What 6 Months of Cannabis Price Data Reveals About Market Maturity

For six months, CannabisDealsUS has recorded the retail price of cannabis products sold across US dispensaries and rolled them into a weekly index. It is an unglamorous exercise: pull menus, normalise categories, average prices, publish. But a price series long enough to have a shape starts to say something the individual weeks cannot. Watched from a distance, the last two quarters describe a market settling down.

The headline everyone expects from a cannabis price tracker is a falling number. Prices in maturing cannabis markets do tend to soften over years as supply expands. That is not the interesting story in six months of weekly data. The interesting story is how the movement changed, even while the price level barely did.

Maturity Shows Up as Steadiness, Not as a Number Going Down

Normalised to a baseline of 100, the index opened the window at around 105 in early December and sat at roughly 104 at the end of June. Six months, and the headline figure went essentially nowhere. A casual reader would call that a boring, static market and move on.

The week-to-week picture tells the opposite story. In December and January the index swung hard: down seven per cent one week, up nearly ten the next, down almost nine the week after. By late May and June those swings had all but disappeared. Several consecutive weeks moved by hundredths of a per cent — 0.02, then 0.02, then 0.12 — and four of the final six weeks changed by less than a sixth of a per cent. The average price barely moved because the market had stopped lurching, not because nothing was happening.

Market maturity appears first as volatility compression: weekly price moves shrink even when the average price level barely changes.

That is the pattern worth naming. Early in a tracked market, aggregate prices jump around because the sample is thin and idiosyncratic — a few dispensaries running promotions, a category with too few products to average cleanly, one large retailer changing its menu. As coverage broadens and steadies, those individual shocks average out. The aggregate stops reacting to any single actor. A market whose weekly change routinely sits below one per cent is behaving very differently from one swinging several per cent a week, even when the shelf price is nearly identical.

The Biggest Move Was a Measurement Event

The largest single change in the entire six months was a drop of more than forty per cent in one week in February, with the measured average price falling from the high fifties to around thirty-one dollars. Read literally, that is a market collapse. It was nothing of the sort.

That week, several new low-priced headshop subcategories — scales, lighters, incense — entered the basket for the first time. Cheap accessories dragged the measured average down, then the series normalised over the following fortnight as the new categories settled into their own baselines. The index carried a low-sample flag on those weeks precisely because the system knew the composition had changed. The number moved because the ruler changed, not because the thing being measured did.

This is the part of running a price index that never makes the headline, and it is the part that matters most. A defensible index has to know the difference between the market moving and its own basket changing. Without that discipline, every re-categorisation, every new data source, every seasonal gap becomes a phantom trend that a careless reader will happily narrate as a crash or a boom.

A defensible price index must separate real market movement from sampling artefacts introduced when new product categories enter the basket.

I have written before about the messy reality underneath this kind of series — the twenty-plus state schemas and inconsistent menus that have to be reconciled before a single average can be trusted, in the ETL lessons from building a US cannabis price index. The February episode is that same problem viewed from the output end. Clean-looking numbers are the product of a great deal of unglamorous work deciding what belongs in the basket and flagging the weeks when the answer changed.

Why This Is a Data Problem, Not a Compliance Problem

Most conversations about cannabis as a business fixate on compliance. Licensing, testing, track-and-trace, packaging rules. That work is real and unavoidable, but compliance only decides whether an operator is allowed to sell. It says nothing about what to charge, and pricing is where margins are won or lost.

Pricing well requires comparable, trustworthy data across dispensaries, categories, and time. That is a data-engineering problem, not a legal one. An operator who can see that the market has stopped moving week to week is in a very different position from one reading a single noisy snapshot and reacting to it. The compression in this series is, among other things, a signal to stop treating every weekly wiggle as information. Most of the early swings were noise. The recent flatness is closer to the truth of the market.

Compliance decides whether an operator can sell; only comparable pricing data decides what they should charge.

This is why CannabisDealsUS exists as a data platform rather than a compliance tool, a decision I set out in why I run a cannabis data platform alongside an AI governance tool. The operators who will do well are the ones who treat pricing as a measurement discipline — who know their basket, trust their series, and can tell a real move from an artefact. Everyone else is pricing blind and calling the noise a strategy.

What Six Months Cannot Tell You Yet

A window this size has honest limits. Six months captures one turn of the seasons, not a full year of demand cycles. It cannot separate a maturing market from a quiet one, and it cannot yet distinguish structural price softening from a temporary plateau. Volatility compression is a strong signal, but a single stable quarter is not a trend until the next one confirms it.

What the series can already do is set a baseline and enforce a discipline. The next surprise in the data — and there will be one — can now be measured against a steady period rather than against noise. When the index next moves sharply, the first question will be the right one: did the market change, or did the basket? Six months in, that question has a place to be asked from. That, more than any single price, is what a maturing dataset buys you.