The night before, four negative shocks came in, yet U.S. stocks didn’t fall much.
Last night’s market action is worth taking a closer look at.
Four things happened at the same time: rumors about refined copper tariffs started to loosen, and copper prices plunged; Brent crude broke through $109 per barrel, and the Middle East conflict stirred up the Strait of Hormuz shipping routes; the European Central Bank raised rates by 25 basis points, deposit rates rose to 2.5%, and it also expects euro area inflation to reach 3% in 2026; and the U.S. August PPI year-on-year came in at 5.4%, well above July’s 4.8%—with energy costs as the main driver.
In the past, any one of these four negative factors could have sent the market straight into a free fall. But last night, all four hit at once, and the performance of U.S. stocks was: the Dow fell 0.60%, the S&P 500 fell 0.58%, and the Nasdaq fell 0.65%. Tech stocks also didn’t drop much.
The signal this market setup sends is very clear—bad news is being priced in early. The market may be thinking: I know inflation is coming, but this is still not enough to prove that the AI earnings cycle is over.
This is the most important background for tonight’s CPI.
First, separate them: year-on-year and month-on-month are looking at two different things.
For the 20:30 data tonight, first look at market consensus.
The US August headline CPI is expected to rise 0.4% month-on-month, up sharply from July’s 0.1%; the year-on-year growth rate is expected to stay at 3.4%. Core CPI (excluding food and energy) is expected to rise 0.2% month-on-month, with the year-on-year rate easing from 2.5% to 2.4%.
There’s virtually no suspense about the headline CPI month-on-month jump—August energy prices surged. Brent crude once neared $110, and WTI broke above $100. The energy subcomponent contributed most of the increase. JPMorgan expects August energy prices to rise sharply 2.5% month-on-month, with a 4.2% increase in transportation fuels being the main driver.
But what really needs to be unpacked are the different implications behind year-on-year and month-on-month.
For the year-on-year comparison, look at how much prices have risen over the past year. For month-on-month, look at the change over the most recent month. Core CPI year-on-year fell from 2.5% to 2.4%, which looks like “inflation cooling.” But that decline could be mainly due to last year’s base effect, and doesn’t necessarily mean that the pricing pressure from the past month has truly eased.
So tonight’s focus is not the headline CPI, but the core CPI month-on-month reading. Market consensus is 0.2%, but Natixis’ chief US economist Christopher Hodge said the more precise forecast for core CPI is 0.19%. Given the current clear divergence among Federal Reserve policymakers on the rate path, a reading below 0.20% may be enough to avoid a September rate hike.
0.19% versus 0.20%—a difference of just 0.01 percentage point. But in the current environment, that 0.01% could be what determines whether the Fed hikes rates next week.
Break it down by subcomponents: is the pressure driven by energy, or is it widespread?
When looking at CPI subcomponents, core needs to answer one question: is the current price pressure truly just localized volatility at the energy end, or has it already spread into broader economic areas?
Housing. Housing accounts for roughly one-third of the CPI weight, and is the biggest support item for core inflation. In July, the housing CPI index rose 0.1% month-on-month. Rent and equivalent rent for homeowners increased by 0.3%. In the past few months, the stickiness in housing inflation mainly came from delayed release of rent data that reflects earlier upward pressure on home prices. Whether this trend continues is the key to whether tonight’s core CPI can come in below expectations.
Core services. Core services prices excluding housing are the Fed’s most watched “super-core” indicator. If this subcomponent’s month-on-month growth accelerates, it means potential second-round transmission from energy shocks through transportation, production, and service costs—more worrying than energy volatility itself.
Energy. The big jump in August energy prices is already a given fact. The question is the transmission range. If only the energy subcomponent pulls up the headline CPI, while core CPI stays moderate, the market’s reaction may be relatively subdued—because the volatility of energy prices naturally tends to be higher than core inflation, and the Fed usually “sees through” such shocks. But if higher energy prices have already seeped into core services areas such as transportation, airfare, and logistics, the situation is completely different.
How will the market react? Don’t treat the script as a foregone conclusion.
Based on current data and market pricing, you can think through a few scenarios in advance, but don’t treat them as inevitable scripts.
Scenario 1: Both are on the hot side, with core also hot. If core CPI month-on-month exceeds 0.3%, the market may significantly increase pricing for a September rate hike. The US dollar and Treasury yields would strengthen, weighing on US stocks and gold. CME data shows the market-implied probability of a 25-basis-point September hike has risen to 71.3%. If core CPI comes in above expectations, this probability could approach over 90%. But note: if after the hike is delivered the dot plot is not further revised upward, you could still see a rebound driven by a “bad news already priced in” dynamic.
Scenario 2: The headline is hot but the core is cool. This is the most delicate situation. Headline CPI rises due to the big jump in energy, but core CPI month-on-month is below 0.2%, suggesting the energy shock hasn’t spread. The market may first be startled by the overall number, then realize that core inflation is still manageable. Stock index futures might drop first and then stabilize. In this case, the subcomponent data matters more than the headline figure.
Scenario 3: Both are below expectations. Core CPI month-on-month is below 0.19%, easing rate pressure. US stocks and gold get support. But gold is also influenced by multiple factors—real yields, the dollar, and safe-haven demand—so it doesn’t necessarily rise just because CPI falls. Currently, gold is trading around $4,400. The core contradiction is that four pricing chains are changing at the same time: energy prices, inflation expectations, the rate-path, and the dollar exchange rate.
Scenario 4: The data basically meets expectations. This is the most likely case. The market has already priced in a fair amount of rate-hike expectations in advance. The 10-year US Treasury yield even touched 4.969% at one point, while the 30-year rose to 5.368%. If the data lands near consensus, the key is how much the market has already priced in—if pricing is sufficient, there may be little drama; if there’s still disagreement, volatility is more likely to show up after the market opens rather than at the instant the data is released.
Gold’s pricing logic is more complicated than “it goes up when CPI falls.”
For gold, the impact of CPI is not linear.
Gold is facing a typical “dual-attribute conflict.” Safe-haven demand increases willingness to hold, but inflation risk could also lift expectations for nominal and real interest rates. The latter usually suppresses the relative valuation of non-yielding assets.
If price changes are mainly due to a weakening dollar, the correlation between gold and the exchange rate will increase. If they’re mainly due to an energy shock, the market often re-evaluates inflation, bond yields, and policy rates at the same time, and gold’s safe-haven attribute will be offset by rising opportunity costs.
In other words, gold trading isn’t about the inflation level itself—it’s about the spread between inflation and “the policy response.” If higher inflation simultaneously brings stronger tightening expectations, real rates could rise, and safe-haven demand may not fully offset changes in the cost of holding.
A good news item that’s easy to overlook.
Beyond the four negative factors from the night before, there’s another thing worth paying attention to.
On September 10, local time, the US Federal Communications Commission (FCC) issued its final rules. In the 13-page rule text, no Chinese optical module companies—including New Easun, InnoLight (JingJie?), Tongyu Precision, and Tianfu Communications—were mentioned.
A Reuters report on the ban of optical modules was only at the information level; it was not reflected in the rules. Galaxy Communications pointed out that the rules actually being implemented are more about precise restrictions. The risks to optical communications under the earlier policies are expected to gradually decline.
This little essay has been debunked. For A-share optical communications, it’s a positive development that has been covered up by market sentiment.
For people doing dollar-cost averaging (DCA), how should they look at tonight?
Back to the most practical question.
If you’re doing DCA, tonight’s data is worth watching, but there’s no need to overturn your long-term plan every time new data is released. CPI is monthly data; inflation trends are variables at the quarterly or even annual level. A single month’s data rarely changes the long-term direction.
If you’re going to trade short term, you should decide your position size and exit conditions before the release, not wait until the price moves and then scramble for reasons. At 20:30, first look at the initial reaction of stock index futures, US Treasuries, the dollar, and gold. At 21:30, after the US stock market opens, see whether that move can be sustained. The first spike up or plunge down doesn’t necessarily determine tonight’s final direction.
A-shares have already shrunk volume for a week—this is the moment for a potential turning point. To anticipate the impact of this rate hike, the large A-share market has already fallen for a month in advance. If CPI doesn’t come in far above expectations, the suppressing factors could turn into support.
First look clearly at the data tonight, then see how the market interprets it. Don’t snatch a few seconds fewer trades—maybe you won’t earn fewer profits. But if you get it wrong and rush to recover, that’s when one data night can easily turn into a long war.