As of October 26, 2026, the board reads 85% bull / 15% bear. Every piece of evidence for and against gold, weighted by impact × confidence × freshness. Items decay with a 5-trading-day half-life and retire below 0.4 weight — this board cannot argue last month's tape. One event can only sit on one side: dedupe is enforced, and evictions are logged below.
Levels and one-month trends in real yields, the dollar, oil and Fed pricing, weighted by how much each moves gold. Full detail in the factor ledger below.
Weekly moves in prediction markets plus scheduled astro dates, weighted by impact × confidence × freshness. It reacts to news, not to levels.
The two lenses measure different things, so they can disagree: Lens 1 is what is pressing on gold today, Lens 2 is what moved this week.
Every major driver of the gold price, with three things kept strictly apart: what it reads today, which way that reading is pushing gold, and how much gold would move for a typical swing in that driver — its implied weight.
The biggest lever on gold: what it costs to hold a metal that pays nothing.
| Factor | Today's reading | Push today | Implied weight on gold |
|---|---|---|---|
US 10Y real yield (TIPS)
Real yields are falling, so the opportunity cost of holding gold is easing.
Model elasticity: about −0.11% on gold per +1bp of real yield (−11% per +100bp). Most of why gold sits below its January record.
FRED DFII10 · 2026-10-23 |
2.66% -10bp in a month |
Bullish push |
Computed in your browser from the simulator engine (±0.50pp real yield).
|
Fed hikes priced in (Kalshi)
Markets price a real chance of further hikes, which lift real yields.
Model: +18bp on the real-yield path per 2026 hike and +7bp per 2027 hike, so hikes work through the real-yield channel above.
Kalshi KXFEDDECISION · 26 Oct 2026 |
28 Oct 7% · 9 Dec 71% · 27 Jan 32% |
Bearish push |
Computed in your browser from the simulator engine (±1 hike in 2027).
|
10Y breakeven inflation
Rising inflation expectations support gold as a hedge; falling ones remove that bid.
Context rather than a separate model input: this is what the CPI slider approximates.
FRED T10YIE · 2026-10-26 |
2.34% +0.01pp in a month |
No clear push |
Context only — not a simulator input, so it carries no implied weight. |
Gold is priced in dollars and sold as an inflation hedge. Both channels cut two ways.
| Factor | Today's reading | Push today | Implied weight on gold |
|---|---|---|---|
Broad US dollar (Fed index)
A stronger dollar makes gold dearer for everyone else and pulls the dollar price down.
Model beta −1.1×: a 1% fall in the dollar adds about 1.1% to gold. This is the Fed's broad index, not the ICE DXY.
FRED DTWEXBGS · 2026-10-23 |
121.3 +1.0% in a month |
Bearish push |
Computed in your browser from the simulator engine (±3.1% dollar).
|
US CPI inflation (y/y)
Hotter inflation is feeding a more hawkish Fed faster than it feeds hedge buying.
Two-way: hotter CPI adds hedge demand (+0.004%/day per pp) but also pushes real yields up (+8bp per pp). In this calibration the reaction channel edges it, net mildly negative.
FRED CPIAUCSL · 2026-09-01 |
3.7% +0.3pp vs last month's y/y |
Bearish push |
Computed in your browser from the simulator engine (±1.0pp CPI).
|
Oil and war reach gold through the dollar and real yields, not directly; fiscal stress is the hedge channel.
| Factor | Today's reading | Push today | Implied weight on gold |
|---|---|---|---|
Brent crude
Brent is at or below ~$120, where oil's pass-through to the dollar and real yields is small, so it is roughly neutral for gold.
No direct beta (≈0 in 16 months of data). Oil lifts the dollar (+0.18% per +10% Brent) and real yields (+2.4bp per +10%), so spikes weigh on gold; above $120, a further +0.6bp per $1 and a quadratic growth drag.
FRED DCOILBRENTEU (a few days' lag) · 2026-10-20 |
$115.22 -0.8% in a month |
No clear push |
Computed in your browser from the simulator engine (±$20 Brent).
|
War / Middle East escalation
Prediction-market evidence on the Gulf tilts toward the safe-haven bid.
Two-faced: gold is the haven in some paths, the dollar is in others (the dollar rose in 4 of the last 8 oil spikes and gold fell with it). War 0 to 100 now spans roughly $250 of median outcome, down from about $470 before the Oct 2026 recalibration.
Kalshi / Polymarket via the evidence board · 26 Oct 2026 |
US-Iran military escalation risk (0.185); Hormuz traffic normalizing (0.115) |
Bullish push |
Computed in your browser from the simulator engine (±15 on the war index).
|
US fiscal stress & debt
Not read live. It matters because when a sell-off is about US credit itself, gold decouples from yields and becomes the hedge.
Model: up to a 30% cut in gold's real-yield beta as stress rises, plus +0.0006%/day of hedge drift per index point.
Model assumption |
No live feed: a model assumption (index 40 of 100) |
No clear push |
Computed in your browser from the simulator engine (±20 on the fiscal index).
|
The buyers that put a floor under dips.
| Factor | Today's reading | Push today | Implied weight on gold |
|---|---|---|---|
Central-bank buying
Record official-sector buying is the structural bid under every dip.
Model: demand drift scales with the monthly pace (30t/month baseline), and the dip-buying floor strengthens below $3,950 and again below $3,800.
World Gold Council (entered by hand) |
130t year to date |
Bullish push |
Computed in your browser from the simulator engine (±16 t/month).
|
These are evidence, not model inputs, so they carry no implied weight above. Their weight in the headline scoreboard is earned from measured forecast skill.
| Signal | Reading | Push today | Note |
|---|---|---|---|
| Trend regime (weekly / daily) | NEUTRAL / BEAR | No clear push | Weekly and daily trends disagree, so the net trend read is neutral. |
| RSI weekly | 50.6 | No clear push | Neutral range (30–70). |
| RSI daily | 23.1 | Bullish push | Oversold below 30. |
| RSI monthly | 45.0 | No clear push | Neutral range (30–70). |
| 50-day vs 200-day average | $4,224 vs $4,341 | Bearish push | Death cross: the 50-day sits below the 200-day. |
| Vedic astro bias | Bearish | Bearish push | Scored like every other pillar; it currently carries 19% of the headline weight. Unproven is the honest default. |
| Pillar | Earned weight | Scored calls | Hit rate |
|---|---|---|---|
| Price action | 72% | 16 | 58% |
| Macro & flows probation | 4% | 0 | — |
| Prediction markets probation | 4% | 8 | 37% |
| Astrology | 19% | 28 | 54% |
Board history: 85% bull share, up from 52% around 2026-10-12.
Each card says which side it sits on and why. Sides come from the week's change in odds times the market's effect on gold, so a title can sit under Bull when its odds fell.
Pre-registered astro window, Oct 31
A scheduled, pre-registered astro date, not news: counted at a fixed weight with confidence capped at 0.5.
Pre-registered astro window, Nov 09
A scheduled, pre-registered astro date, not news: counted at a fixed weight with confidence capped at 0.5.
Weekly move +4.0pts, implied probability now 0.185
Odds rose 4.0pts this week. More of this is bullish for gold, so the move counts as bullish.
Weekly move -2.0pts, implied probability now 0.115
Odds fell 2.0pts this week. More of this is bearish for gold, so the move counts as bullish.
Weekly move -4.0pts, implied probability now 0.45
Odds fell 4.0pts this week. More of this is bullish for gold, so the move counts as bearish.
Weight = impact × confidence × freshness. Impact (1–5) is how much the event can move gold; confidence (0–1) is how sure we are it's true and framed on the right side; freshness decays with a 5-trading-day half-life.
Astro items are capped by earned skill. An astrology item's confidence can never exceed the pillar's live Brier-derived skill. Today that's 19% pillar weight — the board enforces humility automatically.
One event, one side. Every item carries an event_key. If two claims about the same event land on opposite sides, the higher-weighted framing wins and the eviction is logged publicly above.
Update cadence. The board rebuilds every 6 hours from bot state, prediction markets, and the astro calendar. Pillar weights recalibrate every Sunday from the scoreboard. Intra-week regime breaks fire Telegram alerts immediately.
Nothing survives on reputation. Items retire below 0.4 weight. A claim from three weeks ago is gone unless re-registered with fresh evidence — which resets its timestamp and its accountability.
Not investment advice. This is a scoreboard of arguments, not a recommendation. The board has been wrong before and will be again — that record lives on the methodology page.