Every morning at 06:00 GMT, we deliver a structured raw dump of every UK iXBRL filing in the last 24 hours, fully tagged, ratio-extracted and signal-flagged. Operational benchmarks, Ghost-Tech anomalies, employee-growth deltas. Backtestable five years deep, point-in-time clean.
A raw data dump every morning at 06:00 GMT, every UK company that filed in the last 24 hours, ratio-extracted and signal-flagged. Be first to trade on the 24-hour pulse.
Compare any UK company's operating metrics against live SIC-peer averages, revenue-per-employee, gross margin, staff-cost ratio. Surfaces operational alpha bureaus miss.
Detect high-revenue entities with suspiciously low staff counts, shell companies, hyper-efficient AI-driven firms, or transfer-pricing artefacts. Either way, it's tradable.
Alpha signals are the factor surfaces we publish daily against the structured iXBRL stream. The headline number is filing-to-signal latency, under five minutes from a Companies House publication event to a tagged, ratio-extracted, signal-flagged record in your S3 bucket. That latency is the alpha: by the time bureau-level data and second-hand reports settle into commercial datasets, a quant desk wired into the Pulse has had between five and forty-eight hours to model the filing.
The factors we expose include rapid employee growth, threshold crossings, ghost-tech anomalies, margin-decay clusters, going-concern flags, director-resignation clusters and sector-peer outperformance. Each is point-in-time clean, we publish the version of the company state that was knowable on the morning of the filing, not the latest revision. Five years of point-in-time history is available, with no look-ahead bias, for backtesting.
Alpha signals are sold as a feed (raw daily delivery) or as a per-factor derivative product (a backtested series for a single thesis). The feed suits desks that prefer to assemble their own factors; the derivatives suit smaller teams that want a tested signal off the shelf. Either way, the underlying iXBRL filing and the exact ratio that fired are cited on every record, every signal is traceable end-to-end to a Companies House publication.
The Daily iXBRL feed is a structured raw dump of every UK iXBRL filing accepted by Companies House in the previous twenty-four hours, delivered at 06:00 GMT to your S3, GCS or SFTP endpoint. Volume runs at approximately 340,000 records per business day across all UK private and public-listed entities filing that day, though sample-day volumes range from 80,000 (Boxing Week, August Bank Holiday) to over 600,000 (the September and December filing peaks).
Each record is fully tagged: company number, registered name, filing reference, accounting-period dates, every iXBRL element value, every derived ratio (current, quick, debt-to-equity, EBITDA margin, revenue-per-employee, staff-cost ratio) and the full set of Pulse signal flags that fired on that filing. Delivery formats are JSON, Parquet and CSV; the schema is point-in-time stable and versioned.
For quant desks the feed is the cleanest form of UK private-company alt data on the market. There is no overlap with bureau data (we are upstream of the bureaus), no overlap with web-scraped alternative data (we are sourcing the regulator), and no scraping fragility, Companies House is the regulator and the format is their published standard. The five-year point-in-time history that comes with the feed lets the desk backtest any factor without look-ahead bias.
Sector benchmarking is the comparative layer over the feed. For every UK SIC code we maintain live distributions for revenue, EBITDA margin, gross margin, revenue-per-employee, staff-cost ratio and current ratio, refreshed on every new filing. A single company is then placeable on the distribution: percentile rank, Z-score, and percentile movement vs the prior year.
For a quant model this turns a raw filing into a comparative alpha signal. A £10M-turnover engineering firm in isolation is noise; the same firm sitting at the 95th percentile of revenue-per-employee inside SIC 28140 with a positive twelve-point Z-score movement YoY is a tradable signal. The benchmark layer lets models run percentile- and Z-score-based factors directly off the feed without rebuilding the comparison set in-house.
Region-level distributions sit alongside the SIC distributions: London, Manchester, Edinburgh, Belfast, Birmingham. Combined, they support strategies that rotate around sector–region pairs (high-margin engineering in the North, distressed retail in the South-East, ghost-tech in the financial-services cluster around the City of London). Region distributions are a published product in their own right, see the regional benchmarks index for the full set.
Tell us your factor — Ghost Tech, Rapid Growth, Margin Decay, Threshold Crossing — and we'll send a 5-year point-in-time sample with no look-ahead bias.
Request backtest sample
