Platform

Price Forecasts: Monthly & Daily

A budget, a hedge and a trade entry all need the same thing: a defensible view of where a price is heading, one you can put in front of someone who will ask where it came from. The honest form of that view is a range, not a point — any single number implies a certainty no one has about a commodity price six months out. So the forecast pages give you both: a central path to anchor the plan, and a band around it that tells you how much room the plan needs. This page is about reading the band as carefully as the line.

Monthly or Daily: pick the grain for your decision

The grain toggle sits on every commodity's forecast page. Monthly is the strategic picture, cut to the rhythm of contract cycles and budget rounds; Daily is tactical timing, for the week where you have to decide whether to buy now or wait. Same commodity, same underlying engine, two resolutions — switch freely depending on the decision in front of you, and do not use one to judge the other.

Anatomy of the forecast chart

• Actual price history — what really happened, drawn from the market data itself.
• The model's fitted predictions — how the model has tracked that same history. This is the part most forecast charts leave out, and it is the part that lets you judge the fit with your own eyes before you look forward at all.
• The forward forecast, wrapped in a shaded confidence band — the central path plus the range around it.

The band is time-varying, not cosmetic. It widens the further out you look, and it widens when the market is volatile; it tightens when the market is calm. That behaviour is the platform being honest about rising uncertainty rather than drawing a flattering fixed margin around a line — and it is why the band, not the line, is the part you plan against.

A worked example, with illustrative figures rather than a real forecast: suppose the central path for next quarter sits at $100 with a 90% band of $92 to $112. You budget toward $112, because that is the cost you have to be able to absorb, and you treat drifts down toward $92 as buying opportunities rather than as evidence the forecast was wrong — both edges are inside what the model expected. Read the width the same way every time: a narrow band means the model can pin this market down and you can plan tightly against it, while a wide band is a standing instruction to hold extra cushion, stagger your buying, or both.

The tables under the chart

Two tables sit side by side below the chart. The history table lists the model's predicted values next to the actuals for the same periods, so you can check the fit yourself row by row instead of taking a smooth line on trust. The forecast table lists each future period with its low and high confidence bounds, which is the form you want when you are writing numbers into a budget. The Daily view adds a per-horizon accuracy readout, so the record for the exact lead time you rely on is on the same screen as the forecast itself — the fuller per-commodity picture lives in Forecast Accuracy & Why You Can Trust the Data.

Model versions you can see

A model-version badge tells you exactly which model produced the numbers on screen, and a version selector lets you load an earlier one and compare. That matters for a practical reason: if a newer model tells a meaningfully different story about the same market, you want to see that difference rather than discover it as an unexplained shift in next month's budget. Versions are visible by design, and nothing is swapped silently underneath a chart you have already shown to your board.

Context on the same page

A model's view is more useful next to the market's own. The forecast detail page also surfaces the futures chain — the market's forward curve of contract prices across delivery dates — alongside COT positioning, which shows how commercial hedgers and speculators are leaning, and a commodity news feed. When the model and the curve agree, you have a stronger case to take to an approval meeting. When they disagree, that gap is the thing worth understanding before you commit, and it is usually the most interesting thing on the page. Every term here is defined in the Analytics Glossary.

1. Pick your commodity — Open its forecast detail page, from the forecasts workspace or straight from your Favorites group.
2. Choose Monthly or Daily for the decision at hand — Budget round or annual contract, use Monthly. Deciding whether to buy this week, use Daily.
3. Read the centre line as the best estimate — It is the model's most likely path, and it is the number to quote — as long as you quote the range with it.
4. Plan against the 90% band, not the point — Take the upper edge as your worst case and the lower edge as your opportunity level. This single habit is what separates a plan that survives a bad quarter from one that has to be reopened.
5. Check the model-version badge — Note which version you are looking at, so that a later conversation about the same commodity starts from the same numbers.
6. Export the series if you need it in your own model — Pull the full forecast series out as CSV or Excel and drop it into the budget file or planning model you already work in.

Exporting forecast data

Export the full forecast series to CSV or Excel from any forecast detail page, and it lands in the tools you already use — the budget spreadsheet, the planning model, the board deck — with no manual re-keying and no transcription errors to hunt for later.