Tonight at 20:30, the U.S. will release its August PCE data.
You’ve probably already seen a reminder: this time, the methodology has changed.
I’m going to break this sentence down, because it determines how you’ll look at that number tonight.
1. Two things happen tonight at the same time.
The first thing is temperature—that is, how much U.S. prices actually rose in August.
The market’s expectation for core PCE month-over-month is 0.3%, and the actual figure for July is 0.2%.[9]
The second thing is the ruler—that tool statistical agencies use to measure prices.
Tonight it will switch to a different set of algorithms.
The U.S. Bureau of Economic Analysis will kick off an annual revision with the August data.[2]
The re-calculation covers Q1 2021 through Q1 2026.
All the PCE historical data for these years have to be re-calculated.
In the PCE price index, three service subcomponents changed methodology.[1]
They are investment advisory and investment management services, legal services, and computer software and accessories.
II. Which ruler is being switched
First, among the three items, the one that matters most is the most consequential.
Investment advisory and investment management services rose 21.6% year over year over the past 12 months.[6]
It is the second-largest contribution item in core PCE.
The old method simply uses the producer-side price index to deflate it.
That is, treat the increase in management fees as the increase in service prices.
The problem is that asset-management fees are charged based on the size of assets.
As assets rise, costs naturally rise—so this item keeps looking like it’s surging.
With the new method, they changed their thinking; it doesn’t look at fees.
Instead, it uses total industry hours to estimate the real quantity of services.
Then, divide nominal spending by the quantity of services to back out the price.
Growth in hours lags far behind growth in the size of assets.
So for the same category, the measured increase in quantity will narrow sharply.
Using the new methodology to re-calculate, this item’s year-over-year figure is left at only 9.0%.[6]
Same animal—switch the ruler, and it changes from 21.6% to 9.0%.
That’s the literal scene of the tiger turning into a cat tonight.
The other two are computer software and accessories, and legal services.
Their revision directions are different; among them, legal services were revised upward.[1]
III. Taken together, core PCE will be dragged down by lower quantity
Goldman’s estimate is that year-over-year core PCE for May was revised down from 3.4% to 3.2%.[6]
Down by 0.2 percentage points, while JPM gives 3.3%.
So Goldman lowered its December core PCE forecast from 3.2% to 3.0%.[6]
Even today, the official magnitude of the downward revision has not been disclosed.
So these numbers are investment-bank estimates, not official results.
What you should truly be wary of is a fact pointed out by UBS.
Among the three categories that were revised, two happen to be in the top four contribution items of core PCE.[6]
Investment advisory and investment management services rank second, and computer software and accessories rank fourth.
And for a sequence that also has measurement issues, but whose contribution to inflation is neutral and negative, none made it into the list.[6]
They include spectator sports, household operations, photo finishing, and computer prices.
IV. Three hard pieces of evidence that the tiger is still here
Changing the ruler only changes the historical scale; it can’t change these three things below.
First, the quantity hasn’t moved—everything that’s rising is price.
In July, official nominal PCE rose 0.2% month over month, while real PCE rose only 0.0%.[3]
In other words, the physical quantity of consumption hasn’t increased at all.
So that entire 0.2% comes from prices.
Second, the money is flowing to the part that is hardest to bring down.
In July, service consumption increased by $86.2 billion, while goods consumption decreased by $49.9 billion in the same period.[3]
Service prices are the stickiest component of inflation.
Shifting the spending structure toward services is like adding support to inflation.
Third, month-over-month momentum is accelerating, not slowing.
The forecast for August core PCE month over month is 0.3%, higher than July’s actual 0.2%.[9]
Last night, there was also an even more direct side proof.
The U.S. Conference Board consumer confidence index for September fell to 81.9, the lowest since 2014.[5]
The market expectation is 89.
On the list of complaints consumers have, the top item is cost of living.
V. This round of revisions also leaves behind one side effect
CPI is not affected by this methodology change.[6]
So over the next few months, core CPI is likely to look more resilient than core PCE.
The two official inflation indicators will start talking past each other.
For trading, this isn’t an academic issue.
When the two indicators diverge, the longs will pick the softer argument.
The shorts will pick the harder argument.
And you need to know why these two numbers are separated in advance.
VI. At 8:30 tonight, where should you look?
First, look at the month-over-month figures—don’t focus only on year over year.
Year over year, tonight it could be lower—that’s the ruler’s effect, not an inflation concession.
Second, look at the revision crosswalk released that day by the Bureau of Economic Analysis.[4]
The official statement explicitly says it will publish a comparison table on the release day.
It maps revisions to source data and conceptual changes.
Third, back to the end where your position is truly priced.
On September 29, the U.S. Treasury’s official curve shows three sets of numbers.
10-year nominal yield is 5.26%, and 30-year is 5.59%.
10-year real yield is 2.91%.[7]
Switching to a new ruler won’t make these three numbers smaller.
When writing this, Binance XAU/USDT perpetuals are trading at $4,181.
Bitcoin is at $83,324.[8]
Their valuations are all tied to the same denominator.
This denominator is the real interest rate—not the statistical bureau’s ruler.
After that, there’s another date to watch: the Nonfarm Payrolls report on October 2.[5]
Finally
Tonight you’ll see a lower inflation number on your screen.
But that’s a new ruler—not a smaller tiger.
Treating adjustments to statistical definitions as a concession to inflation is the easiest mistake to make tonight.
—MK keeper
Data cut-off: September 30, 2026 at 10:40 CST (Beijing time). Historical PCE-related data and revision methodologies come from official releases by the Bureau of Economic Analysis (BEA), with revision magnitudes reflecting Goldman and UBS research estimates and non-official results; U.S. Treasury nominal and real yields come from the Treasury’s daily curves, latest as of September 29; consumer confidence comes from the Conference Board’s September reading; gold and bitcoin prices are from Binance XAU/USDT and BTC/USDT perpetual realtime data (24/7 trading; perpetual levels and futures levels cannot be mixed). The conditions and position discipline mentioned in the article are used for risk management and do not constitute investment advice.
Sources
[1] BEA — Annual Update of GDP, Industry, and State Stats Publicly Available Starting Sept. 30（2026-08-17）— Annual Update of GDP, Industry, and State Stats Publicly Available Starting Sept. 30 | U.S. Bureau of Economic Analysis (BEA)
[2] BEA Survey of Current Business — Preview of the 2026 Annual Update of the National Economic Accounts（June 2026）— SCB, Preview of the 2026 Annual Update of the National Economic Accounts, June 2026
[3] BEA — Personal Income and Outlays, July 2026（BEA 26–39）— Personal Income and Outlays, July 2026 | U.S. Bureau of Economic Analysis (BEA)
[4] BEA — Information on 2026 Annual Updates to the National, Industry, State, and County Statistics — Information on 2026 Annual Updates to the National, Industry, State, and County Statistics | U.S. Bureau of Economic Analysis (BEA)
[5] Investing.com Finance: US September consumer confidence index falls to 81.9, the lowest since 2014 — US September consumer confidence index falls to 81.9, lowest since 2014 - Finance - Investing.com Finance
[6] Trading Desk News relays the Goldman (Manuel Abecasis) and UBS (Alan Detmeister) research reports: Estimated impact of BEA methodology revisions on core PCE — US PCE Calculation Faces 20-Basis-Point "Paper Cooling" Overhaul; Wall Street Warns: Selective Methodology Changes Risk Credibility — BigGo Finance
[7] U.S. Treasury Daily Treasury Real Yield Curve & Daily Treasury Yield Curve Rates 2026 — Daily Treasury Rates | U.S. Department of the Treasury
[8] Binance Futures public market data — XAUUSDT / BTCUSDT tickers and open interest — fapi.binance.com/fapi/v1/ticker/24hr?symbol=XAUUSDT
[9] CoinGlass Economic Calendar: US Aug core PCE and ADP forecast — CoinGlass