Editorial standards & corrections
cotreport is a research service, not a signal seller. These are the standards behind every number, and how we handle it when something is wrong.
Where the data comes from
Positioning data is the U.S. CFTC's public Commitments of Traders report (legacy, disaggregated and TFF). Prices are continuous front-month proxies; roll gaps are included and no costs or slippage are assumed. cotreport is not affiliated with, endorsed by, or sponsored by the CFTC. Every figure traces back to these public sources. See the methodology.
Nothing is fabricated
Every statistic on the site is computed from the data above and carries its sample size, baseline, and correction status. We do not invent contract counts, outcomes, testimonials or ratings. Where a market's data is unfit to measure (for example VIX, whose front-month proxy is the wrong instrument), we publish no outcome rather than a misleading one.
How results are graded
Rules are fixed in advance, applied mechanically, and confirmed out of sample: trained on 2006–2018 data, then tested on 2019-onward data the rules never saw. Findings are corrected for multiple testing (Benjamini–Hochberg FDR at q=0.10). We publish the strategies that failed alongside the ones that held, including our own ideas.
How corrections work
Numbers are frozen per research release (r1, r2, …) so a published figure never silently shifts underneath a reader. When a method changes or an error is found, we issue a new release and document what moved and why on the methodology page, rather than quietly editing old numbers. If you spot an error, write to support@cotreport.ai and it gets checked.
What we always publish
Every case, every baseline, and every correction. The weekly free file, one complete backtest file, the full 44-market ledger and the entire methodology stay open, so any claim here has a free surface you can inspect before paying. The map is measured and public; the trade is yours.
Statistics, not advice · Not affiliated with the CFTC · results frozen per research release