FinUties · Market Regime · SOFR upper tail narrows to 3.62%

Updated at 25 August 2026.

Scope

This edition of Market Regime interprets cross-asset conditions through FinUties money-market and positioning feeds. It is explanatory research, not investment advice. Conclusions are conditional on the endpoints that returned usable observations this cycle; where feeds were empty, the narrative states the gap rather than inferring missing levels.

Introduction

This edition asks: why are positioning and rates turning together (or not), and into what regime — concentrated open interest or a broader rates read?

Evidence answer (as-of 2026-05-19 UTC): SOFR’s 75th percentile held at 3.58% and its 99th percentile printed 3.62%—three basis points below the prior stored reading of 3.65%—while the 25th percentile stayed at 3.52% and the 1st percentile edged to 3.50% from 3.51%.. That is the read from prints already in this edition — not a forecast and not advice.

SOFR’s 75th percentile held at 3.58% and its 99th percentile printed 3.62%—three basis points below the prior stored reading of 3.65%—while the 25th percentile stayed at 3.52% and the 1st percentile edged to 3.50% from 3.51%. FinUties planned five core New York Fed rate endpoints and two CFTC catalogue paths for this pull; only the secured overnight financing rate series delivered parseable observations. Companion feeds for the effective rate, overnight bank funding rate, tri-party general collateral rate, and broad general collateral rate returned no economic fields; CFTC metadata and product-hierarchy facts arrived without position-level observations. That asymmetry frames the central question: whether money-market tail compression and the extreme dealer-short positioning documented in prior Market Regime episodes still describe the same regime label, or whether leadership has narrowed to a rates-only signal. The deep dive below walks through what the SOFR distribution implies for funding dispersion, what prior CFTC readings still bind in memory, and where cross-domain confirmation remains incomplete.

Try this query

Reproduce the P0 pull for this edition:

• REST: GET /api/v1/positioning/legacy_futures-facts?limit=25
• MCP query args:

{"domain": "positioning", "resource": "legacy_futures-facts", "limit": 25}

Provenance: P0 domain positioning. As-of this edition 2026-05-19 UTC. Informational only — not investment advice.

Deep dive

Money markets and overnight funding

The SOFR distribution this cycle shows a stable core with a narrower tail. The 25th and 75th percentiles are unchanged at 3.52% and 3.58% respectively, anchoring the middle of the distribution. Against that fixed inter-quartile spine, the 1st percentile edged one basis point lower to 3.50% (prior: 3.51%), and the 99th percentile fell three basis points to 3.62% (prior: 3.65%). The full 1st–99th band therefore spans 12 basis points, down from 14 basis points on the prior pull—a modest compression of upper-tail dispersion rather than a wholesale repricing of the centre.

Central-rate readings within the same series show a slight downward drift on one observation and deliberate stability on another. One percent-rate print registered 3.53% (prior: 3.55%), accompanied by roughly $3.16 trillion in reported volume; a second percent-rate observation held at 3.55% with volume near $3.16 trillion. The coexistence of a two-basis-point dip and an unchanged central print is consistent with a market where the modal trade cluster is stable but the upper tail is less elevated than the last stored edition—not with acute funding stress or a broken corridor.

Mechanistically, a tightening upper tail with an unchanged 75th–25th block typically signals less dispersion in large overnight repo trades, not necessarily a change in policy stance. When only one rate endpoint among five planned money-market feeds is observable, however, FinUties cannot cross-check whether EFFR, OBFR, BGCR, or TGCR corroborate the same compression. The SOFR reading alone supports a label of orderly, concentrated overnight funding; it does not, by itself, confirm a broader cross-asset regime.

Positioning and futures-market structure

The CFTC domain contributed catalogue structure without fresh position observations this cycle. Metadata returned twelve rows but no parseable positioning fields; product-hierarchy facts listed active contract families—including agricultural indices, energy proxies, and spread products—but carried empty value arrays for each label. FinUties therefore cannot update managed-money or dealer positioning shares from this pull alone.

Prior Market Regime editions on this node remain binding context. The most recent stored narrative documented dealer short shares above 85% across equity-index, Treasury, and SOFR futures, alongside a sharp rotation in pollution allowances where managed-money long share of open interest rose from 5.7% to 25.4% (+10.5 percentage points). Other prior prints showed 100% dealer short share on two-to-seven-year interest-rate swap contracts and the Bloomberg Commodity Index, 78.1% on European currency unit futures, and 85.8% on non-major crypto contracts. That pattern describes narrow, dealer-dominated hedging leadership in rates and macro-beta instruments—a positioning extreme that prior editions linked to the same SOFR percentile band now re-printing at 3.58–3.62%.

The tension this cycle is explicit: rates data suggest tail compression; positioning memory suggests persistent dealer short concentration. Without a fresh CFTC observation set, FinUties cannot test whether those extremes have mean-reverted, intensified, or rolled into adjacent contract classes (as the pollution-allowance swing might imply). Leadership therefore reads as narrow on positioning (dealer-heavy, product-specific) and broad only at the level of stable overnight funding—not as a unified cross-domain regime confirmation.

Structural backdrop: OECD fiscal aggregate

General government gross debt for the OECD aggregate stands at 112.0% of GDP, unchanged from the prior stored reading (observation dated end-2025). This is a slow-moving fiscal stock, not a cyclical market signal; FinUties cites it as structural context for Market Regime rather than as a trigger for this edition’s label. An unchanged aggregate debt ratio neither amplifies nor contradicts the SOFR compression narrative—it confirms that the sovereign balance-sheet backdrop against which rate and positioning cycles play out has not shifted between pulls.

Continuity and build-up

This cycle’s new information is modest but directional: SOFR’s 99th percentile fell three basis points to 3.62%, the 1st percentile ticked one basis point lower to 3.50%, and one central percent-rate reading dipped to 3.53% from a prior 3.55%, while the 25th and 75th percentiles held steady at 3.52% and 3.58%. What persists unchanged is the core funding band, OECD aggregate gross debt at 112% of GDP, and—by absence of contradicting data—the dealer-short concentration described across prior Market Regime editions on equity, Treasury, SOFR, and macro-index futures. What accumulates across editions is a longitudinal story of tight overnight rates coexisting with extreme futures positioning: each successive pull that reprints the 3.58–3.62% upper band without fresh CFTC position data adds another observation that money markets look orderly while positioning extremes remain unverified for mean reversion. The pollution-allowance managed-money swing (+10.5 pp) in the prior edition further suggests leadership may be rotating at the margin even as rates stability dominates the observable surface.

Synthesis

On the evidence available, Market Regime this cycle aligns partially across domains. Money-market data support a low-dispersion, stable-centre funding regime with a slightly cooler upper tail. Positioning memory from prior pulls supports a dealer-heavy, narrow-leadership futures structure that has not been refreshed. The research questions posed at planning—regime-label agreement, breadth of leadership, and whether positioning extremes match rate behaviour—receive qualified answers: rates and remembered positioning are not contradictory (tight funding can coexist with dealer hedging), but they are not fully confirmed either, because four of five planned rate feeds and all CFTC position fields were empty. Confidence in a single regime label is therefore conditional and domain-split until positioning feeds repopulate or companion NY Fed rates return observations.

What to watch

• SOFR tail percentiles: whether the 99th percentile stabilises near 3.62% or re-expands toward the prior 3.65% reading, and whether the 1st percentile holds at 3.50%—jointly signalling continued compression or renewed dispersion.
• Companion NY Fed rate feeds: reappearance of observations on effective, overnight bank, tri-party, and broad general collateral rates would allow a cross-check that SOFR compression is system-wide rather than series-specific.
• CFTC positioning refresh: dealer short shares on equity-index, Treasury, and SOFR futures—the binding prior signal—need updated prints to test whether extremes persist or begin to unwind.
• Product-level rotations: managed-money share shifts in less-central contracts (as seen in prior pollution-allowance data) may indicate narrow leadership migrating before it appears in headline macro futures.
• OECD aggregate debt: unchanged at 112% of GDP for now; any future revision would alter the slow fiscal backdrop against which rate and positioning cycles are interpreted.

Methodology and confidence discipline

FinUties treated this edition as a cross-domain regime read requiring agreement between rates and positioning, per the stream research plan. Endpoint insights were interpreted numerically where observations existed; empty feeds were recorded as gaps, not imputed. Direction-of-change statements on SOFR percentiles and percent-rate readings were assessed against stored prior pulls on this node and against prior Market Regime blog editions where positioning context was last observed. Where dual percent-rate observations appeared within the same pull, both were reported and the direction of change was assessed against stored priors rather than collapsed into a single synthetic level.

Confidence is highest on SOFR distribution mechanics (multiple percentile observations with prior anchors) and moderate on regime-label inference (single-domain confirmation). Confidence is low on positioning updates and on whether futures extremes still match rate behaviour, because CFTC position fields were absent this cycle. OECD debt is cited at aggregate scope only. No endpoint observations were invented for thin or empty feeds.

FinUties data sources

• /api/v1/rates/nyfed-bgcr-series
• /api/v1/rates/nyfed-effr-series
• /api/v1/rates/nyfed-obfr-series
• /api/v1/rates/nyfed-sofr-series
• /api/v1/rates/nyfed-tgcr-series
• /api/v1/cftc/metadata
• /api/v1/cftc/product_hierarchy-facts