We build proprietary models that measure how investors discuss financial assets in public forums — and turn that into systematic signals for trading our own capital.
Proprietary NLP models read public financial discussion from licensed data sources and score, post by post, directional stance and quality of reasoning on the assets discussed.
Per-post scores are aggregated into time-weighted, per-asset sentiment indices. The output is purely numerical and aggregate — no individual-level tracking, no content redistribution.
The indices feed rule-based entry and exit decisions on liquid US ETFs and equities, executed exclusively with proprietary capital. We have no external clients and sell no data.
All source data is obtained through official, licensed APIs under their respective developer agreements.
We analyze markets, not people. Signals exist only at the asset level; underlying content is never displayed, resold or shared.
Privately held and self-funded. Our only customer is our own trading book — which keeps our incentives simple.