AdCho · About Us
The right brain — and the ground it stands on.
AdCho is a qualitative research layer over QuantOrb, developed by kfchow.com — the lab behind the LLM Value Leaderboard. We read the SEC filings behind the money flow — institutional 13F holdings, 10-K/10-Q report text, 8-K earnings press releases — together with market news and sentiment and the quantorb 13F & money-flow datasets, and turn them into AI-powered, document-grounded readings: per-ticker tear sheets.
AI-powered readings of the SEC record — no price targets, not investment advice.
Why two brains
Facts and judgment live on different planes.
Reading a market takes both facts and judgment — and blurring the two is how research loses your trust. So we built two sites. QuantOrb is the left brain: deterministic formulas, a 28-year back-test, SEC-sourced institutional data, 0% AI predictions — the ground you can verify. AdCho is the right brain: AI that reads the same record and writes it up in plain language — the reading, always labelled as AI. AdCho never ships an EDGAR product: it ships readings of one, with AI-powered insights.
QuantOrb is the ground, AdCho is the reading of it — verify the ground, and treat the reading as one input to your own judgment. Context, not proof.
The quantitative half
AdCho is the qualitative layer. The numbers live at QuantOrb — where the money is flowing, and what the record says about how those signals have behaved.
Sources & provenance
Every document citation carries its EDGAR accession number and its ingestion path.
Facts (verifiable)
SEC EDGAR (13F-HR, 10-K/10-Q, 8-K) via the QuantOrb document corpus, and the quantorb 13F & money-flow datasets.
Market context (live feeds)
Market prices, technicals and ticker & global news flow (Yahoo Finance — Reuters & Bloomberg items as syndicated), macro indicators (Federal Reserve FRED), event markets (Polymarket) and X social sentiment — ingested per engine run, with per-lane gaps labelled on every sheet.
Reading (AI-powered)
AdCho synthesis by a multi-analyst AI engine. Qualitative context, not proof.
How this engine was validated
The multi-analyst architecture behind AdCho has been studied in academic research.
A 2025 preprint from UCLA and MIT researchers (arXiv:2412.20138) describes a multi-agent LLM framework for financial trading — analysts, traders, and risk managers collaborating in a simulated trading floor. The study's authors report that their framework achieved at least a 23.21% cumulative return and a 24.90% annual return on three sampled stocks over a three-month backtest window, surpassing rule-based baselines by a margin of 6.1%. The authors note the Sharpe ratios (5.6–8.2) exceeded their expected empirical range and attribute this to the short window with few pullbacks.
AdCho's engine is an independent re-build of this multi-agent architecture, adapted for qualitative research rather than trading. The architecture draws on the same academic preprint that framed this field — the study's findings are the authors' own, not AdCho's performance claims.