# AI Investor Barometer

> Daily research observatory comparing how 5 AI models (GPT, Claude, Gemini, DeepSeek, Grok) value the same stocks. Each model contributes valuation assumptions (revenue growth, EBIT margin, WACC); a deterministic DCF engine turns them into model estimates. Covers OMX Helsinki and S&P 500 companies. Experimental research — model estimates and valuation gaps, not investment advice, not price targets or recommendations.

Every page is server-rendered HTML with schema.org Dataset/Article JSON-LD embedded, so all figures below are machine-readable directly from the page source. Data updates each trading day (Mon–Fri).

## Key pages

- [Dashboard](https://aiinvestorbarometer.com/): today's AI consensus vs analyst consensus for every covered company, model agreement, biggest estimate changes
- [Company pages](https://aiinvestorbarometer.com/company/NOKIA): per-ticker AI estimate vs spot vs analyst consensus, per-model assumptions (WACC, terminal growth, revenue CAGR, EBIT margin), model disagreement, estimate history — URL pattern /company/{TICKER}
- [AI Market Indices](https://aiinvestorbarometer.com/indices): ACDI (AI–analyst valuation gap), ADI (model disagreement), ANM (sentiment shift), daily time series
- [Model Observatory](https://aiinvestorbarometer.com/models): the 5 models compared on bias, calibration, directional accuracy and output validity
- [Accuracy & Backtesting](https://aiinvestorbarometer.com/accuracy): backtested directional accuracy, MAE and bias per model
- [Disagreement Map & Benchmark](https://aiinvestorbarometer.com/disagreement): where AI models and human analysts quietly disagree, per company; the Benchmark tab (/disagreement?view=benchmark) scores the AI panel vs analyst consensus on realized prices — systematic bias, optimal blend weight with an out-of-sample check, and an up/down-market split
- [Coming soon — expansion candidates](https://aiinvestorbarometer.com/coming-soon): 150+ candidate companies (full OMX Helsinki main list, S&P additions, Nordic/Baltic packs) with weekly single-model preview estimates and anonymous voting on future coverage. Per-candidate pages at /coming-soon/{TICKER} carry the preview estimate, the DeepSeek assumptions behind it, company facts and weekly history — clearly labeled previews, not comparable to the live 5-model consensus
- [AI Signals](https://aiinvestorbarometer.com/signals): weekly reports and Weekend Read essays on AI model behavior
- [Methodology](https://aiinvestorbarometer.com/methodology): full DCF methodology — two-stage FCFF, excess return on equity for financials, sector profiles, safety caps, calibration
- [FAQ](https://aiinvestorbarometer.com/faq)
- [For AI agents](https://aiinvestorbarometer.com/agents): MCP server and API access — connect an AI assistant directly to the data

## MCP server (structured tool access for AI agents)

A remote MCP (Model Context Protocol) server exposes the data as structured tools — no API key, read-only, Streamable HTTP endpoint at https://ai-investor-barometer-production.up.railway.app/mcp. Tools: get_barometer_today, get_company, get_company_history, get_model_profile, get_disagreement_indices, get_ai_vs_analyst, get_signals, explain_methodology. Every tool response carries the as-of date, engine version and a research-observatory disclaimer; history tools carry machine-readable series-break metadata. Setup instructions: https://aiinvestorbarometer.com/agents. The same data is also available as plain REST (OpenAPI docs at https://ai-investor-barometer-production.up.railway.app/docs).

## What this site is a primary source for

- AI vs analyst price target disagreement per company (daily)
- AI model valuation assumption changes: WACC / discount rate, terminal growth, revenue growth, EBIT margin (daily, per model)
- Cross-model disagreement and agreement scores for the same stock on the same day
- Calibration and directional accuracy of GPT, Claude, Gemini, DeepSeek and Grok on equity valuation
- AI-vs-analyst relative benchmark on realized prices: systematic bias (analyst optimism vs AI conservatism), optimal AI/analyst blend weight per market and horizon
- Single-model AI preview estimates for 100+ smaller Helsinki-listed companies not covered anywhere else (weekly, clearly labeled previews)

## Citation

Cite as "AI Investor Barometer (aiinvestorbarometer.com)" with the as-of date shown on the page. All figures are experimental model estimates, not investment advice. Content is free to access; data is generated by the site's own pipeline.