Gold just ripped +18% in a straight line since our AI research detected it early September.
I flagged this setup in our newsletter on September 4, 2025 (Read the archive here: Link) - the technical breakout + low retail participation on the breakout were screaming a higher leg up before the move took off.
But now everyone's pinging me with the same question: Should I add more here?
The long term fundamental story remains rock solid: inflation hedging, Fed rate cuts, reserve diversification, and relentless central bank accumulation.
Here's the issue: the macro thesis hasn't changed, but tactical positioning absolutely matters. When a trade gets this vertical this fast, you need to assess whether you're early to a structural shift or late to a crowded room (ie, what's the short term and medium term setup look like?).
The challenge for investors? Raw positioning data is scattered across multiple sources. COT reports lag by weeks. ETF flows don't tell you if vol is bid or complacent. RSI might be overbought, but what actually happened the last time it printed these levels?
I'm going to show you how I used AI to synthesize gold's volatility structure, positioning data, and historical RSI analogs into a single tactical framework for short to medium term setups.
The goal: answer whether this trade is tactically exhausted or if there's still room to run before the next consolidation.
Here's how these three metrics tell us if gold is too crowded:
- Volatility dynamics — If implied vol is rich vs realized, markets are hedged and nervous (room to run). If implied is cheap, everyone's complacent (crowding risk).
- Positioning metrics — COT longs, ETF holdings, and open interest show if speculative capital is maxed out or if there's still dry powder.
- RSI historical analogs — When RSI hit similar extremes in the past, did gold consolidate shallow or collapse hard? That pattern reveals if momentum can persist or needs a reset.
Let's dig in.
The Prompt
We prepared a 3 page prompt here:
Gold_Crowding_Prompt_Davewang.pdf
The Result
Full results here: Link
AI surfaced three key findings that frame the tactical setup:
1. Volatility structure shows hedged momentum, not complacency.
GVZ spiked to ~32.8 on October 16 as gold hit fresh records. Meanwhile, 30-day realized vol sits at just ~17%.
Translation: Implied vol is trading rich to realized. Markets are paying up for protection, not getting comfortable. This is a momentum-charged, hedged tape.
Historically, this pattern precedes one of two outcomes: either a vol-expansion consolidation as positioning resets, or an over-the-top extension if structural flows persist. It does not signal complacency - which means the trade isn’t sleepwalking into a reversal.
2. Positioning is elevated but not at prior blow-off extremes.
AI computed a directional crowding score across three dimensions:
- ETF holdings: Within 2% of the November 2020 peak (~3,857t vs ~3,929t). Very full, but not maxed.
- COMEX open interest: Rising to ~487,732 contracts, confirming fresh speculative inflows.
- COT managed money longs: ~493 tonnes as of late September - elevated, but still below prior ATH extremes seen at past peaks.
Crowding score: High, but not extreme.
The kicker? Central banks remain net accumulators. 95% of surveyed central banks expect global reserves to rise over the next 12 months, and 43% plan to buy. That’s a structural, price-insensitive bid that puts a floor under any tactical shake-out.
Bottom line: Positioning is crowded via ETFs and futures, but COT hasn’t hit the blow-off levels that typically mark final tops.
3. RSI extremes point to consolidation risk, not reversal.
Daily RSI printed >80 this week - statistically stretched. AI mapped three historical analogs where RSI hit similar levels:
- August 2011 (RSI upper-70s/low-80s near $1,900): Gold dropped -11% in 1 month, -19% in 3 months.
- August 2020 (RSI >80 near $2,075): Gold fell -9% in 1 month, -14% in 3 months.
- April 2024 (weekly RSI highest since 2020): Shallow consolidation, then the rally resumed into 2025.
Pattern: When RSI crosses 80, gold typically consolidates for 1-3 months with mid-single to mid-teens pullbacks - unless structural buyers stay engaged (like 2024).
The 2024 case is the key exception: extreme RSI did not require a deep reversal because central bank flows kept the bid intact. That’s the exact dynamic we’re seeing now.
So what’s the tactical call?
AI assigns probability-weighted scenarios for the next 1-3 months:
- Base case (45%): Range-bound with a shallow retrace. Gold consolidates -8% to -12% to reset RSI while ETF holdings and central bank demand stay firm. Think digestion, not distribution.
- Bull case (40%): Extension after a brief pause. If real yields drift lower and ETF inflows persist, gold re-tests and marginally exceeds highs after volatility compresses.
- Bear case (15%): Hard reset. A jump in real yields + USD squeeze flips ETF flows negative, collapsing implied vol and triggering a >15% drawdown toward prior multi-month support.
Key risks to monitor:
- Real-yield spike or USD squeeze that forces ETF outflows
- Policy surprises that re-price the Fed path and crush implied vol
- COT positioning surge to prior extremes (watch for updated CFTC data)
- Central bank demand wobble beyond idiosyncratic slowdowns
How I'm Positioning on $GLD:
Gold is tactically crowded but not at a final top. The fundamental story remains intact, but the tape is vulnerable to a 1-3 month shake-out. If you’re already long, this is a tape to manage position size and prepare for volatility, not chase at current levels (I personally took a bit of profits on Friday). If you’re waiting to add, the base case suggests a better entry within the next few months as momentum cools and RSI resets.
Personal
I was invited by CLSA (one of largest investment banks in Asia) to speak on a webinar for their VIP clients on "AI in Investment Decision Making".
If you work on the buy-side, I may be able to get you an invite.
Reply to this email if you're interested and I can send a request to the organizers :)
