Facteus’s Aug. 5, 2026 announcement is less about a single new dataset and more about scale-plus-continuity: 15 million freshly added U.S. card accounts bring total coverage past 120 million accounts with over eight years of history, while the Onyx UPC product is set to hit 30,000-plus stores and $265 billion in tracked annual spend by the end of Q3, according to the company. For buy-side data teams that have watched card-panel providers lose or reshuffle bank partners mid-backtest, that framing is the point — panel churn, not raw latency, is usually what breaks a signal a quant desk has spent months validating.
The claimed 92% correlation with Census Monthly Retail Sales and 1.8% MAPE across the top 100 tracked tickers are the kind of headline stats every transaction-data vendor now publishes, and they’re worth treating as a starting bid rather than a verdict — Facteus is self-reporting against benchmarks it chose, with no independent audit cited in the release.
A vendor’s own backtest is marketing until a client reruns it on an out-of-sample period.
The practical takeaway for sourcing teams: this is a reasonable moment to request a coverage comparison against whatever card-panel and UPC providers are already in the stack (Consumer Edge, M Science, Numerator-style retail panels, et al.), specifically checking bank-partner overlap and how the new 15-million-account cohort skews by income and geography — Facteus flags high- and lower-income cohort feeds as new, which is exactly where representativeness claims tend to get tested hardest.
"When a panel changes, the impact goes far beyond coverage; it can quickly become a model-integrity problem," said Lorn Davis. "Our goal is to give investors a signal they can audit, backtest and continue relying on as their strategies evolve."
