The Cockpit · August 2026

Edito

Once again, the pessimists made sense and the optimists made money

Psychology has turned up a pattern worth knowing. Pessimists look smarter. In 1983 Teresa Amabile handed people real book reviews and asked them to rate the reviewer, and the harsh reviewers came back rated more intelligent and more expert. Here is the part that stings. Separate judges had already scored the kind reviews as the better writing. Same quality. Nastier tone. Cleverer author, apparently.

Finance has its own version, and it is measured in prices rather than opinions. Paul Tetlock scored the daily mood of the Wall Street Journal’s market column and matched it against the market. Journal of Finance, 2007. Gloomy columns pushed prices down. Then the prices came back. His word for it was noise.

Fast forward to August 2026. It was a very good month for investors. Every risk asset I track went up. Technology gained 6.4 per cent in the month. The S&P 500 gained 2.7. All seven model portfolios rose, in all three currencies. The equity ones led by a distance: up 14.3 per cent this year in dollars, 13.0 in euros, 12.6 in francs.

Bonds are where it gets interesting.

Financial news at the end of August 2026 was dominated by worry about high interest rates. The thirty-year Treasury yield closed the month at 5.25 per cent. It has not been that high since July 2007. What followed 2007 was the crash of 2008, and a financial crisis that spread across the world in 2009.

Now look at what the bond funds actually did.

2022 is the crash people remember. The thirty-year yield went from 1.9 to 4.4 in ten months. The long Treasury fund fell 36.6 per cent from its peak.

This year the same yield moved from 4.84 to 5.25, the same fund is down 2.8 per cent, and six of the seven bond lines I track finished August higher.

A high yield and a rising yield are two different things. 2022 was the climb. This is the plateau. A bond bought at 5 per cent and left at 5 per cent pays 5 per cent.

None of that makes the pessimists fools. They are right that tails exist. They are wrong about how often. Take September, historically the worst month of the year for equity markets. Since 1928 it averages a 1.2 per cent loss, the only month with a negative long-run average. Its median is a gain of 0.1. A handful of disasters carry the average. Most Septembers are dull.

This is why I spend the time building these Monthly Cockpit Reports. Emotion is expensive, and a monthly set of measured facts is the counterweight. The pull towards bad news is not a failing of the press either. It sits in the reader. Soroka, Fournier and Nir wired people up in seventeen countries and found the body reacts harder to negative stories than to positive ones. Gloom travels because we carry it further.

So the worries are real, and I will not talk you out of them. Yields are higher. Debt is higher. The AI build-out leans on borrowed money, and in August Nvidia structured a $500bn compute platform with six financial institutions. It might look like circular financing on steroids, and the shares fell about 3 per cent on the news. Earnings carry paper gains that never touched cash, $98bn at Alphabet and $53.4bn at Amazon.

The other half is just as real. Remove those two companies and the seven largest technology firms still grew earnings 43.2 per cent. The other 493 grew 31.8 per cent, their best since late 2021. Across the index 86 per cent beat estimates, against 78 per cent over five years. The Atlanta Fed’s nowcast had third-quarter growth at 4.8 per cent on 1 September, with the caveats this issue sets out. An outstanding quarter and a fast-running economy do not stop being facts because the headlines are about rates.

My aim is a picture of the financial world as close to reality as I can get it. Nothing here is a recommendation. It is an examination. I am very happy to exchange views, so do not hesitate to reach out.

The Cockpit · August 2026

The Seven Model Portfolios

Seven classic portfolios, from Fixed Income through to Equity, read as references rather than recommendations. Each is passive, total-return and rebalanced quarterly, indexed to 100 with a simulated-expense slider. The graph, the performance and risk statistics, the contribution by holding and the full compositions sit below, frozen as of 31 August 2026. They are a risk dial to set a real allocation against, not products.

The Cockpit · August 2026

Equity Watchlist

One thousand companies, the S&P 500 plus the 500 others we consider the most important in the world, priced to the close of 31 August 2026. Sort any column to see where the month actually moved, filter by sector, region, currency or size, and open a name to read its ten-year history. Three companies that belong here are absent, and the reason is worth stating rather than hiding. International Holding and Phoenix Group cannot be priced from either of our data feeds. ChangXin Memory listed in Shanghai on 27 July and would enter near the thirtieth largest, but it has a ticker and no usable prices yet. An instrument that reports its own blind spot is more use than one that leaves the gauge blank.

The Cockpit · August 2026

Top News

The August developments that mattered most for markets, read to the close of 31 August 2026. Every figure was checked against a filing, an official publication or our own series. Where a claim rests on press reporting, the outlet is named.

Brent ended August at $90.49, a fifth of a per cent above where it started, after trading between $79.36 and $94.39 inside the month. A 19 per cent range and no net move. WTI closed at $85.76. The Strait of Hormuz stayed shut throughout. The year is where oil shows up: Brent is up 48.7 per cent since December and WTI 49.4 per cent.^[1] Copper closed at $6.59 a pound, up 2.5 per cent on the month and 17.1 per cent for the year.^[1] Agriculture ran hardest of the three. The Bloomberg Agriculture Spot Index gained more than 13 per cent, its best month since July 2012, after attacks on Black Sea ports and summer heat across the United States and Europe.^[3] Wheat reached a three-year high. Core CPI cooled to 2.5 per cent year on year, its lowest since March 2021.^[4] Core PCED did not move: 3.3 per cent.^[5]

The US Treasury spent August trying to hold down its own long end, and did not manage it. On 19 August it doubled its buybacks of ten- to thirty-year debt to $4bn an operation.^[6] The thirty-year yield fell nine basis points on the news. It was back above 5.27 per cent before the month ended.^[7] Kevin Warsh gave his first Jackson Hole address as chairman on 28 August and declined to offer guidance. He described the Federal Reserve as committed to a discipline rather than to a decision.^[8] July's meeting had already produced three dissents, all of them in favour of a rise. The next decision comes on 16 September, with new projections. Prediction markets have moved a long way since the speech. On Kalshi, the odds of a 25 basis point rise were near 30 per cent before Jackson Hole. On 2 September, two days after this issue's cut-off, they stood at 58 per cent, against 42 per cent for no change and 2 per cent for a larger move, on $31m of volume.^[9] The iShares 20+ Year Treasury Bond ETF (TLT) finished the month up 0.7 per cent and is down 2.8 per cent for the year.^[2]

Long-dated government borrowing costs rose across the developed world in August. Japan's ten-year yield touched 3.00 per cent on 31 August, its first visit to that level since 1996.^[10] The thirty-year set a record for the tenor. Scott Bessent told Finance Minister Satsuki Katayama and Governor Kazuo Ueda that Japan should raise rates next, which Bloomberg reported on 31 August.^[11] In public, the same day, he said he would not tell them what to do. Europe moved the same way. The French ten-year stood at 4.08 per cent on 21 August against 3.24 per cent for the German Bund, a gap of 84 basis points.^[12] Euro-area inflation ran at 3.3 per cent in August, the fastest since September 2023.^[13] French debt is above 116 per cent of output.

The Atlanta Fed's nowcast has third-quarter growth running near 5 per cent, and the quarter before it was measured at 1.5. GDPNow opened the quarter at 5.0 per cent on 30 July, reached 6.2 per cent on 3 August, and stood at 4.8 per cent on 1 September.^[14] The Bureau of Economic Analysis put actual second-quarter growth at 1.5 per cent annualised.^[15] The model therefore describes an economy running roughly three times as fast as the one just measured. It is worth being precise about what that number is. GDPNow is a nowcast assembled from the data released so far, not a forecast, and the Atlanta Fed says plainly that early-quarter readings rest on limited data and move a great deal. Its average absolute error is 0.77 percentage points on the final estimate of a quarter, with a root-mean-square error of 1.17, and a reading taken five weeks in carries more than that. On 31 August the President said growth could reach 20 per cent, a rate the United States has hit once since the war.^[16] Set the nowcast beside the labour market and the two instruments point opposite ways: hiring has averaged 20,000 a month over three months, and July's payrolls fell outright. Which of the two is describing the economy is the question the September meeting turns on.

Payrolls fell in July, and almost nothing in the rates debate acknowledged it. The report landed on 7 August. Employers cut 23,000 jobs against a consensus of 85,000 added. Government shed 53,000 while private payrolls grew 30,000. Unemployment edged down to 4.1 per cent, but for an unwelcome reason: the labour force shrank by 264,000. People left rather than found work. Average hourly earnings rose 3.2 per cent year on year, the slowest since May 2021. Job growth has averaged 20,000 a month over the last three months, against 60,900 for the year to date.^[17] The Federal Reserve meets on 15 and 16 September with core inflation stuck at 3.3 per cent and hiring close to a standstill. The market has moved towards tightening regardless. On CME FedWatch the odds of a September rise fell on the 7 August payroll print, still favoured no change on 25 August, and reached 66 per cent by 31 August.^[18] Payrolls fell, and the pricing went the other way.

August belonged to technology, and the year depends entirely on which technology you mean. The technology sector (XLK) returned 6.4 per cent in August and is up 29.9 per cent for the year. Invesco QQQ gained 4.2 per cent and 17.0 per cent. The Vanguard Value ETF returned 2.2 per cent in August and 19.0 per cent for the year. So value beat the Nasdaq 100 over the year and trailed the technology sector by eleven points, and Vanguard Growth is up 8.8 per cent, less than half of value. Ask whether value or growth is winning and each instrument gives a different answer. The S&P 500 broke out of the range it had held since 14 May, set successive records mid-month, then gave part of that back. It closed at 7686.14 on 31 August.^[19] Korea and Taiwan returned to the front of the global rankings, carried by semiconductors. Europe and Japan lagged, and a firmer dollar reduced what was left for a dollar-based holder.

The quarter was excellent, and two companies made it look more excellent than it was. FactSet's update of 28 August, once every large company had reported, put earnings growth at the seven largest technology companies at 118.5 per cent, their best since at least the end of 2020. Alphabet's figure carried a $98bn gain in other income, mostly unrealised gains on shareholdings. Amazon's carried $53.4bn, largely from its investment in Anthropic. Both report on a GAAP basis, so those gains sit inside earnings per share. Remove the two and the seven still grew 43.2 per cent. The other 493 companies grew 31.8 per cent, their own best since late 2021. Across the index 86 per cent beat estimates, against a five-year average of 78 per cent.^[20] The headline flatters. The quarter underneath it is strong on its own.

Nvidia made two announcements seventeen days apart and the market treated them as opposites. On 10 August it unveiled a $500bn financing platform with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, structuring compute the way toll roads are structured. The shares fell about 3 per cent, read as circular financing. On 26 August it reported revenue of $96.2bn against $92.2bn expected, guided the current quarter to about $108bn, and Jensen Huang forecast roughly 70 per cent revenue growth for fiscal 2028, well above what analysts were carrying. The shares rose sharply the following session and finished the month up 10.0 per cent.^[21] The market paid for a longer disclosure horizon and charged for financial engineering, in the same company, in the same month. Tim Cook left the Apple chief executive's office on 31 August after fifteen years. John Ternus succeeded him the next day.^[22]

Moderna rose 176.97 per cent on 19 August, six times its previous best day, and fell about 20 per cent the next. It and Merck had reported that their personalised mRNA cancer vaccine succeeded in a phase 3 melanoma trial.^[23] Intraday reports that day carried gains of 85, 125 and 151 per cent, all filed before the close. Software staged the month's other reversal. The S&P 500 software and services subsector had fallen more than a third from its October 2025 peak, on the argument that AI agents would destroy per-seat pricing. In August it rose 13.5 per cent, taking the quarter to 28.4 per cent. Salesforce raised guidance and had its second-best day on record.^[24] The same companies, the same business models, and two opposite verdicts inside a single year.

The iShares Silver Trust (SLV) returned 14.8 per cent in August and SPDR Gold Shares (GLD) 9.9 per cent, and both are worth less than they were in January. Spot gold ended at $4,452 an ounce, well below the $5,586 record set early in the year. Across 2026 GLD has returned 3.1 per cent and SLV has lost 6.7 per cent. A reader who knows only that metals had a strong August has the year backwards. Central banks bought a record 288.9 tonnes in the second quarter. China added in July for a twenty-first consecutive month. Poland took 82 tonnes in the first half. Forty-five per cent of central banks surveyed expect to add more over the coming year, also a record.^[25] The official buying is steady. The price is not.

Bitcoin gained 25.1 per cent in August and is still down 10.2 per cent for the year. It ended at $78,549 against $62,814 on 31 July, its largest monthly rise of 2026. The rally undoes part of a long fall. Bitcoin peaked at $96,929 on 14 January and closed August 19 per cent below that.^[26] The rest of the market ran harder. Solana added about 46 per cent, its best month in ten, while Ethereum and XRP each gained more than 30 per cent. The Securities and Exchange Commission proposed two exemptions for token offerings on 19 August, the day the White House met industry executives.^[27] Bitcoin exchange-traded funds have still lost more than $7bn this year.^[28] Separately, the Treasury is counting on stablecoin issuers to absorb its bill supply, which ties the asset class to how the government funds itself. That link showed up in a share price. Circle, which issues USDC, returned 52.6 per cent in August, the third-best performance in the whole Socrates 1,000.^[29] A stablecoin issuer earns the yield on the reserves backing its coins, so a larger bill supply at a higher policy rate widens the spread it collects. Buying Circle is, in part, a position on how much the government needs to borrow.

Chinese listings raised more than $54bn in 2026 by the end of August, passing the whole of last year, and the year's defining listing came in July. ChangXin Memory, China's largest maker of DRAM, priced its Shanghai STAR offering at 8.66 yuan and raised 57.9bn yuan, about $8.6bn. That is the largest listing on the mainland since Agricultural Bank of China in 2010 and the biggest in Asia this year. It closed its first day, 27 July, at 49.00 yuan, up 466 per cent on the offer price, and ended August at 58.01.^[30] It held 7.67 per cent of the global DRAM market at the end of 2025, behind Samsung, SK hynix and Micron, and has gained pricing power as AI demand has tightened memory supply. A country that could not buy advanced memory now lists its own maker at a scale that puts it among the largest companies in China. Unitree Robotics, the first listed maker of humanoid robots on the mainland, debuted in Shanghai on 19 August. Shein priced its Hong Kong offering on 31 August but did not begin trading until 1 September, so it belongs to the September issue.^[31] Money supply grew 7.7 per cent in the year to July, with M1 at 4 per cent. The central bank widened cross-border renminbi and foreign-exchange operations for multinationals.^[32] The policy tone stays on stability and supply, not on stimulating demand at home.

September has the worst record of any month, and the record is thinner than the reputation. Since 1928 the S&P 500 has averaged about -1.2 per cent in September, the only month with a negative long-run average, and it has fallen in 56 per cent of years. Nine of the forty worst months on record are Septembers. Then look at the middle rather than the mean. The median September is about +0.1 per cent, and the index has risen in 52 per cent of them since 1925. The average is carried by a few disasters: September 1931 at -29.6 per cent, September 2022 at -9.4 per cent, September 2008 at -8.9 per cent.^[33] Take the handful of extremes out and the month is ordinary. An average built from a fat left tail describes the tail, not the year you are about to live through. The election calendar is the firmer pattern, and the midterms fall on 3 November 2026. Since 1945 midterm years have averaged 3.8 per cent against 10.9 per cent for the other three years of the presidential cycle, with the first three quarters averaging -0.9 per cent and the fourth quarter 6.4 per cent. The twelve months after a midterm have averaged roughly 16 per cent.^[34] Both patterns rest on about twenty observations, which is a thin base for a decision, and neither tells you what this September will do.

The Cockpit · August 2026

Methodology & sources

This report is a set of instruments, not a set of instructions. What follows is how each figure is built, where the data comes from, and how it is checked, so that any reader can trace a number to its origin and judge it independently.

How the figures are built

The seven model portfolios and the asset-class series are total-return, indexed to 100 at the start of each period, and expressed in US dollars. Where a simulated expense is shown, it is applied as a daily drag on net asset value, and rebalancing follows the frequency selected in each widget. Yields, spreads and inflation are plotted as reported, without adjustment.

Sources

The data in this report comes almost entirely from two public sources: Yahoo Finance for market prices, and FRED, the Federal Reserve Bank of St. Louis, for the macro series. Fund fact sheets are the reference used for cross-checking. This is public, end-of-day data, not an institutional terminal feed.

1. Market prices and total returns: Yahoo Finance and EODHD, using dividend-adjusted closing prices for each ETF and index. The Swiss-listed sleeve, the Italian MIB fund and the two euro-hedged commodity ETCs are taken from EODHD, which prices them where Yahoo does not do so reliably.
2. US Treasury nominal and real (TIPS) constant-maturity yields, and the effective federal funds rate: FRED (Federal Reserve Bank of St. Louis).
3. Moody's seasoned Aaa and Baa corporate bond yields, and ICE BofA investment-grade and high-yield option-adjusted spreads: FRED.
4. US consumer price index (CPI-U): US Bureau of Labor Statistics, via FRED.
5. Portfolio weights and fund characteristics: the respective ETF issuers' published fact sheets, used for cross-checking.
6. LME copper cash settlement: the daily table published by Westmetall. The copper card is the LME cash price, not the COMEX future.

News and events

Sources for the Top News items, numbered to match the references in that tab.

1. Brent, WTI and copper closes, monthly ranges and returns: our own key_levels.csv at 31 August 2026.
2. ETF total returns (XLK, QQQ, IVV, VTV, VUG, GLD, SLV, TLT): our own etf_adjusted_close.csv at 31 August 2026, dividend-adjusted closes.
3. Bloomberg Agriculture Spot Index, August 2026.
4. US consumer price index for July 2026: US Bureau of Labor Statistics, released 12 August 2026.
5. Personal consumption expenditures price index for July 2026: Bureau of Economic Analysis, released 28 August 2026.
6. US Treasury buyback operation announcement, 19 August 2026.
7. US Treasury constant-maturity yields: Federal Reserve H.15.
8. Kevin Warsh, address at the Jackson Hole economic symposium, 28 August 2026.
9. Kalshi, “Fed decision in September?” market, read 2 September 2026. This reading is two days after the issue's 31 August cut-off and is dated as such in the text.
10. Japanese government bond yields: Japanese Ministry of Finance, 31 August 2026.
11. Bloomberg, “Bessent Tells Japan Officials Rate Hikes Are Needed”, 31 August 2026, sourced to NHK.
12. French and German ten-year government bond yields and the OAT–Bund spread, 21 August 2026.
13. Euro-area inflation flash estimate for August 2026: Eurostat.
14. Atlanta Fed GDPNow, readings of 30 July, 3 August and 1 September 2026, and the Atlanta Fed's own accuracy note (average absolute error 0.77 points, RMSE 1.17, on the final estimate of a quarter).
15. US real GDP, second quarter 2026: Bureau of Economic Analysis.
16. CNBC, “Trump says U.S. growth could hit 20%”, 31 August 2026.
17. Employment situation for July 2026: US Bureau of Labor Statistics, released 7 August 2026.
18. Implied probabilities for the September 2026 FOMC meeting: CME FedWatch, readings dated 7, 25 and 31 August 2026.
19. S&P 500 index close at 31 August 2026: our own key_levels.csv.
20. John Butters, FactSet, “‘Mag 7’ Companies Reported Earnings Growth Above 100% Boosted by Investment Gains”, 28 August 2026.
21. NVIDIA financing platform announcement, 10 August 2026; second-quarter results and third-quarter guidance, 26 August 2026.
22. Apple, chief executive transition effective 1 September 2026.
23. Moderna and Merck, phase 3 melanoma trial announcement, 19 August 2026.
24. S&P 500 software and services subsector, August 2026, and Salesforce fiscal second-quarter results.
25. Central bank gold purchases for the second quarter of 2026 and the 2026 central bank gold survey: World Gold Council.
26. Bitcoin closes, August and year-to-date returns, and the 14 January 2026 peak: our own key_levels.csv at 31 August 2026.
27. Securities and Exchange Commission, proposed token-offering exemptions and disclosure regime, 19 August 2026.
28. August returns for Solana, Ethereum and XRP, and bitcoin exchange-traded fund flows: market reporting. We hold no series for these.
29. Circle's August return and its rank in the Socrates 1,000: our own watchlist.json at 31 August 2026.
30. ChangXin Memory: Shanghai STAR offer price and proceeds, first-day close of 27 July 2026, close at 31 August 2026, and global DRAM share at the end of 2025.
31. Hong Kong and Shanghai listing proceeds to 31 August 2026, the Unitree Robotics debut of 19 August 2026, and the Shein Hong Kong pricing of 31 August 2026.
32. Money and credit data for July 2026: People's Bank of China, released 17 August 2026.
33. S&P 500 September seasonality since 1928: mean, median, frequency of decline, and the individual months of September 1931, 2008 and 2022.
34. Midterm-year statistics for the S&P 500 since 1945, quarterly split and the twelve months following, and the United States midterm election date of 3 November 2026.

How it is checked

This report is produced with the help of an AI assistant working under the supervision of Jonathan Castella, CFA, who authors it and is responsible for what it says. Before publication, the underlying data passes hygiene checks, and individual figures are spot-checked against issuers' fact sheets and the primary series. These steps reduce error, but they do not eliminate it: mistakes remain possible.

Corrections and liability

This report is educational. It examines frameworks, regimes and history; it is not investment advice, and it recommends no security, fund or allocation. Nothing here should be relied upon as the basis for a financial decision, and the author accepts no liability for any decision taken, or any consequence arising, from its use.

Accuracy matters to the publication. If you find an error, or have a correction, a question or a comment, it is welcome: write to socrates@socrates-on-investing.com.