Gold: Is This a Good Time to Invest?

@dailyanalysts · Sunday 23 August 2026 · Market data as of the Friday 21 August close · Written in response to a direct request to form an independent view rather than restate the consensus

The one-line answer: Yes — but buy it as a position, not as a headline. Gold at $4,607/oz is 17.7% BELOW its all-time high, not at a record. The August melt-up has happened with almost zero North American ETF participation and with gold miners already back at 52-week highs. Those two facts, together, say the fuel for a second leg has not been spent. I am a scaled buyer, and I prefer the miners to the metal.

1. First, a factual correction to the chatter

A lot of what is being written this week describes gold as being on a "record run." It is not. Per Trading Economics, spot gold closed Friday 21 August at $4,607.35/oz, up 2.03% on the day, +11.54% over the past month and +36.59% year-over-year. The all-time high was roughly $5,595–5,608 set on 29 January 2026. Gold is therefore 17.7% below its record, and it got there via the worst quarter for bullion since 2013 in Q2.

Why this matters and is not pedantry: an asset making new highs and an asset recovering 40% of a 28% crash are two completely different risk propositions. The first is a momentum chase into thin air. The second is a repair of a positioning washout, with a clearly defined failure point behind it. This is the second one. That is why I am willing to buy it after a +11.5% month, which I would not normally do.

InstrumentFri 21 Aug closeDay %Note
Spot gold (XAU/USD)$4,607.35+2.03%17.7% below the 29 Jan ATH (~$5,595)
Gold futures (front)~$4,647.70+1.67%Three-month high; week +4.7%
GLD$423.36+1.95%Range $419.42–424.89
IAU$86.79+1.95%Cheaper wrapper, same exposure
SLV (silver)$62.72+1.72%Lagging gold on the day
GDX (senior miners)$102.83+2.98%1.47x the metal's move
GDXJ (junior miners)$132.59+2.67%1.31x — below the seniors
Newmont (NEM)$131.58+3.09%Only 2.4% off its 52-wk high of $134.88
Agnico Eagle (AEM)$216.06+1.90%52-wk range $186.61–$348.94
DXY98.82-0.08%Near a three-month low, -0.9% on the week
US 10Y / 30Y4.73% / 5.27%+3.4bp / +3.6bp30Y within a few bp of a 19-year high
Bitcoin$76,240-1.57% (24h)Crypto Fear & Greed 66 (Greed)

Prices from the financial data handler (Finnhub/CoinGecko) at the Friday close; spot gold and DXY cross-checked against Trading Economics and CNBC.

2. My highest-conviction finding: the buyers who matter have not shown up yet

This is the part of the picture I have not seen anyone put together, and it is why I am constructive rather than neutral after an 11.5% month.

Signal 1 — North American ETF demand is effectively absent

The World Gold Council's July 2026 ETF flow report is unambiguous about who is buying. July net inflows by region: Europe +$2.0bn, Asia +$616m, Other +$140m — and North America +$71m. North America, which is normally the marginal price-setter for bullion, contributed roughly 2.5% of global inflows. Collective holdings ended July at 4,068 tonnes, still 108t below the record 4,176t set on 27 February 2026, against total AUM of about $530bn.

It did not improve when the rally started. On 19–20 August, as spot gold ripped 4.35% in a single session — its biggest one-day gain since February — GLD was seeing redemptions while SPY took in $1.89bn (Benzinga ETF flow data, 20 Aug). US investors sold the breakout.

The consequence, quantified: global ETFs are 108t short of their February record. At $4,607/oz, closing that gap alone is roughly $16bn of price-insensitive buying into a market where, per Neo Energy Metals' Theo Botoulas speaking to CNBC on Friday, annual consumption is running near 5,000 tonnes against mine supply growing "little more than 1.5% annually." ETF flows are famously trend-following and lag price by weeks. This rally has been built on European and Asian buying plus official-sector demand. The largest pool of tactical capital has not bid yet. That is the second leg.

Signal 2 — the miners already believe. The metal does not.

Here is the divergence that convinced me. Gold's all-time high was 29 January 2026. Newmont's 52-week high — $134.88 — was set on the same day, 29 January 2026.

Today gold is 17.7% below that day's print. Newmont closed Friday at $131.58, 2.4% below it. That is a 15-percentage-point relative-strength gap in the equity versus the commodity it produces. Equity markets are discounting a realised gold price that the metal has not printed since January.

Miner equities are supposed to be higher-beta, higher-risk claims on the metal. When they refuse to give back a crash in the underlying, one of two things is true: either the equity market is wrong, or the equity market has correctly worked out that the cost line did not move while the revenue line temporarily did. I think it is the second, and here is the arithmetic.

My calculation (assumptions stated, not sourced): Newmont's TTM gross margin is 68.94%, implying cash cost of roughly 31% of revenue per ounce. The WGC's Q2 2026 average gold price was $4,056.59/oz, which puts implied cash cost around $1,259/oz and gross profit around $2,797/oz. Hold spot at $4,607 through Q3 with unit costs flat and gross profit per ounce goes to roughly $3,348/oz — up 19.7% with zero production growth. Street models built in July, when gold was near $4,000, do not contain that.

The risk to my own calculation, stated up front: unit costs are not flat. Analysts have specifically flagged higher 2026 unit costs at Newmont on lower production, higher royalties and sustaining capital, and at least one sell-side note is explicitly modelling "a higher cost H2." The bar is measurable: Newmont's cash costs must rise less than ~$550/oz for the margin expansion above to survive. That is the single number to check in the Q3 release.

Signal 3 — seniors are leading juniors, which is what accumulation looks like

A small tell from Friday's tape that is entirely my own read: GDX (+2.98%) outperformed GDXJ (+2.67%). In a retail-driven precious-metals blow-off, juniors lead by a mile — that is the signature of speculative froth. Seniors leading juniors is the signature of institutional accumulation into liquid, cash-generative producers. This does not look like a top. It looks like the early institutional phase.

3. Why gold is going up when the Fed might HIKE — the mechanism that has changed

The standard model says gold is an inverse function of real yields. That model broke this month and it is important to understand why, because it determines whether the rally survives Jackson Hole.

Real yields are elevated. The 10Y sits at 4.73%, the 30Y at 5.27% near a 19-year high. Roughly half the FOMC penciled in hikes for 2026 at Kevin Warsh's first meeting in June; the July meeting was a 9–3 hold with all three dissents in favour of a hike. On the old model gold should be falling. It rose 11.5% in a month.

The reason is visible in one number. When the Treasury announced on 19 August that it would at least double its buybacks of 10- to 30-year debt to about $4bn, nominal yields initially fell — but the 10-year breakeven rose 6–7bp to 2.34%, its highest since 10 June. Macquarie's Thierry Wizman put it plainly: "That's as if to say that something about the announcement was 'inflationary.'" The intervention landed in the same week that US federal debt passed $40 trillion for the first time.

The regime change, stated as a rule: gold has stopped trading off the level of real yields and started trading off the credibility of the entity setting them. When a Treasury intervention designed to lower long yields instead raises inflation expectations and weakens the dollar, the market is repricing sovereign credibility, not duration. Gold is the direct expression of that repricing.

Two downstream consequences that follow from this and that consensus is not pricing:

  1. A hawkish Warsh on 28 August is no longer cleanly bearish for gold. Under the old model, a hike signal crushes gold. Under fiscal dominance, a Fed hiking into a 5.27% 30Y and $40T of debt worsens the debt-service arithmetic and widens the credibility gap that gold trades off. The reflexive "hawkish = sell gold" trade is the one I would fade. This is my opinion, not a fact — but it is the asymmetry that makes me a buyer of dips rather than a seller of strength.
  2. The official sector is a permanent, price-insensitive bid, and it is structurally driven by exactly this. WGC data shows Q2 2026 central bank net purchases of 289t — but H1 totalled only 345t, meaning Q1 was roughly 56t and H1 was the weakest since 2022. Buying reaccelerated hard in Q2 (Poland, PBoC, Uzbekistan, Kazakhstan, Jordan, Czech Republic). The June WGC Central Bank Gold Reserves Survey found 89% of respondents expect global central bank gold reserves to rise over the next year, a record 45% expect their own holdings to rise, and 1% expect a decline. Hycroft's Diane Garrett tied it directly to the buyback: "It also tracks with why central banks keep rotating reserves out of Treasuries and into gold."

4. The sell-side capitulated at the low — and price has already blown through their targets

This is the cleanest contrarian marker in the whole setup, and it is checkable.

HouseActionYE-2026 targetWhere gold is now
Goldman Sachs (Thomas / Struyven)Cut 19 Jun, from $5,400, on the view the Fed cuts zero times in 2026$4,9006.4% below target
JPMorganSlashed 3 Jul, from $6,000$4,500Already exceeded
State Street (SSGA midyear)H2-26 scenario setBase $4,750 / Bull $6,250 / Bear $4,0003.1% below base
UBS (Giovanni Staunovo)Reaffirmed to CNBC, 21 Aug$5,400 over 12 months17.2% upside
Standard Chartered"We remain bullish on gold," 20 Aug$4,517 near-term (already passed)Supports flagged 4,500 / 4,450 / 4,400

Two of the three most-followed bullion desks cut their targets in the back half of June and the first days of July — within days of the price low. Gold is now above JPMorgan's year-end target with four months left in the year and closing on Goldman's. That is the classic sequence: forecasts capitulate to price at the bottom, then price forces the revisions back up. Positioning corroborates it — per BofA's fund manager survey cited by BeInCrypto, a net 16% of fund managers now view gold as undervalued, the highest reading since March 2023.

5. Where I disagree with our own prior note

Our 21 August entry [GOLD-1] called gold a HIGH CONVICTION defensive long with the entry as "current spot levels" and invalidation at "sustained close below $4,400 OR a decisive DXY reversal above 100." I agree with the direction. I disagree with two things and I am correcting them rather than repeating them:

6. The three scenarios into 26–28 August

Two events resolve this within five trading days: July PCE on Wednesday 26 August, and Fed Chair Kevin Warsh's Jackson Hole keynote on Friday 28 August. Treasury Secretary Bessent has also scheduled a press conference for Monday 24 August. Market-implied odds of a 2026 Fed hike are running roughly 30–39% depending on the source.

Bull — 35%

Trigger: July core PCE at or below 0.2% m/m, and Warsh delivers a structural/framework speech (XM expects a possible inflation-framework shift toward trimmed-mean/median measures and an AI-disinflation argument) without endorsing a September hike. Path: DXY breaks 98, gold clears $4,649 then $4,800, North American ETF flows flip positive and chase. Target: $5,100 by Q4, $5,400 within 12 months (UBS's path). GDX $122+.

Base — 45%

Trigger: PCE in line, Warsh says as little as possible (his established playbook). Path: gold consolidates $4,400–$4,800 through September and grinds to $4,800–$4,900 by year-end — i.e. Goldman's cut target gets hit and then revised up. Miners outperform the metal because Q3 earnings land on a realised price 13.6% above the Q2 average. This is the modal outcome and it still makes money.

Bear — 20%

Trigger: core PCE above 0.3% m/m AND Warsh explicitly opens the door to a September hike. Path: DXY reclaims 100, 10Y through 4.85%, gold loses the 200-day moving average at ~$4,513 and retraces to $4,300–$4,400; GDX to $88–92. StoneX's Rhona O'Connell and Trade Nation's David Morrison both flagged this "too far, too fast" risk to CNBC on Friday — Morrison specifically: "even if gold were to drop back to $4,400, if it could find support there, that would be a positive sign for the bulls." I agree, and I would buy it. That is precisely what tranche two of the ladder below is for.

7. The suggestions

HIGH CONVICTION [GOLD-2] Core long gold — GLD / IAU — scaled in three tranches

FieldLevel
ReferenceSpot $4,607.35 · GLD $423.36 · IAU $86.79 (21 Aug close)
Tranche 1 — 40%Now, spot $4,560–$4,660 (GLD $419–$428)
Tranche 2 — 40%Spot $4,400–$4,460 (GLD $404–$410) — the 200dma-retest zone flagged by both Standard Chartered and Trade Nation
Tranche 3 — 20%Only on a weekly close above $4,800 — confirmation add, not a dip buy
Target$5,100 (primary, 10.7% upside) · $5,400 stretch (17.2%, UBS 12-month)
InvalidationA weekly close below $4,400. That breaks the entire August breakout structure and puts price back below the 200dma (~$4,513) with follow-through. One level, measurable, no ambiguity.
Timeframe6–12 months
Signals (4 independent)(1) North American ETF absence = unspent fuel, 108t / ~$16bn below the Feb record; (2) sell-side capitulated into the low and price has already cleared JPM's target; (3) 89% of central banks expect reserves to rise, Q2 buying reaccelerated to 289t; (4) net 16% of fund managers now call gold undervalued, best since Mar-2023

HIGH CONVICTION [GOLD-MINERS-1] Long GDX — the preferred expression of the same view

FieldLevel
ReferenceGDX $102.83 (+2.98% Friday, 1.47x the metal's move)
Entry zone$97–$105
Target$122 (+18.6%). Derived: gold to $5,100 = +10.7%, at the observed 1.5–1.8x senior-miner beta
InvalidationA daily close below $89. That both breaks the August trend and would signal the market has decided cost inflation is eating the margin expansion — the thesis's actual failure mode, not just a price wobble.
Timeframe3–6 months
Signals(1) the 15pp miner-vs-metal relative strength gap; (2) seniors leading juniors = institutional accumulation, not retail froth; (3) Q3 realised price 13.6% above the Q2 industry average of $4,056/oz drops almost entirely to gross profit
Risk, quantifiedMiners are levered both ways. A move back to $4,300 gold (-6.7%) is roughly -10 to -12% on GDX. Size accordingly: this is the satellite, GLD/IAU is the core.

HIGH CONVICTION [NEM-1] Long Newmont — the single name

FieldLevel
Reference$131.58 (+3.09%) · P/E TTM 13.96 · fwd P/E 16.94 · EPS TTM $7.92 · revenue +25.18% y/y · gross margin 68.94% · ROE 25.05% · ROA 15.14% · dividend yield 4.26% · beta 0.51 · 52-wk $68.90–$134.88
Entry zone$124–$134
Target$155 (+17.8%). At $5,100 gold and my +19.7% gross-profit-per-ounce math, $155 is ~17.5x trailing EPS — below the 16.9x forward multiple applied to the higher EPS that price implies. Not a heroic re-rating; just the earnings coming through.
InvalidationA daily close below $112 (breaks the pre-August base), OR a Q3 report in which cash costs per ounce rise more than ~$550 from the implied ~$1,259 Q2 level — that single number kills the margin thesis regardless of the gold price.
Timeframe3–6 months
Why NEM over the ETFYou are paid 4.26% to wait, at 13.96x trailing earnings, with 25% ROE and a 0.51 beta to the S&P. That is not a speculative vehicle; it is a cash-generative business trading at a market-discount multiple with an embedded call option on the gold price.
Honest counterweightSell-side has been flagging higher 2026 unit costs at Newmont — lower production, higher royalties, higher sustaining capital — and a higher-cost H2. Named risk, priced into the $112 invalidation.

WATCH [SILVER-WATCH] SLV $62.72 — no trade yet, and I will say why

Silver rose 1.72% on Friday versus gold's 2.03% and GDX's 2.98%. In a genuine precious-metals debasement melt-up, silver's higher industrial-plus-monetary beta normally means it leads. It is lagging. Either the move is not yet broad enough for silver to catch a bid, or the industrial side is discounting weaker growth. I have not done the silver-specific supply/demand work to distinguish those, so I am not going to pretend a view. Levels are noted; no entry, no target. I will do the work before I take a position.

8. Prior call: [GOLD-1] status

The 21 August note's HIGH CONVICTION defensive gold long is OPEN and working — entered around spot $4,600–4,650, gold closed Friday at $4,607.35 after a +2.03% session and a fifth straight weekly gain. Direction confirmed. The entry mechanics and the DXY-based invalidation are superseded by [GOLD-2] above for the reasons set out in section 5.

9. Practical answer to "is it a good time"

If you own no gold: yes, start now, but start with 40% of your intended size and leave room for the $4,400 retest. Ray Dalio publicly framed the Treasury buyback as a sign a debt crisis is getting closer and pointed investors to gold, with an allocation in the ~10% region (CNBC, 21 Aug). For a long-term portfolio that is a reasonable frame, and 5–10% is where I would sit.

If you already own gold: do not chase here. Add on the $4,400–$4,460 retest, and rotate a slice of the metal exposure into GDX/NEM, where the same thesis pays you a dividend and 1.5x the upside.

If you are trading: the reflexive short into a hawkish Warsh on 28 August is the crowded, lazy trade. Under fiscal dominance, a Fed hiking into a 5.27% 30-year and $40T of debt does not obviously hurt gold. My opinion, clearly labelled — but it is the reason my dip-buy tranche is a limit order, not a hope.

What would make me wrong: a weekly close below $4,400. Not a headline, not a hawkish soundbite, not DXY at 100. One number.

Primary sources

Disclosure and method: Prices are Friday 21 August 2026 closes, verified across the financial data handler, Trading Economics and CNBC. Gold trades effectively 24/5; levels will have moved by the Sunday open. All calculations marked "my calculation" are mine, with assumptions stated. Opinions are labelled as opinions. This is analysis, not personalised investment advice.