Weekly Market Recap, 19–25 October 2026: Gold and Silver Fall as Crypto Flows Diverge

The weekly market recap for 19–25 October 2026 shows a mixed cross-asset picture. Gold and silver fell during the 19–23 October traditional-market week even as US Treasury yields and the dollar index declined. Bitcoin, Ethereum and Solana also finished lower across the 19–25 October cryptocurrency window, but selected fund flows moved differently: US-listed Bitcoin exchange-traded funds recorded net inflows, while the tracked Ethereum and Solana-linked products saw outflows.

Precious-metals products were similarly split. Estimated outflows from SPDR Gold Shares (GLD) outweighed smaller inflows into iShares Gold Trust (IAU), while iShares Silver Trust (SLV) recorded slightly positive estimated flows despite falling silver prices. These different signals form the backdrop to the Federal Reserve and European Central Bank decisions scheduled for the following week, as of Monday 26 October.

Week at a glance

The following comparisons use the first and last reported observations within each specified source window. Traditional-market prices cover 19–23 October, cryptocurrency prices cover 19–25 October, and US exchange-traded product flows cover 19–23 October.

For the preceding week’s inflation, yield, and fund-flow context, see HFR’s 12–18 October market recap.

Market context: mixed activity and inflation signals

Economic figures gave no single message about the pace of growth or the path of monetary policy. US jobless claims remained relatively low, yet September housing starts fell, and the latest consumer-sentiment survey pointed to continued concern about purchasing power. The flash US composite purchasing managers’ index was still comfortably in expansionary territory, at 57.6 in October, but below September’s 58.4. A PMI above 50 signals rising surveyed activity, not an official estimate of GDP growth.

Europe offered its own contrasts. The euro-area flash composite PMI slowed to 51.9 from 53.1; the UK composite improved to 53.1 from 52.0, even as the UK’s September consumer-price inflation accelerated and retail-sales volumes declined. That combination helps explain why apparently similar dollar crosses need not behave alike: domestic economic releases and expectations about each central bank differ.

The US September CPI data released on 14 October remained relevant background, with headline inflation at 3.7% annually and core inflation at 2.6%. Those were not fresh releases during 19–23 October. Nor did the Federal Reserve, ECB, Bank of England, or Bank of Japan announce a new policy decision during the completed week.

Federal Reserve, US data, Treasury yields and DXY

US claims and housing were not telling the same story

The US Department of Labor reported 194,000 initial unemployment-insurance claims for the week ended 17 October, down from a revised 200,000. The four-week average fell to 197,750 from a revised 198,750. Insured unemployment, which refers to a different measurement week, stood at about 1.728 million for the week ended 10 October, 25,000 higher than its revised previous reading. These series describe different parts of labour-market conditions and should not be merged into a single hiring measure.

The Census Bureau and HUD reported September housing starts at an annualised 1.221 million, down 5.2% from a revised August 1.288 million. Building permits were 1.400 million, down 0.2% from a revised 1.403 million, while single-family starts fell 4.6% to 859,000. Separately, the Bureau of Labor Statistics reported a national unemployment rate of 4.2% in its September state-employment release; five states recorded monthly unemployment-rate declines. The state report was not a new national payrolls release.

The BLS also reported median usual weekly earnings of $1,232 for full-time wage and salary workers in the third quarter, an increase of 1.5% from a year earlier in nominal terms. These quarterly earnings figures offer useful wage context but do not measure month-on-month changes in earnings or inflation.

Consumers remained worried about prices

The University of Michigan’s final October consumer-sentiment reading was 48.0, above the preliminary October estimate of 46.3 but marginally below September’s 48.1. The university’s original Surveys of Consumers historical series showed one-year inflation expectations rising to 4.7% from September’s 4.6%, and five-year expectations moving to 3.5% from 3.4%. These are survey expectations, not realised inflation. The institute’s dated October statement and the original expectations history are distinct sources.

Yields fell across the curve, but the curve flattened

In the US Treasury’s daily par-yield-curve series, the two-year yield moved from 4.77% on 19 October to 4.69% on 23 October, a fall of eight basis points. The five-year declined from 4.94% to 4.82% (12 basis points), while the ten-year fell from 5.15% to 5.00% and the thirty-year from 5.48% to 5.33% (15 basis points each). These comparisons use the same official series at both endpoints, not a mixture of Treasury par yields and market closing quotes.

Because the ten-year yield declined more than the two-year yield, the ten-year-minus-two-year spread narrowed from 38 to 31 basis points. That was a seven-basis-point flattening of a positively sloped curve, not an inversion. Meanwhile, the provider-dated DXY observation slipped from 101.98 to 101.81, a modest 0.17% decline. Falling Treasury yields and a softer dollar can be interpreted together, but the data do not establish a sole cause for either move.

Forex market reaction: EUR/USD, GBP/USD and USD/JPY

Foreign-exchange performance was uneven. Under consistent ChartExchange historical observations within each pair, EUR/USD edged down 0.10%, GBP/USD rose 0.42%, and USD/JPY fell 1.07% from 19 to 23 October. In levels, EUR/USD moved from 1.1213 to 1.1202, sterling from 1.3307 to 1.3363, and USD/JPY from 158.5220 to 156.8190. The DXY and individual pairs use different provider closing conventions and are not the exact same-timestamp decomposition of the dollar basket.

For sterling, UK economic reports supplied an important separate backdrop. The ONS labour-market overview showed unemployment at 4.7% in June–August, employment at 75.3%, regular pay growth of 3.7% year on year and vacancies of 691,000. Preliminary September payroll employment rose 13,000 from the previous month. At the same time, UK CPI climbed to 3.5% year on year in September from 3.1% in August, with CPIH at 3.6%. The ONS producer-prices report recorded annual input and output inflation of 8.3% and 4.6% respectively.

There was a weaker consumer-spending signal: retail-sales volumes fell 0.4% in September after a revised 0.7% rise in August, although third-quarter sales volume grew 0.7% quarter on quarter. October’s flash UK composite PMI improved, suggesting a different picture for surveyed business output. Across the Channel, the slower euro-area composite PMI suggested more subdued momentum. None of these comparisons alone determines the next ECB or BoE decision.

The USD/JPY decline is consistent with the yen strengthening against the dollar by this specific pair-return measure. It does not, by itself, establish that authorities intervened in foreign-exchange markets. The Bank of Japan’s scheduled 29–30 October policy meeting also lay ahead of the reporting cutoff.

Crypto market reaction

CoinGecko provider-dated observations place Bitcoin at $81,875 on 19 October and $80,049 on 25 October, a 2.23% decline. Ethereum fell from $2,405.92 to $2,326.83, losing 3.29%, while Solana slipped from $104.98 to $101.24, down 3.56%. All three finished the seven-date measurement lower, with Ethereum and Solana posting larger percentage declines than Bitcoin.

Those returns are calculated from matched CoinGecko daily historical observations rather than exchange-native technical candlestick closes. That distinction matters particularly for cryptocurrency, where market trading continues through weekends, and different providers have different daily measurement conventions.

For the separate daily-chart perspective on the five assets, see HFR’s 19 October technical outlook. Its historical candle dates and providers are not identical to this recap’s price-return series.

Activity beyond price included the Solana Foundation’s Alpenglow communication dated 21 October. It described the protocol upgrade’s mainnet activation and a finality design target of roughly 150 milliseconds. A target in an original protocol statement is not proof that every transaction achieved that latency. Financial-infrastructure announcements were also notable: the US Office of the Comptroller of the Currency announced a stablecoin implementation measure on 21 October, followed by Federal Register publication on 23 October with a stated future effective date of 18 January 2027. The latter was publication, not entry into force.

Bitcoin, Ethereum and Solana fund flows

The strongest contrast between token prices and selected investment-product demand emerged in US-listed fund data. Farside Investors’ Bitcoin table recorded a net $301.2 million inflow across the five eligible 19–23 October sessions, whereas its Ethereum table showed a $220.7 million outflow. The separately tracked Solana-linked product universe recorded a $17.6 million net outflow. These are differences in both sign and product coverage, not evidence of a uniform institutional retreat.

The daily pattern was far from one-way. Bitcoin products gained $526.7 million on Monday and $524.8 million on Wednesday, but lost $635.6 million on Thursday and $137.9 million on Friday. Ethereum recorded both inflow and outflow days. The reported Solana-linked aggregate for Friday was 0.0, which should not be confused with imputing zero to missing constituent-product records.

Net product inflows and redemptions describe a reported fund-flow universe. They do not demonstrate a same-day spot-token purchase or sale of identical size, and different fund structures, conversions or seed arrangements complicate direct comparisons. The positive Bitcoin aggregate therefore does not contradict a negative Bitcoin price return; it simply tells a different, narrower part of the week’s story.

Weekend crypto update

Crypto weakness continued beyond the Friday close of the traditional-market reporting window. Between CoinGecko’s Friday 23 October and Sunday 25 October dated observations, Bitcoin fell 1.30%, Ethereum 2.43% and Solana 1.46%. These are Friday-to-Sunday comparisons, not a claim about Saturday-open prices. The seven-day reporting window also includes the intervening Saturday 24 October observation for each asset.

Gold, silver and precious-metals ETF flows

The price picture in metals was unequivocally negative by the chosen spot-proxy convention. Gold XAU/USD moved from $4,132.1100 per troy ounce on 19 October to $3,994.7700 on 23 October, down 3.32%. Silver XAG/USD fell from $56.8822 to $53.8492, a 5.33% decline. Silver therefore underperformed gold by roughly 2.01 percentage points. These are ChartExchange historical spot-proxy quotes, not COMEX futures settlements.

Selected US metal-fund flows offered a more complicated picture. ETFIQ estimated GLD net outflows of about $2,307.8 million; IAU recorded a positive $17.1 million. Together, the two funds showed estimated net outflows of approximately $2,290.7 million. By contrast, summing ETFIQ’s displayed daily SLV estimates gives approximately $38.5 million of net inflow. The gold pair is a selected group of US funds, not a measure of all global gold ETFs; the SLV figure is not a global silver-fund total.

These values are estimates based on the ETFIQ methodology: the change in shares outstanding multiplied by the previous day’s net asset value. They are not issuer-certified settled cash flows, fund-trading turnover, changes in assets under management or proof that an equivalent amount of physical metal was bought or sold. On 22 October, the displayed GLD estimate is rounded to approximately −$1.6 billion; the source’s displayed five-day GLD figure is about −$2.3 billion.

There is also a small but important SLV rounding distinction. ETFIQ’s displayed five-day figure rounds to around +$38.6 million, whereas adding its individually published one-decimal daily values produces approximately +$38.5 million. The article uses the latter calculation transparently rather than implying greater precision than the daily data support.

Gold’s negative price return coincided with a negative GLD estimate, while IAU remained slightly positive. Silver’s decline diverged from positive estimated SLV flows. Neither alignment nor divergence establishes price causation, global precious-metals ETF activity or verified changes in physical bullion holdings.

What to watch next: 26–30 October 2026

The next scheduled policy and macro releases are important because they may test the mixed signals from the completed week. These are future calendar entries as of 26 October, not known policy outcomes or reported data results.

The official Federal Reserve calendar places the FOMC meeting on 27–28 October, with the decision scheduled for Wednesday 28 October. The ECB calendar lists its decision for Thursday 29 October, and the Bank of Japan calendar lists its meeting for 29–30 October. The next Bank of England policy decision is on 5 November, outside this forward week. Other US releases and times come from the original Census, BLS, and BEA calendars. No policy decision or economic print scheduled after the cutoff is treated as already known.

For background on why spreads may change around economic news, read HFR’s guide to why forex spreads widen. The scheduled releases do not imply any particular price direction.

Conclusion

The completed week is best understood through its contrasts. Treasury yields and DXY declined while gold and silver fell; the yield curve flattened without inverting; sterling and the yen strengthened against the dollar by their respective pair measures even as EUR/USD edged lower. In crypto, Bitcoin’s positive fund-flow total stood against a negative price return, while Ethereum and Solana-linked products had net outflows. SLV’s estimated inflow alongside falling silver prices offers another reminder that one market indicator rarely tells the whole story.

For the week ahead, the useful question is how the scheduled central-bank decisions and growth and inflation releases alter that fragmented picture. The week just ended supplies a factual baseline, but it does not predetermine the response to those future events.

Risk note

Financial-market reporting is for information and education, not personalised investment advice or a recommendation to buy, sell or hold an asset. Currency, precious-metal and cryptocurrency prices can change sharply. Exchange-traded fund flows do not eliminate price risk, and financial-market movements around scheduled economic and policy releases may be volatile.

Sources and methodology

Traditional-market prices and yields are 19–23 October 2026 historical observations; crypto price returns use 19 and 25 October CoinGecko provider-dated readings, and weekend returns compare 23 and 25 October. US product flows cover the five reported sessions from 19 to 23 October. Historical source data and publication dates are distinguished throughout; no Monday 26 October market session is included.

• Official US data: Treasury par yield curve · Department of Labor weekly claims · Census/HUD housing starts · BLS state labour markets · BLS usual weekly earnings · BLS September CPI (14 Oct context)
• UK and European economic releases: ONS labour market · ONS CPI · ONS producer prices · ONS retail sales · US flash PMI · Euro-area flash PMI · UK flash PMI
• US consumer survey: University of Michigan final sentiment · Original inflation-expectations table
• Dated price histories: DXY · EUR/USD · GBP/USD · USD/JPY · XAU/USD · XAG/USD · Bitcoin · Ethereum · Solana
• Investment-product data: Farside Bitcoin · Farside Ethereum · Farside Solana linked · ETFIQ GLD · ETFIQ IAU · ETFIQ SLV · ETFIQ flow formula
• Primary announcements: Solana Foundation · OCC · Federal Register
• Forward institutional calendars: Federal Reserve · ECB · Bank of Japan · Census · BLS · BEA

ETFIQ daily flow figures are estimates; sums derived from rounded daily observations are approximate, particularly GLD and SLV. GLD plus IAU does not represent worldwide gold-fund flows, and SLV does not represent worldwide silver-fund flows. The research did not establish a matching weekly global-fund universe or dated physical-holdings series, so neither is estimated here. Fund creations and redemptions should not be equated with immediate underlying transactions. DXY, foreign-exchange pairs and spot metals use their stated provider daily conventions rather than a single synchronised global close.