A decision-support tool for swing positions in gold-linked and anchored stocks. It answers two questions: when is a stock unusually cheap against the thing that drives it (a buy zone), and when has it run unusually far ahead (a sell zone). Everything on it was tested against up to 20 years of history before it earned a place; the failed ideas are documented in the project repository and deliberately absent here.
Every stock here has an anchor — the asset that statistically drives it (gold for the miners and royalty companies, bitcoin for COIN, the uranium fund URA for UEC, sector funds for MET and ZION). The z-score measures how far the stock has drifted from its normal relationship with that anchor, in standard deviations, using the last 250 trading days.
Gold's stretch above its own 200-day average, ranked against 20 years of history. This is the one gold-wide warning that tested well: readings at or above the 99th percentile (118 days in 20 years) preceded flat-to-negative gold over the following 1–6 months. The January 2026 top scored 100.0 on this gauge on its exact peak day.
| Element | Meaning |
|---|---|
| Green / red bands | Periods the stock was in its buy / sell zone |
| Solid dark line at 0% | The zero line on family charts — above it, up since the window began |
| Dashed red line + date | All-time high (drawn when within reach; always stated in the card text) |
| Thin dashed lines "high / low" | The highest and lowest close within the visible range |
| Dashed line "your stop" | 12% below your average cost, once you've entered a position |
| 1d · 5d · 2wk ranges | 30-minute bars, refreshed at each server run |
| 1mo – 1y ranges | Daily closing prices |
All choices persist in this browser and affect nobody else.
Two ways to track a holding, both stored only in your browser — never on the server:
The ledger uses average-cost accounting: buys raise your share count and blend your average cost; a sell realizes gain or loss against that average. Cards show shares, average cost, value, unrealized and realized P&L; the portfolio card at the top sums every position, with realized gains broken out.
Delivered by the free ntfy app. Alerts fire on changes, never repeat daily, so silence means all quiet.
| Alert | When |
|---|---|
| 📈 Buy zone | Any watched stock enters its buy zone |
| 📉 Sell zone | A stock you hold enters its sell zone |
| ⚡ Provisional (~3:30pm ET) | A zone change looks likely on intraday prices — the 4:30 run confirms or cancels |
| 👀 Gold closing in | Gold stretch crosses the 95th percentile |
| 🚨 Gold froth | Crosses the 99th — the January-2026-top zone |
| 🚨 Stop breach | Owner's position breaks its stop (checked every 5 minutes) |
| ⚠️ Pipeline problem | The data feed broke — don't trust the page until cleared |
| ✅ Monthly heartbeat | 1st of the month: proof the system is alive |
The public topic shown on the dashboard carries market signals only — never anyone's positions. Subscribe in the ntfy app with that topic name.
Everything above survived adversarial backtesting on 2006–2026 data; these did not, and were removed: a six-method composite scoring system (lost to buy-and-hold in 24 of 24 configurations), mechanical stop-losses as a signal (cost 255 points of return; kept only as disaster insurance), a sell-and-rebuy "ratchet" (0 of 108 configurations), moving-average trend gates and macro-regime gates (predicted nothing, or the opposite). One honest limit applies to all of it: the evidence comes from a period in which gold mostly rose. Signals are decision support — the decisions, and the results, are yours.