Zero-Touch Divergenceforward observation
A divergence detector. It compares a higher-timeframe price extreme against the oscillator that measures how stretched price is from its anchored VWAP, and when the two disagree it reports a bias: a direction, the level that would prove it wrong, and the level it aims at. It never proposes an entry.
How it reads a chart
- 1Wait for price to reach an extreme against its anchored-VWAP band while the oscillator is also at an extreme.
- 2Record that range as the reference, and wait for the oscillator to come back to zero.
- 3On the next excursion, compare its extreme to the reference: a higher price on a weaker oscillator is a regular divergence, a weaker price on a stronger oscillator a hidden one.
- 4Report the bias that implies, with the range extreme as the level that ends it and the zero-touch level as the level it aims at, and keep reporting until it is contradicted, times out, or arrives.
What lands on the card
- Whether a divergence bias exists on this instrument at this timeframe, stated either way. Most bars carry none, and that is the method's ordinary state rather than a failure to read the chart.
- When one is live: its direction and kind, the price beyond which it is wrong, and the price it aims at.
- How the last few biases ended, reached, contradicted or timed out, so a run of timeouts is visible rather than hidden behind a single live read.
Where it goes quiet
- Hourly, four-hour and daily charts only. The rule gates itself to those in the original indicator, and a request for anything faster is refused rather than quietly served a different object.
- A bias is not a trade. Nothing here says where to enter, and the entry layer that was built on top of this failed its own tests; that record ships separately with its own method.
- It fires a few times a month per instrument. A read that says no bias is the common one.
What the testing found
The rule was verified bar-for-bar against the chart it was designed on, and its port matches the original to machine precision. That is verification of the detector, not of an edge: a bias is graded by what the market does after it, and that record is short. A six-month mechanical grade of its hourly signals on one pair came back two won, five lost. This method is published under forward observation so that record can grow in public, and every read it produces is tracked to its own outcome like any other.
Tracked outcomes
No outcomes tracked for this method yet. Every read it publishes is graded to a win, a loss or an expiry, and appears here once the first ones resolve.
CAT is an educational community. Every analysis shows how our system reads a chart — it is not investment advice and not financial advice, not a recommendation to buy or sell, and carries no performance promise. Trading involves substantial risk of loss.