MONITOR
| TICKER | LAST | CHG | 12M FCST | VS NO‑CHG |
|---|
SECURITY
UNIVERSE
Any series. Any country. Any contract. Type a name on the command line, or browse here; the first lookup fetches the history and the forecast, the next reader gets the cache.
SERIES
Pick a country or a commodity group.
| TICKER | SERIES | UNIT |
|---|
Cross-indicator correlation
Pearson r — column observed N months after row.
−1
+1
gray = insufficient overlap
Forecast accuracy by horizon
MAE at h=1 to h=N from rolling backtests. Flat = stable; rising = noise compounds with horizon.
Backtest data still warming up — accuracy is computed on the same 14-day cycle as forecasts.
80% interval coverage
When the model says "80% chance the value falls in this band," does it actually? Bars = empirical coverage from rolling backtests. Reference line = the honest target.
No coverage data yet.
well calibrated · ≤ ±5pp
off · ±5–15pp
miscalibrated · > ±15pp
Backtest replay
Scrub a cutoff. See what the model said from there vs what actually happened. Solid = history + actuals; dashed = forecast; shaded = 80% band.
No backtest data yet — accuracy refresh runs on the same 14-day cycle as forecasts.
Indicator comparison
Indexed to 100 at start of window. Solid = history, dashed = forecast. Hover for raw values.
Pick up to 3 indicators above to overlay them on one normalized chart.