Every trader has levels that price "always comes back to". Checking that by hand on a month of history takes a long time, so usually nobody checks.
The agent counts how often price returns, on the bars of your own chart, from one message, and shows exactly what it counted.
How often does price return to the previous US session POC in the first two hours of the next US session? The US session is 9:30-16:00 New York time. Count it on my BTCUSDT 5m chart over all loaded history. Use code. Show how many sessions you took, how many returns there were and what you counted as a return. Don't change my charts.
Set the history depth on the chart beforehand.
What happens
The agent counts directly on the bars of your chart and answers with a number and its definitions.
How often does price return to the previous US session POC in the first two hours of the next US session? The US session is 9:30-16:00 New York time. Count it on my BTCUSDT 5m chart over all loaded history. Use code. Show how many sessions you took, how many returns there were and what you counted as a return. Don't change my charts.
There was a return in 18 of 44 sessions, 40.9%, counting US sessions Monday to Friday and using Friday's POC for Monday.
For BTC I also computed a variant with a daily window that includes weekends:
| Variant | Sessions checked | Returns | Frequency |
|---|---|---|---|
| Mon-Fri | 44 | 18 | 40.9% |
| All calendar days | 62 | 19 | 30.6% |
Computed with code over all 18,109 loaded BTCUSDT 5m candles, Sep 4 to Nov 5, 2026. The first session serves as the base for the POC.
What was counted:
- POC is the price with the highest total volume by price level over 9:30-16:00 New York time. When volumes were equal, the lower price was taken.
- Return means at least one candle opening between 9:30 and 11:25 satisfies
Low ≤ previous POC ≤ High. A touch by a wick and an open right at the POC are included; repeated touches count as one return. - For each pair I checked all 78 candles of the previous session, the 24 candles of the check window and that volume by price level was present. Nothing is missing.
- The switch to standard time is taken into account. In the Mon-Fri variant, weekday holidays were not excluded.
Calculation code · Results by session
I didn't change your charts: the code ran as a preview calculation without adding an indicator.
Look at the definitions: what the POC is, what counts as a return, which days were dropped and why. Without them, 40.9% means nothing. Here the agent computed a second variant on its own, with weekends, and the frequency came out different. Ask the agent to always show them.
This is only how often price touches the level. It says nothing about the result of trades. The Hypothesis template computes the historical result after costs.
What you need
- A crypto exchange connection
- An agent that can run code. Statistics computed directly in the chat are unreliable.
These are historical statistics. Past results do not guarantee future results, and the trading decision is yours.