About

Who builds NCSX —
and what we refuse to do

SpreadBench is an independent data benchmark: it turns public secondary trades in AI-infrastructure (neocloud) bonds into NCSX, a continuous, tiered, fully replicable credit-spread index.

Methodology v0.1.1 DOI 10.5281/zenodo.23931330 Press kit
Thesis

Why we exist

The market prices compute credit every day, but the answer is scattered across CDS quotes, single-bond yields and footnotes in sell-side reports — no common measure, no history, no public method. We fill that layer with public trades: no guessing, no ratings, no recommendations.

Aggregate

FINRA TRACE public secondary trades plus the US Treasury curve — spread per bond, then face-value weighted into an index.

Publish

Full methodology, formulas, version and DOI are public; data is CC BY 4.0, citation needs no permission.

Replicate

Every basis point traces back to a CUSIP and a trade. REST API, MCP and static JSON — people and machines read the same numbers.

Red lines

What we do not do

These three do not bend with commercial pressure — they are the precondition for being citable.

No credit ratings：We do not score issuers or set outlooks — that is a rating agency's job.
No investment advice：No recommendations, no price targets, no directional calls.
No resale of raw trades：We deliver SpreadBench-derived products only, never TRACE raw records.
Team

Team and collaboration seats

We organise by seats: each seat publishes its scope and the background it needs, and is filled seat by seat. These are the 10 seats currently on the roster.

Head of Index Methodology01

Head of Index Methodology

Owns the rulebook: tier thresholds, constituent entry/exit and replication audits — and explains where every basis point comes from.

Head of Credit Research02

Head of Credit Research

Neocloud credit fundamentals, bond-by-bond coverage and the weekly note — turning market narrative into spread moves.

Data Engineering Lead03

Data Engineering Lead

TRACE ingestion pipeline, identifier verification and the reliability of the T+1 17:30 ET publishing window.

Data Licensing & Compliance04

Data Licensing & Compliance

TRACE licence terms, redistribution boundaries and the line between CC BY 4.0 and commercial licensing.

Platform & API Engineer05

Platform & API Engineer

REST endpoints, static rendering, caching and rate limits — keeping the JSON an agent reads identical to the number on the page.

AI & Agent Engineering06

AI & Agent Engineering

MCP server, agent consumability and cross-model consistency — so machines citing us do not get it wrong.

Statistical QA07

Statistical QA

Driving identifier verification from 5/20 to 20/20, outlier screening and replication spot-checks.

Media & Communications08

Media & Communications

Press kit, journalist relationships and citation hygiene — making "NCSX-Composite (SpreadBench)" the standard form.

Product & Commercial09

Product & Commercial

Subscription tiers, enterprise licensing and the early-bird programme — this seat guards the "free tier stays free" line.

Developer Relations10

Developer Relations

Embeddable badges, sample code and community answers — lowering the cost of a first successful API call.

Seat-based by design: the entries above are functional seats and role briefs (3D figures are illustrative), not completed hires. Each seat is replaced with a named person once the collaborator is confirmed.

Contact

How to reach us

One mailbox, three purposes. We reply on business days.

Media & citation

Data, charts, methodology explainers or interviews
hi@spreadbench.com

Data & licensing

Subscriptions, enterprise licensing, embedded distribution
hi@spreadbench.com

Partnerships & feedback

Methodology challenges, data corrections, seat collaborations
hi@spreadbench.com

SpreadBench is backed by Chainfir Capital; index construction and publication operate independently.