Public markets, connected to institutional books

Three things, kept together. A daily market record: par SOFR, €STR, EURIBOR and SONIA curves stripped from the public tape; swap spreads to Treasury and to the ECB AAA curve; forward-start swap flow and traded forward coupons; swaption normal-volatility surfaces; SOFR cap and floor surfaces by strike and term; TIPS asset swaps and inflation swaps; CDX and iTraxx index levels with roll-aware changes. A record of institutional books: the derivative positions life insurers disclose in NAIC Schedule DB, parsed to position level across annual and quarterly statements, with their collateral regimes and product footprints. And the connection: each day, the tape is scored for prints whose structure — tenor, forward gap, strike against forward, size, clearing status — matches what those books hold. A match is a shape, never a name; the public tape does not carry counterparties or direction, and the site never attributes a print to a firm.

The morning note on each page is written from the day’s derived figures; every number in it appears in the tables beneath it. Research notes collect the longer analyses that arise from the series.

The Observatory is a research product of ruxton.ai (Ruxton Advisors), created and edited by Terry Leitch, who ran liability-hedging programmes on the buy side before building this. It is not affiliated with any exchange, data vendor, insurer or university. Requests, corrections and questions: rates@ruxton.ai, or the Requests section at the foot of any page, where replies are posted in public.

It means the inputs are open: every series on the site can be rebuilt by anyone from the public sources above, and the method for each series is written out on its page. The derived series are published as JSON files (below) and may be used with attribution. It does not mean the raw tape is redistributed — DTCC’s and ICE’s terms are respected, so no trade-level records appear here — and it does not, at present, mean the pipeline code is published. If that changes it will be announced on the home page.

On weekdays the pipeline runs at 07:30, 10:00 and 18:00 ET. The 07:30 pass fetches the previous day’s final cumulative tape from DTCC (rates and credits), Treasury and ECB reference data, re-derives every series, writes the morning notes, and publishes; the two later passes add the day’s intraday slices and mark the result provisional until the next morning’s final tape replaces it. On Saturday the 07:30 pass lands Friday’s final tape and writes the Week in review, which stays on the page bar through Monday and is then archived. Traffic and feedback reports run with each pass. Every number in a morning note is a value from the derived series; where the model that writes the notes is unavailable a templated summary is used and labelled as such.

Curves. Par rates are IQR-trimmed medians of outright spot-start prints by tenor each day; forwards and discount factors come from an annual bootstrap of the par curve. Swap spreads are par swap minus the Treasury par yield at the same tenor.

Forward starts. Classified by the gap from trade date to effective date and by tail; traded forward coupons are compared with the par-implied forward from the same day’s curve; package legs are separated from outright prints by the tape’s package indicator (populated from 2026).

Options. Swaptions are priced to a Bachelier normal volatility against the day’s forward and annuity, with a venue rule that corrects straddle legs on which the whole-straddle premium was reported. Caps and floors are priced as caplet strips to a flat normal volatility, with payment frequency from the tape.

Credit. On-the-run series are the latest expiration with sufficient prints; indices quoted in price are converted to equivalent spread with a flat-hazard ISDA-style model (40% recovery, quarterly, SOFR discounting); changes across a roll are computed within series.

Europe. EUR and GBP series are the US repository’s view of those markets: levels are informative, volumes are not market-wide.

Statutory books. Schedule DB is parsed from the companies’ posted PDFs; positions are deduplicated across statements on insurer, trade date, maturity and notional so that execution histories can be reconstructed; the parser’s coverage of each statement is reported on the Insurers page.

A print is counted as insurer-linked when its structure falls inside the ranges of structures the statutory books disclose — for example a receiver swaption two to seven years into a ten-to-twenty-year tail struck well below the forward, or a swap starting ten or more years forward with a 10–30-year tail, or a bond-dated asset swap. The ranges are published on the Insurers page beside the positions they come from. The count says that the market traded shapes that insurers hold; it does not say that an insurer traded them. Direction is inferred from the filings, not read from the tape.

The tape is one US repository, so counts understate the market while levels are representative. Notionals are rounded and capped, so notional changes are floors. Package premiums are reported on one leg or not at all, so roughly half of swaption and cap prints cannot be priced. Statutory parsing is imperfect: direction and coupon are unreadable for a share of positions, descriptions are sometimes run together, and positions that have been reinsured leave the schedules together with their liabilities. Daily statistics assume independent days. Every page states which of these apply to it.

Derived aggregates only from DTCC public dissemination data, ICE Trade Vault public data, Treasury and ECB reference series, and insurers’ public statutory statements. No raw trade records or licensed data are republished. Nothing on this site is investment advice. Requests and corrections: rates@ruxton.ai.