Reyna is the desk's commodities and energy specialist. The model ingests live data from 64 physical supply nodes — crude oil storage at Cushing and Rotterdam, LNG terminals in Sabine Pass and Qatar, copper smelters in Chile and China, grain silos across the Corn Belt and Mato Grosso — and pairs them with weather forecasts, shipping AIS data, refinery utilization, and inventory releases from the EIA, USDA, and their global analogues.
What makes commodities different from the rest of the desk: they are physical. A barrel of oil has to be somewhere. A bushel of corn has to come from a field. A ton of copper has to pass through a port. Reyna is built around that physicality — the model thinks in terms of storage, flow, and disruption, not narrative.
Accuracy sits at 68.7%, a little below the desk average. The reason is honest: commodity markets are dominated by low-probability, high-magnitude events — a pipeline freeze, a hurricane, an OPEC surprise — and Reyna can only model the base rate of those events, not predict them individually. The model is calibrated to be prepared rather than to predict.
“Polar vortex forecasts 9 days out showed high-confidence severe cold across Plains + Upper Midwest. Storage drawdown would overshoot consensus. Positioned before the freeze.”
“Chinese smelter utilization dropped to 71%, lowest since 2016. Treatment charges collapsing. Supply tightness wasn't priced. Positioned before spot caught up.”
“Cushing inventories building faster than seasonal norm. US production flat. Demand indicators weak on 3-week basis. Front-month overheld by spec longs.”
“Black Sea tension premium looked underpriced heading into spring planting window. Misread the hedging flow — it was selling, not buying.”
“Summer driving season approach. Refinery maintenance heavier than usual. Gulf Coast utilization at 88%. Crack spreads should expand into Memorial Day.”
“Central bank buying + real yields turning. Thesis correct on both catalysts but ETF outflows capped the run. Direction right, magnitude moderate.”
WTI physical hub inventory at 25.1M barrels — above 5-year seasonal average. Building faster than demand can absorb.
Utilization rate at 71% — lowest since 2016. Treatment charges collapsing. Copper concentrate tightness signal.
Working gas in storage at 112% of 5-year average. Warm winter, strong production, weak industrial demand. Bearish setup.
USDA reports 34% planted vs. 22% 5-year seasonal. Weather window opened early. Yield risk skewed lower if frost returns.
ECMWF and GFS converging on a low-pressure system crossing North Sea platforms next week. Wind energy output will spike, then crash. Offshore oil loadings may delay 48-72 hours.
Early planting means more acres exposed to late frost. 10-day ECMWF shows 40% probability of sub-freezing overnight lows in Iowa and Illinois corridors around May 3.
Dry window improving but still 6% behind seasonal pace. Second-crop corn planting delayed. Will start mattering for July–September supply outlook if weather stays unfavorable.
Reyna is built around the insight that commodity markets are unusual: they are driven by physical constraints that equity and bond markets don't have. A barrel of oil must be produced, transported, stored, refined, and consumed — and at each step, there are tanks, pipelines, refineries, and terminals with real capacity limits. The model's core task is to maintain a live picture of where the global supply chain has slack and where it doesn't.
The ingestion is wide and external: NOAA weather data, ECMWF medium-range forecasts, USDA crop progress reports, EIA petroleum and gas reports, LNG cargo tracking via AIS, satellite-derived refinery utilization, Chinese smelter activity, Australian iron ore shipping manifests, and the CFTC Commitment of Traders report. Reyna cross-references these with price and curve structure to identify dislocations between physical reality and futures pricing.
Reyna publishes supply-side disruptions, inventory anomalies, weather catalysts, and positioning extremes. Every call identifies the specific physical signal, the cash vs. futures spread implied, and the event that would resolve the trade.
Reyna does not forecast OPEC decisions, major wars, or sanctions. These are political events outside the model's scope, and the model is calibrated to detect them after they happen — through physical flow anomalies — not before. The accuracy reflects that honest limitation.
OPEC+ surprises, sanctions, wars. Reyna cannot predict these — only respond to the physical aftermath. Calls made within 72 hours of a geopolitical event should be re-evaluated.
USDA reports weekly, EIA weekly, China monthly. A commodity can move significantly between releases. Reyna's read is only as fresh as its slowest feed.
When ECMWF and GFS disagree by more than 48 hours on a storm track, Reyna explicitly lowers confidence. Weather-driven calls require model consensus.
// FREE · DAILY · DELIVERED BEFORE MARKET OPEN