Every Signal,
Every Report Type
From raw CFTC filings to actionable positioning intelligence. Our pipeline processes 7 report families across 100+ markets and computes 10+ derived analytics on each.
Positioning Analytics Engine
Every feature is computed server-side from CFTC raw data, delivered through a real-time Flutter dashboard.
AI Bias Inference
A lightweight classification model analyzes net positioning across commercial, non-commercial, and retail cohorts. It considers positioning extremes, momentum, and cross-report-type confirmation to output a regime classification: bullish, bearish, or neutral. Each classification includes a momentum qualifier — accelerating, decelerating, or extreme — so you can gauge conviction.
COT Indices (26W & 52W)
For every market, we compute the classic COT index — current net position minus the minimum over the lookback window, divided by the range — for commercial, non-commercial, and retail cohorts. The 26-week window captures medium-term positioning trends; the 52-week window provides a broader historical context. Values range from 0 (short extreme) to 100 (long extreme).
Z-Score Analysis (52W & 156W)
Standard deviation-based positioning analysis. The 52-week z-score highlights positions that have moved more than 2 standard deviations from the mean — a statistically significant shift. The 156-week (3-year) z-score captures regime changes over longer cycles. Both are computed for commercial and non-commercial cohorts separately, giving you a multi-perspective view of positioning extremity.
Smart Money Divergence
Our proprietary divergence index fires when commercial hedgers (smart money) and non-commercial speculators move in opposing directions over consecutive weeks. Historically, these divergences often precede significant price reversals. The signal is strongest when commercials are adding longs while speculators pile into shorts — the classic "smart money vs. dumb money" setup.
Speculative Bias & Conviction
Tracks the directional conviction of non-commercial (speculative) traders across four levels: Neutral, Mildly Bullish/Bearish, Bullish/Bearish, and Aggressively Bullish/Bearish. Determined by the magnitude and consistency of net position changes relative to open interest. When speculators reach extreme conviction levels, it often signals positioning is stretched.
Dumb Money Index (Retail Positioning)
A contrarian indicator derived from the net positioning of non-reportable (retail) traders. When the dumb money index reaches extreme levels (above 80 or below 20 on the 52-week scale), institutional flow often reverses direction. Tracked alongside commercial positioning to identify potential turning points.
Seven CFTC Report Families
Each report type offers a unique lens on institutional positioning. Pro and Enterprise tiers unlock the full dataset including options.
| Report Family | Coverage | Markets | Tier |
|---|---|---|---|
| Legacy (Futures) | Commercial, non-commercial, retail positions | All | Free |
| Legacy (Futures + Options) | Combined futures & options positioning | All | Pro |
| Legacy (Options Only) | Pure options-derived positions | All | Pro |
| Disaggregated (Futures) | Producer/merchant, swap dealers, managed money, other reportables | All | Pro |
| Disaggregated (Futures + Options) | Combined disaggregated positioning | All | Pro |
| Disaggregated (Options Only) | Pure options from disaggregated | All | Pro |
| Traders in Financial Futures (Futures) | Dealer/intermediary, asset manager, leveraged funds, other reportables | Financials | Pro |
| Traders in Financial Futures (Futures + Options) | Combined financial traders | Financials | Pro |
| Traders in Financial Futures (Options Only) | Pure options for financial traders | Financials | Pro |
| Supplemental (Futures + Options) | Concentration ratios, top-4 and top-8 trader positions | Major markets | Pro |
How It Works
From raw CSV to actionable signal — our data pipeline is fully automated.
Retrieval
Every Friday at 15:30 ET, our Cloud Function fetches the latest CFTC CSV files from cftc.gov for all 7 report families. Historical data is cached locally during the initial backfill.
Processing
Raw DataFrames are routed through per-report-type processors that compute net positions, indices, z-scores, speculative bias, dumb money index, and concentration ratios.
Loading
Processed data is upserted to PostgreSQL (via Firebase Data Connect) through auto-generated GraphQL mutations. Each record is keyed on [cftcId, reportDate] for idempotent updates.
Dashboard
The Flutter web app queries Data Connect via generated SDK. Data is rendered as interactive line charts with COT index, z-scores, and AI bias overlays. Auth-gated tiers control report type and market access.
Ready to See Inside the Markets?
Join traders who track institutional positioning weekly. Free tier available.
Launch Dashboard