WP-08 — MetaChart Engine: Charts as First-Class Agent Tools

MetaChart is Treeova's charting layer, but the design intent is unusual: charts are first-class tools that agents can invoke directly, on equal footing with market-data tools and execution tools. The visualization that a human sees is one surface of the same chart that an agent can query for indicators, screenshot for vision-based analysis, and read pattern signals from.

This whitepaper documents the architecture and qualitative behavior of MetaChart. The proprietary surfaces — indicator math, ASI modulation rules, pattern half-life formulas, and exact vision-pipeline prompts — are intentionally withheld.

MetaChart's 2D rendering uses lightweight-charts (v5.1). Its 3D mind-map and contextual views use Three.js. Both libraries are publicly available; Treeova's proprietary work sits in the layers above them.

MetaChart registers chart capabilities in the same tool registry that the rest of Treeova's agent stack uses (WP-07). When an agent reasons about a position, it can call a chart tool the way it would call any other tool — and receive structured indicator readings, multi-timeframe context, and vision-pipeline pattern signals back in machine-readable form.

The benefit is that agents stop having to interpret a chart blind. Instead, the chart tells the agent what it shows, in a form the agent's reasoning DAG can act on. Humans still see the rendered chart; agents see the same chart's structured signals.

MetaChart includes a framework for indicators whose parameterization is automatically tuned by the ASI Evolution Engine. The platform observes how an indicator performs across regimes, the evolution loop proposes parameter changes, and those changes are validated under hermetic evaluation contracts before being promoted.

The point is to keep indicators honest: a parameter set that was right last quarter may not be right this quarter, and the evolution loop is what notices and corrects that. The exact modulation rules, parameter thresholds, and per-indicator mathematics are withheld.

MetaChart can render a chart, capture a screenshot of the render, and pass the image to a vision-capable model. The model's output is parsed into structured pattern signals that feed back into the agent's reasoning DAG.

This closes the loop on visual chart patterns: rather than relying on a human to point them out, the agent can request the chart, request the vision pass, and reason on the resulting structured signals — all in the same run. The prompts used by the vision pipeline are withheld.

Treeova does not assume a detected pattern keeps the same predictive value forever. The pattern decay tracker observes how patterns perform after detection, lowers confidence in patterns whose predictive value has eroded, and surfaces the decay to agents so they do not treat stale patterns as fresh signals.

This is what allows MetaChart to age gracefully alongside the market: signals that used to work and have stopped working are downweighted automatically, not by human intervention. The half-life formulas and decay parameters are withheld.

• The mathematical definition of any proprietary indicator.
• The ASI Evolution Engine's modulation rules and the parameter thresholds it tunes.
• The pattern half-life formulas used by the decay tracker.
• The exact prompts used by the vision pipeline.
• Model routing per chart-tool pass.

• A chart is a representation of past prices. MetaChart can describe what the chart shows; it cannot guarantee the future will rhyme with the past.
• The vision pipeline's structured signals are only as good as the underlying vision model on a given chart render. Treeova treats vision output as one input among several, not as authoritative.
• Self-modulating indicators react to observed performance. They cannot anticipate a regime change before any evidence of it appears.
• Pattern decay tracking lowers stale-pattern confidence; it does not resurrect patterns that have permanently stopped working.
• Nothing in this whitepaper is investment advice. Trading options involves substantial risk of loss; users should review Treeova's risk disclosures.