Core PCE lands with gold 21% below its January high and a market that has stopped trading it off real yields

Upcoming catalystiThe event that could move the share price.

The Bureau of Economic Analysis published the Personal Income and Outlays report for August at 12:30 UTC on 30 September 2026. Core PCE, the inflation gauge the Federal Reserve targets, rose 0.2% on the month and 3.0% over the year, against a 3.4% consensus, and headline PCE rose 0.3% and 3.4%. It was the last major reading on that gauge before the 28 October rate decision, and market odds of a rise at that meeting fell to about 38% from about 51%. The event is complete.

Event date and timeiWhen the event is expected. Shown in your own time zone, or in UTC if you switch the clock at the top of the page. If the exact hour is not confirmed, you only see the date.

30 September 2026, 12:30 UTC, which is 8:30 in the morning in Washington. The date and hour are on the BEA release calendar. Federal funding runs to 11 December under the continuing resolution signed on 2 September, so nothing about this release is at risk from a lapse on 1 October.

Main contract affected

Gold (GC) on COMEX, December 2026 delivery. Worth stating precisely, because the feed's quote is not the nearest month: October 2026 is trading at $4354.6 and December is the liquid contract, so every figure on this card refers to December. It does not expire until December, well after the release. Silver (SI) was the alternative and is rejected because silver carries an industrial demand story of its own that has nothing to do with a US inflation print, which would muddy the reading rather than sharpen it.

Current priceiWhere the shares traded when this write-up was made, and when that was.

$4394.7 as of 22 Sept 2026, 18:11 UTC

Expected directioniWhich way we expect the token price to go after the event. Positive means it rises, Negative means it falls, Mixed means it could go either way.

Mixed, and the outcome is what mixed was for. The print landed on the soft side this card said would lift gold, at 3.0% against a 3.4% consensus, and the metal rose more than $20 to about $4,210 within minutes. It then gave all of it back and closed the session about 0.7% lower near $4,156, because the ten year Treasury yield reached 5.306%, its highest since June 2007, so the easing in real yields the positive case needed did not arrive with the soft print. Two sessions later gold is near $4,182, which is inside the half a percent this card called the most likely single outcome. The condition for the positive case was met and the payoff was not, which is the shape a genuinely two sided event produces.

Action before the eventiWhat we suggest you do before the event. Buy, Watch or Avoid.

The useful thing to settle before Wednesday morning is not which way to lean but how much weight to give the print at all. Gold has spent three months not doing what the rates market said it should, and anyone sizing a position off the old relationship between this contract and real yields is sizing it off a relationship that has not held this year.

Why it mattersiHow the event reaches the price. Through demand, supply, how much the network is used, fees, or the rules the token trades under.

Core PCE is the number the Fed has said it is steering by, and the Fed is in a hiking cycle rather than a cutting one. A firm print hardens the case for another rise on 28 October, which lifts real yields, and a non-yielding metal is the classic loser from that. What makes this card interesting is that the classic result has stopped arriving.

Catalyst confidenceiHow sure we are that the event happens as described. Scored out of 100. Higher means more sure.

The event has happened. The Bureau of Economic Analysis published the August report at 12:30 UTC on 30 September 2026, on the date and at the hour its own calendar carried. Confidence that the event occurs: 100%.

Price-reaction confidenceiHow sure we are that the token moves the way we expect. Scored out of 100. Higher means more sure.

High, because the reaction has happened, and the card read it correctly. Gold jumped more than $20 to about $4,210 on the 12:30 UTC release on 30 September 2026, closed the session about 0.7% lower near $4,156 as the ten year yield reached 5.306%, and traded near $4,182 on 1 October. The net move across the two days this card allowed itself is under one percent, which is what the 50 figure said was most likely. The reason it stayed small was the one the card named: the real yield channel this event travels down was pulling the other way at the same moment. Confidence in the reaction: 85%.

Expected moveiHow far we think the price moves, and how long the move may last.

A firm core print takes it down $30 to $70 an ounce, which is 0.7% to 1.6%. A soft one lifts it by the same, $30 to $70, or 0.7% to 1.6%. Either way that is a fraction of a normal session here rather than a multiple of one: the average true range over the last fourteen days is $106.2, or 2.42% of the price, so the expected move runs from about a quarter of an ordinary day's range to two thirds of it. The reason it is that small is the whole argument of this card and is set out under What the market may be missing.

What exactly is happeningiThe event itself in full. Who is involved, what changes, and the amounts and conditions that matter.

The BEA publishes personal income, personal spending and the PCE price index for August. The market reads three lines: headline PCE year on year, core PCE excluding food and energy year on year, and the monthly change in core. Core PCE matters more than CPI here because it is the series the Federal Open Market Committee's own projections are written in. The committee voted 12 to 0 on 16 September to raise the target range by a quarter point to 3.75% to 4.0%, and Chair Kevin Warsh signalled that a further rise this year is on the table. He has also shortened post-meeting communication since May and declined to publish his own forward guidance, which pushes weight onto the data and away from anything the Fed says between meetings. This print is the last core PCE reading before 28 October.

Market expectationsiWhat most traders currently expect to happen. Bullish means they expect a rise, Bearish a fall, Neutral neither.

Bearish going in, and wrong on the number. Consensus for August core PCE was 3.4%, up from July, and the Cleveland Fed nowcast agreed at 3.40%. The print was 3.0% on the year and 0.2% on the month, so both were four tenths too high on the annual rate, which is a large miss for a figure that well anchored. What the market did with it is the part this card cares about. Odds of a rise on 28 October fell to about 38% from about 51%, so the print moved the thing it is supposed to move. Gold still finished the session lower, because the ten year yield went to its highest since June 2007 in the same hours.

How much is already priced iniHow much of the expected outcome the share price already reflects before the event happens.

The hawkish case is very well priced, which is the first reason this is a mixed call rather than a negative one. Market-implied odds of a Fed hike moved from roughly 30% in late July to 84% by the middle of September, and then the hike was delivered. The 10 year real yield has pushed past 2.6% and ended last week at 2.67%, the highest in more than twenty years, and the 30 year Treasury yield has climbed above 5.2% for the first time since June 2007. The next meeting on 28 October is already widely expected to bring another rise. The gold price carries that. The contract is at $4394.7, which is 35.9% of the way up a 52 week range of $3718.1 to $5586.2, and 21.4% below the high at the top of it. It is down 6.59% over the last twenty sessions. A market that had not absorbed the hiking cycle would not be trading a fifth below its January peak. Positioning has come down with it but is still one-sided. The CFTC report for 8 September showed managed money holding 145,804 gold longs against 10,832 shorts, a ratio of more than thirteen to one, with the long side easing to 142,394 by 15 September against open interest of 409,899.

What else lands that dayiOther news and events on the same day that could push the price around, and whether they help this trade, bury it, or work against it.

A crowded morning, and two of the three compete directly. The ADP National Employment Report publishes at 12:15 UTC, fifteen minutes ahead of PCE, and speaks to the same question about how much room the Fed has. The third estimate of GDP lands in the identical 12:30 UTC minute as the PCE data. It is also the last day of the quarter, which brings rebalancing flows that have nothing to do with inflation into the same session. The day is crowded. A labour number arriving a quarter of an hour early can set the dollar before the inflation print is out, and whatever this contract does by the close will have three plausible authors.

Bullish scenarioiWhat would have to happen for the price to rise, and why the market would read it that way.

Core PCE comes in below the 3.4% consensus, holding at July's 3.3% or lower, or the monthly rate prints 0.1%. That would be the first real sign of cooling in the Fed's own gauge after 65 months above target, and it takes 28 October from near-certain back to a question. Real yields ease and gold rises $30 to $70, which is 0.7% to 1.6%, with the larger end requiring the labour number fifteen minutes earlier to point the same way.

Neutral scenarioiThe outcome that simply meets expectations and moves the price very little.

Core lands at the 3.4% consensus with the monthly rate at 0.2%, a tenth above the last print, which is what the tariff pass-through story implies. Nothing changes about October. Gold moves less than half a percent on the print and the session is decided by quarter end flows instead. This is the most likely single outcome.

Bearish scenarioiWhat would have to happen for the price to fall. A delay, a disappointment, or supply nobody expected.

Core firms above the 3.4% consensus, or the monthly rate prints 0.3%. That confirms tariffs are still feeding through and locks 28 October in. Real yields extend their twenty year highs, and gold falls $30 to $70, which is 0.7% to 1.6%.

Result needed for a big moveiWhat the outcome has to be for the stock to move a lot instead of a little.

For anything beyond the range above, the print has to change the destination rather than the next step. A core reading at or above 3.5% would put a second hike into the conversation rather than a first, and a reading at or below 3.0% would be the first evidence that the tariff impulse has peaked. Neither is the base case, and the move either way would be a multiple of the range quoted here rather than inside it.

Other contracts affected

Silver (SI), same direction as gold and more of it, because silver trades the same macro with a higher beta and less central bank support beneath it. Platinum (PL), same direction and weaker still, since its price is set more by autocatalyst demand than by real yields. Palladium (PA), same again and weakest of the three, for the same reason with a narrower demand base.

Main risksiWhat could go wrong, and what would prove this call wrong.

Medium risk overall, and the shape of it is unusual for a data card. The first risk is attribution rather than direction. With ADP fifteen minutes ahead, the third GDP estimate in the same minute and quarter end flows through the afternoon, a reader can be right about the inflation print and still see the price do something else for reasons this card correctly named but could not weight. The second is that the argument below, that this market has stopped trading off real yields, is an observation about the last several months rather than a law. Relationships that break can mend. A sufficiently hot print, on a day when the labour number agrees, could remind everybody what the old model said. The third is that a mixed call is the honest answer to this event and it is also the least useful one to hold a position with. The value here is in the sizing rather than the side.

What the market may be missingiSomething we think investors have overlooked or read wrong.

Gold has stopped paying attention to the thing this release is supposed to move it through. Look at what the last two months did. Market-implied odds of a hike climbed from roughly 30% in late July to 84% by mid-September. The Fed then delivered, unanimously. The 10 year real yield reached 2.67%, its highest in over twenty years, and the 30 year crossed 5.2% for the first time since June 2007. Under the relationship that governed this metal for decades, every one of those is a reason to sell it. Gold is up 18.42% over the last 252 sessions and rose 1.28% in the last five, including the week of the hike itself. The reason is a change in who sets the price at the margin. Central banks bought 288.9 tonnes net in the second quarter, 62% more than a year earlier and the strongest second quarter in the World Gold Council's series, with Poland adding 51 tonnes and China 33. Over 2022, 2023 and 2024 they took more than a thousand tonnes a year, and 863 tonnes in 2025, which is roughly a quarter of annual mine production. A reserve manager building a position measured in decades does not sell it because core PCE printed a tenth high. The buyer that used to set this price was a rate-sensitive Western ETF investor. That buyer now shares the market with one that is not rate-sensitive at all. The tell is that the analysts have already moved and the calendar has not. Goldman Sachs holds $4900 for year end, JPMorgan $4500 for the fourth quarter, HSBC an average of $4560 and Bank of America $4360, and all four held those numbers through the entire repricing of hike odds described above. Nobody adjusted a target for the hike, because the hike was never what their targets rested on. The second thing under-weighted is that the curve agrees. February 2027 trades at $4427.1, which is $37.7 above the front month, about 0.9% over two months or roughly 5% a year, and the strip keeps climbing through April at $4450.2 and June at $4497.9. That is plain cost of carry at a 4% funds rate plus storage, and it is what a metal with no physical scramble looks like. So the support under this price is not a shortage. It is accumulation. It is also worth knowing that a long futures position pays that 5% a year at every roll, a cost that never appears beside a year-end target.

TrendiWhether the price is above or below its own 50-day and 200-day averages. Above both is an uptrend, below both is a downtrend.

Down over the medium term and directionless in the short. The contract is 21.4% below its 52 week high and has fallen 6.59% over the last twenty sessions and 5.58% over 120, while still holding a gain of 18.42% over 252. The last five sessions added 1.28%, which is a bounce inside a correction rather than a change of trend.

Market structureiThe shape of the chart. Rising highs and rising lows mean an uptrend, falling ones mean a downtrend, and neither means the price is stuck in a range.

A market working out where its floor is. The high at $4755.0 on 25 August gave way to a sharp break, and since then the range has been defined by the low at $4273.3 on 16 September, made on the day of the Fed decision, and the recovery highs around $4439.8 on 18 September. The last six sessions have all closed between $4332.8 and $4424.9, a band of under 2.2%, which for a contract with a 2.42% average daily range means the market is coiled rather than trending.

Support and resistanceiPrices where buyers have stepped in before, and prices where sellers have. A price often stalls or turns at these.

Support first, because that is where the structure is. The 50 day moving average at $4338.3 and the 100 day at $4358.0 sit only $19.7 apart and both lie within 1.2% below the current price, which makes them one shelf rather than two levels. Under them is $4273.3, the 16 September low, which is 2.65% below the current price and is the level the whole recent range is built on. Resistance is the 20 day moving average at $4461.8, which is 1.65% above, and then a genuine cluster: the 3 September high at $4558.5 sitting almost exactly on the 200 day moving average at $4554.9, both a little under 3.9% above the current price. Getting through that pair would be the first thing this chart has done in a month that means anything.

VolumeiWhether trading is picking up or dying down. Rising volume behind a price move means more people agree with it.

The last twenty sessions have averaged 181,055 contracts a day against 152,460 over sixty, and recent daily volume has run between 142,919 and 249,081. The heaviest session in the window was 16 September at 249,081, the day of the Fed decision, and the second heaviest was 28 August at 274,603, the day the market broke down from its high. Volume is arriving on the decisive days rather than on the drift, which is what an orderly correction looks like.

RSI and momentumiWhether the price has been rising or falling hard lately, and whether that push is getting stronger or running out.

RSI(14) is 47.3, which is as close to neutral as this indicator gets, and it is the single most informative number on the chart. There is no oversold bounce signal and no overbought warning. Momentum is neutral, and the market is genuinely undecided rather than leaning.

Moving averagesiThe average price over the last 20, 50, 100 and 200 days. Trading above them is generally a healthy sign, below them a weak one.

Tangled, which is the definition of a market without a trend. The price sits below the 20 day at $4461.8 and well below the 200 day at $4554.9, while holding above the 50 day at $4338.3 and the 100 day at $4358.0. The 200 day is still above the 100 day, which says the longer-term damage from the January peak has not been repaired.

VolatilityiHow much the share price swings day to day right now, compared with how much it usually swings.

High by this contract's own standards, and that matters for reading everything above. The average true range over fourteen days is $106.2, or 2.42% of the price, and realised volatility over the last 30 sessions is 20.3% annualised. Gold ordinarily moves half a percent to one percent in a day and a two percent day counts as a large one, so a 2.42% average range is a market in a genuinely disturbed state.

Breakout and breakdown levels

Upside, a close above $4461.8 opens the $4554.9 to $4558.5 pair, and only a close above that pair repairs the chart. Downside, a close below $4338.3 puts the 16 September low at $4273.3 in play, and a close under $4273.3 breaks the only floor the last month has built.

Technical biasiWhat the chart alone suggests, ignoring the news. Rated Bullish, Neutral or Bearish.

Neutral, and unusually literally so. RSI at 47.3, the price inside a 2.2% band for six sessions, above two moving averages and below two others, with the last five sessions up and the last twenty down. A chart that is about to be handed a catalyst and has no opinion of its own about which way to take it.

What would invalidate this biasiThe price level or chart change that would prove the technical read wrong.

There is no bias here to invalidate, so the question is what would create one. A daily close outside the $4273.3 to $4461.8 band, in either direction, turns this from a neutral chart into a directional one.

Inventory leveliHow much of the physical commodity is sitting in storage, compared with what is normal for the time of year. Tight means there is less than usual, which supports the price.

Ample. COMEX vaults held 27.38 million ounces of gold on 9 September, of which 15.12 million ounces were in the registered category on 8 September, and exchange-traded funds held 4,068 tonnes at the end of July. Gold is not a market that runs short in the way an energy or an agricultural contract does, and nothing in the visible stock says otherwise right now.

Inventory trend

Building. COMEX stocks were essentially flat between 3 and 9 September, at 27.3776 million and 27.38 million ounces, while the ETF side has been adding: Kitco reported on 21 September that ETF holdings had extended a buying streak to eight consecutive days. The end-July figure of 4,068 tonnes was still below the record of 4,176 tonnes set on 27 February 2026, so the rebuild has not recovered the whole of the spring outflow.

Spread structureiWhether the market pays more for the commodity today than for delivery in a few months. Backwardation means today costs more, which is what a market short of supply looks like. Contango is the opposite, and it means holding a long position slowly costs you money.

Contango, and unremarkably so. February 2027 is $37.7 above the front month, roughly 0.9% over two months or about 5% annualised, with April at $4450.2 and June at $4497.9 continuing the climb, and October 2026 sitting below the front at $4354.6. That is the cost of carrying metal at a 4% policy rate plus storage and nothing more. There is no backwardation anywhere on this curve and therefore no sign of physical tightness.

Production trend

Rising. Mine supply is forecast to increase 2.4% year on year to 3,907 tonnes in 2026, after rising 2% to 3,815 tonnes in 2025, which was itself an all-time record. Output is strengthening in every region except Oceania and Europe.

Demand trend

Stable, with the two halves pulling against each other and both worth naming. Central bank buying is accelerating, at 288.9 tonnes in the second quarter and 62% above a year earlier. Against that, Metals Focus forecast in June that total gold demand would fall this year even as supply increased, while still expecting the average price to rise 43% in 2026. Those are not contradictory: the composition of demand is changing faster than the total.

Managed money positioningiWhether speculative traders are mostly betting the price rises or falls, from the weekly CFTC report. A crowded bet often means a smaller move than expected on good news, because most buyers have already bought.

Long-heavy, and thinner than it was. The CFTC report for 8 September showed 145,804 managed money longs against 10,832 shorts, more than thirteen to one. By 15 September the long side had eased to 142,394 against open interest of 409,899. The crowd is still on one side, but it has been trimming through the correction rather than adding into it, which is why this card does not lean negative on positioning the way the soybean one does.

Seasonality

Neutral. The autumn is conventionally the friendlier half of the year for physical gold demand in Asia, but the demand this card is about is financial rather than jewellery, and no seasonal pattern in reserve manager buying or in ETF flows could be established here worth trading on.

Dollar effect

A headwind, and the strongest single one against this contract. The Fed raised rates on 16 September and signalled another rise may come, which lifted the dollar, and the 10 year real yield at 2.67% is the highest in more than twenty years. Every one of those makes a dollar-priced, non-yielding asset harder to hold. The point of this card is that it has been making it harder for months without making it cheaper.

Freight and logistics

Not part of this story. Moving gold is cheap relative to its value and the metal that matters to this price is sitting in central bank vaults where it already is. There is no freight reading to give and none is invented.

Warehouse stocks

Stable. COMEX gold inventory was 27.3776 million ounces on 3 September and 27.38 million on 9 September, with registered stock at 15,110,183 ounces on 3 September and 15.12 million on 8 September. Neither category is moving in a way that suggests a delivery squeeze or an exodus.

Central bank gold buying

Rising, and this is the most important line on the card. Net central bank purchases were 288.9 tonnes in the second quarter of 2026, 62% above the same quarter a year earlier and the strongest second quarter in the World Gold Council's series, with Poland adding 51 tonnes and China 33, China's largest single quarter since late 2023. Annual buying ran above a thousand tonnes in 2022, 2023 and 2024 and 863 tonnes in 2025, which is roughly a quarter of what the world's mines produce. This is the buyer that does not read the PCE report.