July’s Market Extremes
The S&P 500 barely moved, but beneath the index, July produced historic moves across individual stocks, semiconductors, and overseas markets.
July was a month of extremes.
You would not know it from looking at the S&P 500. The index declined just 0.13%, with a maximum closing drawdown of only 3.42%.
Beneath that relatively calm surface, however, individual stocks and entire market segments experienced some of their largest moves in decades—and, in certain cases, ever.
How extreme was July? Let’s take a look.
| July extreme | Historical context |
|---|---|
| Intel fell 35.40% during July. | Its second-worst calendar month in the available history since 1980, exceeded only by September 2000. |
| Microsoft rose 15.51% on July 30. | Its best day since October 2008. The move added approximately $450 billion in market capitalization—the largest one-day gain by a company at the time. |
| Amazon rose 15.32% on July 31, while Apple fell 7.35% on the same day. | Amazon recorded its largest one-day gain since April 2012. |
| IBM fell 25.21% on July 14. | The worst one-day decline in its available history, exceeding even its loss during October 1987. |
| South Korea’s KOSPI fell 10.84% on July 28 and another 5.98% on July 29, before rising 17.91% on July 31. | The index still lost 22.19% during July and experienced a maximum closing drawdown of approximately 34%. It was the third-worst calendar month in the available history since 1996, behind October 1997 and October 2008. |
| The PHLX Semiconductor Sector Index lost 20.61% during July. | Its worst calendar month since October 2008. |
Returns are measured close to close. Historical rankings refer to the available daily price histories described in each row.
These were not variations of a single event. They were evidence of a market whose averages concealed what investors actually experienced.
As enormous winners and losers offset one another, the index muted the violence occurring beneath it. The market appeared relatively stable in aggregate while becoming extraordinarily unstable at the level of individual securities, sectors, and portfolios.
When the Hedge Was the Same Bet Twice
It is no coincidence that a month like this also forced Leopold Aschenbrenner’s Situational Awareness fund into a crisis-mode unwind of its public-equities book.
As I wrote in Leopold Aschenbrenner’s Hedge Was the Same Bet Twice, the portfolio appeared diversified by ticker and direction, but its AI-infrastructure longs and software shorts depended on the same underlying relative-value thesis. When that relationship reversed, both sides lost together, and leverage turned a market reversal into a financing and liquidity crisis.
This is the difference between a hedge that has the opposite sign and a hedge that has a genuinely different failure mode.
A long position and a short position may point in opposite directions on a statement. Economically, however, they can still depend on the same relationship continuing to hold. If that relationship breaks, the apparent diversification can disappear precisely when it is needed most.
July did not require the entire market to crash to create a portfolio-level crisis. The S&P 500 could remain nearly flat while the exposures inside a concentrated portfolio moved violently against one another.
Risk lived in the portfolio’s coordinates—not in “the market” as an abstraction.
Where the Self-Improvement Inner Loop Thrives
Months like July are also where Runtime’s self-improvement inner loop thrives.
Our platforms operate through a continuous process of observation, measurement, testing, and improvement. An extreme month produces something ordinary markets cannot: a dense collection of edge cases—broken correlations, violent reversals, unusual dispersion, and hedges behaving very differently from expectations.
Those observations become proprietary evidence. They are logged, tested against our existing assumptions, and incorporated into the process only after validation.
That is why a month like July makes Runtime’s platforms stronger.
Not because one extraordinary month proves a system, but because the self-improvement inner loop now carries a broader record of how markets behave when familiar relationships break.
Experience becomes data.
Data becomes training data corpus.
And validated learning becomes a stronger platform.
Engineered compounding,
Deniz Erkan
