Observed Demand Index methodology
The IDO sums up, as a percentile from 0 to 100, how much market activity that has already happened a set has compared with the rest of the live sets. It does not represent intrinsic value, quality or a price forecast: it is not a prediction or buying advice — it is a dated snapshot of the observed buying pressure.
The variables (observed facts only)
- Amazon price events (90 days). Every real price change observed is a market event. Source: synced history (includes Keepa).
- Multi-shop stock transitions (90 days). Sell-outs and restocks observed in the shops in Spain that we follow: how fast the market empties.
- Set followers. Interest in buying later, with real friction (following on Telegram or in the account).
- Shops with fresh stock (≤7 days). Market depth: in how many observed shops it is available now.
We leave out on purpose: the price itself (it is the result of supply meeting demand, not the demand), the number of reviews (it arrives weeks late) and LEGO.com’s binary stock (noise that the supply chain can manipulate).
How they are combined: zero arbitrary weights
Nobody decides “by hand” how much each variable weighs. After taking the logarithm, trimming outliers (percentiles 5-95) and converting to standard scores, the weights are the first principal component (PCA) of the covariance matrix: pure algebra on the data. The result is mapped to strict percentiles against the universe of live sets.
| Variable | PCA weight (computed on October 28, 2026) |
|---|---|
| Amazon price events (90 days) | 0.5499 |
| Multi-shop stock transitions (90 days) | 0.6079 |
| Set followers | 0 |
| Shops with fresh stock (≤7 days) | 0.5728 |
Joint variance explained by the index: 84% · universe: 987 live sets · this table is regenerated with each computation — what you see is what scores the set pages.
How to read it
- 0–20: Marginal demand.
- 20–40: Below-par demand.
- 40–60: Median demand.
- 60–80: High demand.
- 80–100: Exceptional demand (top 20% in observed buying pressure).
Limitations (read them)
- It does not predict. A high IDO today does not guarantee resale value or availability tomorrow.
- No observed activity: if a set has no price event, stock change, follower or shop with stock in the window, we do not give it a percentile: the set page says “no observed activity” instead of a number. Sets with the same activity share a percentile.
- Cold start: sets that have just arrived have little observed trace and their percentile is less reliable for the first few weeks.
- 90-day window: older activity does not count; a “quiet” classic can score low while having steady demand.
- Latency: the sources consolidate with a 24-48h lag; the computation date is visible in each module.
How it fits our promise
This site does not publish ROI or resale value projections — and the IDO is not one: it describes what has already happened, with the same dated sources as the rest of the set page (general methodology). We are affiliates: if you buy through our links we receive a commission at no extra cost to you; the IDO is computed the same way whether or not the set has an affiliate link.