Proof of Capability · built before any award · public data only
The engine beneath the analysis.
Every number below was produced from one public document, the County Treasurer's monthly investment report for March 31, 2026. No privileged access, no data the County hasn't already published. We parsed all 401 coupon securities, reconstructed each one's cash flows, solved the market yield its reported price implies, and computed the portfolio's risk analytics from the ground up. This is the engine that produces the headline findings, and the fact that it runs on public data is the strongest evidence of our independence.
How the engine works
Three independent cross-checks
An engine built from public data is only credible if it reproduces the numbers the County publishes separately. It does, on the two it can fully see, and it precisely quantifies the third.
| Check | HSG recast (public data) | County / report | Verdict |
|---|---|---|---|
| Aggregate market value | $9,220M | $9,220M | Reproduced to the dollar |
| Weighted-avg maturity | 2.32 yr* | 2.31 yr · 842 days | Match within 0.01 yr |
| Portfolio duration | 2.10 yr* | 1.55 yr (option-adj.) | Gap = call optionality |
* Portfolio-level figures add the ~$1.3B cash / money-market sleeve (par-valued, ~0 duration) to the 401-security coupon book so they are comparable to the County's pool-wide statistics. The coupon book alone carries 2.64-yr WAM and 2.42-yr modified duration.
The duration gap is itself a finding.
Our to-maturity duration (2.10 yr at the pool level) runs above the County's reported option-adjusted duration of 1.55 yr. The difference is the embedded call optionality in the agency book: callable agencies behave as if they mature sooner when rates fall, shortening effective duration, but they also cap the upside (negative convexity). On a +100 bp shock the two measures diverge by roughly $59M (our −$223M to-maturity vs. ~−$164M option-adjusted). Public data gets the engine ~95% of the way; the last mile, exact call schedules, is precisely what the security master in the engagement supplies.
The recast, in one picture
Left: each of the 401 securities, repriced from public cash-flow data, plotted by solved market yield against time to maturity. The coherent term structure that emerges, Treasury below agency below corporate below supranational, each on today's curve, is the engine validating itself. Right: par by maturity bucket; the long-duration tail, seen security by security.
What the engine surfaces
| Sector | Count | Par | Market yield | Mod. duration |
|---|---|---|---|---|
| Federal agency | 200 | $5,471M | 4.22% | 2.56 |
| Corporate note | 150 | $1,921M | 4.26% | 2.42 |
| Supranational | 33 | $1,023M | 4.35% | 2.40 |
| U.S. Treasury | 9 | $700M | 3.83% | 1.94 |
| Bank CD / time deposit | 9 | $175M | , | 0.19 |
Why this matters for the County
Three things follow from an engine that runs on public data. First, independence: our evaluation does not depend on the County handing us its pricing feed, we can check the County's numbers against the market ourselves. Second, capability, demonstrated not asserted: you are looking at the working recast weeks before a contract exists. Third, transparency: every figure traces to a public source and a documented cash-flow assumption, the same discipline we bring to the full engagement, where the County's machine-readable holdings and a commercial security master turn this proof into a production system.
All 401 securities repriced from public data, the 75-month total-return series, and the EIP classification, the ~95% proof, built before any award. The production shell that takes it to 100% lives on the Total Return page.