FinzData Weekly

FinzData Weekly: Liquidity, Capital Flows & Positioning, September 26, 2026

Liquidity, capital flows and positioning for the week to September 26, 2026, built from the FinzData database. Data, not investment advice.

Download the deck (PDF)   PowerPoint · CSV · JSON

This week in five lines

• Fed net liquidity stands at $5,770bn as of 2026-09-23, falling over the past four weeks (−9bn).
• The Treasury General Account is rising over four weeks (+26bn) to $977bn.
• M2 money supply is 6.1% versus a year ago at $23,306bn.
• Most crowded positioning: US Treasury bond (Leveraged funds, index 100) and Copper (Managed money, index 100).
• Largest weekly swing in net % of open interest: British pound (Asset managers) at −18.9pp.

Liquidity

Fed net liquidity is $5,770bn as of 2026-09-23, down 9bn over four weeks, with bank reserves at $2,930bn and the overnight reverse repo at $0.63bn. The Treasury General Account rose to $977bn while M2 money supply is 6.1% versus a year ago at $23,306bn.

Capital flows (TIC)

Net foreign purchases of US long-term securities were $40.616bn in 2026-07, down from $208.899bn prior, with a trailing 12-month total of $1,754.38bn. Net foreign purchases of US Treasuries were −$3.56bn while foreign official net purchases of Treasuries were $25.537bn.

Positioning (CFTC Commitments of Traders)

TFF report (financial futures)

Disaggregated report (commodities)

Crowding and weekly swings

Most crowded (by |COT index − 50|)

• US Treasury bond (Leveraged funds): COT index 100, flag: Crowded long.
• Copper (Managed money): COT index 100, flag: Crowded long.
• Soybeans (Managed money): COT index 100, flag: Crowded long.
• Corn (Managed money): COT index 99, flag: Crowded long.
• Copper (Producers/merchants): COT index 2, flag: Crowded short.

Largest weekly swings (by |w/w pp|)

• British pound (Asset managers): −18.9pp to -47.3% of open interest.
• S&P 500 (Asset managers): +12.7pp to 49.2% of open interest.
• Nasdaq-100 (Leveraged funds): −8.2pp to -12.0% of open interest.
• S&P 500 (Leveraged funds): −7.2pp to -19.7% of open interest.
• Russell 2000 (Leveraged funds): −6.4pp to -26.0% of open interest.

Most crowded positioning by COT index: US Treasury bond (Leveraged funds, index 100), Copper (Managed money, index 100), and Soybeans (Managed money, index 100). Largest weekly swing in net % of open interest: British pound (Asset managers) at −18.9pp, with net positioning at −47.3% of open interest.

Method and sources

All figures are point-in-time for the issue date. COT positioning uses rows dated on or before issue date minus four days (Tuesday positions published the following Friday). Liquidity and TIC series use rows dated on or before issue date minus two days. This lets us rebuild past issues exactly as they would have appeared at the time.

• Liquidity and flows: latest observation, plus the change versus the last observation on or before 7, 28, 91 and 365 calendar days earlier. Monthly TIC flows also show the prior month and a trailing-12-month sum; foreign holdings show the level and the 12-month change.
• Positioning: net = long − short contracts per trader group; net % of open interest = 100 × net ÷ total open interest; w/w change in net contracts and in net % of open interest (percentage points); COT index = 100 × (net − min) ÷ (max − min) over the last 156 weekly rows (50 when max equals min); percentile of net % of open interest versus all history since 2010. "Crowded long" when the COT index is 90 or above, "Crowded short" when 10 or below.
• Sources: Federal Reserve H.4.1, US Treasury, FRED, CFTC Commitments of Traders, and the US Treasury TIC survey, all stored in the FinzData database.

Explore the underlying series: COT positioning, datasets, explore, and the RSS feed.

Pull the same data

import finzdata as yf

c = yf.Client()

# Net liquidity: Fed assets minus TGA minus reverse repo
liq = c.liquidity("FED_NET_LIQUIDITY")

# COT positioning: S&P 500 futures, TFF report
cot = c.cot(report="tff_fo", market="13874+")

# Treasury International Capital flows
tic = c.tic()

Built for AI, quants, data scientists and financial analysts who backtest. Clean data. Real insights. Real alpha. finzdata.com. This material is for information only and is not investment advice.