework: Smart money concepts, explained plainly

This page is educational. It is not financial advice, and nothing here is a recommendation to buy or sell anything.