Technical Analyst reads weekly chart images for stocks, indices, crypto or forex and works through a fixed sequence: trend, support and resistance, 20, 50 and 200 week moving averages, volume, and chart patterns. It then builds two to four scenarios with probabilities that sum to 100 percent, target levels and invalidation levels.
A second mode checks whether a weekly chart shows reversal evidence for a crowded market, using key reversal, failed extreme and failed breakout rules, with an optional script that pulls weekly prices. The reports describe chart structure and are not investment advice.
What You Get
- One Markdown report per chart named by symbol and date, following a fixed template with ten sections
- Two to four scenarios, each with a probability, supporting factors, target levels and an invalidation level
- Trend, support and resistance, moving average and volume readings written with specific price levels
- A contrarian confirmation verdict in JSON and Markdown with swing levels, when you ask for that mode
Common Use Cases
- Reviewing the weekly chart of an index or stock before a weekly planning session
- Comparing several crypto or forex weekly charts with the same method and a report for each
- Checking whether a market flagged as crowded is showing a weekly reversal pattern
What Makes It Unique
The analysis follows one disciplined sequence and forces probabilities that sum to 100 percent, each with an invalidation level, so scenarios are comparable from chart to chart. The contrarian mode adds a deterministic script and a conservative rule: when the chart read and the script disagree, the verdict becomes not confirmed.
Under The Hood
In chart mode the skill reads your images, loads a bundled framework reference and a report template, and writes Markdown files into a reports folder. It needs no key and no network access in this mode. The contrarian mode can run a Python script from the shell that calls the Financial Modeling Prep API for weekly price bars, which needs an API key, and writes JSON and Markdown reports. When prices are missing or too short, the script returns an insufficient-data verdict and invents nothing.