Comment Cault estime la valeur de vos LEGO — méthodologie complète

Mechanics

Why LEGO gains in value

The underlying mechanics are simple: supply is limited and eventually dries up.

While a set is in production, it's available everywhere at retail price, and its second-hand value stays close to that price. The day LEGO stops production, supply freezes. As boxes are opened, built or damaged, sealed new copies become scarce — and collectors start searching for them.

Historically, retired sets have appreciated at an average of about 8 to 11% per year, a pace that has sometimes exceeded that of classic assets. But this average hides enormous differences: some sets explode, others stagnate. The whole challenge of our model is to distinguish the two.

The key point

As long as a set is still on sale at LEGO, its price doesn't take off. The gain only starts at retirement. Our model enforces this strictly: a newly released set is estimated at its retail price, no more. If you see a new set selling above retail, that's transient speculation — not a fundamental appreciation.

Depreciation risk during production. Conversely, a set in production can slightly lose value — clearance sales, promotions, quiet end-of-life. The projection for a set still on sale is therefore asymmetric: the low range goes down to −25%, the high range barely reaches +2%. Nothing comparable to post-retirement appreciation.

Announced retirement date. As soon as a set's retirement date is official, our model anticipates the spike: it projects a flat plateau until that date, then the retirement jump and the structural curve beyond. Example: a set retiring in two months will have its projection incorporate the expected +25% jump, whereas a set with no known date stays flat.

Trajectory

A three-phase curve

The key moment is the set's retirement. Once it runs out in stores, the market shifts to resellers and the price jumps — often 20 to 30%. This is the “retirement spike”. Then the value follows three stages:

The spike — immediate jump when production stops (sometimes followed by a small correction when the initial rush fades).
Rapid appreciation — for 6 to 24 months, value rises quickly as new stock disappears.
Stabilisation — growth slows and settles into a steady, more predictable pace.

This is the chart at the top of the article: flat in stores, then a rise that accelerates then slows towards a cruising pace.

Under the hood · the curve formula

V(τ) = P₀ · (1+π) · exp[ g∞·τ + (g₀−g∞)·λ·(1−e^(−τ/λ)) ]

To read: P₀ is the retail price, (1+π) the retirement spike (π ≈ 0.25). In the exponential, a fast early rate g₀ relaxes to a cruising rate g∞ (≈ 8%/year) over a duration λ. τ = years since retirement. This formula only activates after retirement: before that, V = P₀, full stop.

Context

A set doesn't live in isolation

The curve above is the set's own trajectory. But its real price, month by month, also moves with what surrounds it. We therefore decompose each movement into four ingredients:

The global LEGO market. When the whole second-hand market rises or cools, your sets follow. We measure this "general mood" with an index built across all sets.
Theme hype. A theme can be "hot" at a given moment. Important: we don't decide a theme is trendy — we measure it in the recent co-movement of its prices. No theme is favoured in advance.
Set-specific hype. A temporary hype (a sought-after minifig, a fashion effect) that inflates the price then fades.
Noise. The irreducible part, the market's randomness.

own trajectoryglobal marketthemehypetotal movement

Example: a month where the price gains +1.8%, broken down by its different causes.

Under the hood · the decomposition

Δx_month = μ(structural) + β_mkt · f_mkt(market) + β_thm · f_thm(theme) + η(hype) + ε(noise)

To read: the monthly price change (in log) is the sum of the set's own drift μ, its market sensitivity β_mkt multiplied by the market movement f_mkt, the same for its theme, hype η (transient, reverting to zero) and noise ε. The market and theme factors are estimated from data, never fixed by hand.

Factors

What sets a good set apart

Beyond context, certain intrinsic characteristics make a lasting difference:

Theme and licence. Strong licences — Star Wars, Harry Potter, Lord of the Rings, Marvel — appreciate more strongly and durably than house themes.
Exclusive minifigs. A huge lever: a minifigure found only in one set can pull the entire value upwards.
Size — surprisingly. Small sets and very large ones outperform medium sizes.
Exclusivity. A limited edition set already starts above retail price.
Re-release risk. Risk factor #1: if LEGO re-releases the model, the original can drop suddenly. A set that is itself a re-release often appreciates less.

“Mr. Gold, originally sold for under €3, trades today at around €2,000.”
A single exclusive minifig can dominate the value of an entire set.

Under the hood · re-release risk

E[V(τ)] = V_base(τ) · [ 1 − d · (1 − e^(−h·τ)) ]

To read: we treat re-release as a lurking risk, not slower growth. d ≈ 0.30 is the drop magnitude (≈ −30%) and h is the risk of a successor appearing. Real example: the 2007 UCS Millennium Falcon lost nearly 30% when a successor was rumoured.

Minifigs

A separate model for minifigures

Standalone minifigs follow a different logic from sets. What matters here is exclusivity — the number of sets it has appeared in — and the production status of those sets.

In production (one parent set still on sale) — +2%/year. As long as it can be bought in a new set, its value stays contained.
Exclusive (present in a single retired set) — +10 to +15%/year. Buyers who want this minifig have no other option — supply dries up quickly.
Rare (2 or 3 sets, all retired) — +8 to +13%/year. Slightly less exclusive, but still rare.
Common (4+ sets) — +1%/year roughly. Remains accessible through multiple channels, appreciation is modest.
Re-used in a new set — Stagnation or slight decline. The arrival of a new set containing it restarts supply.

“An exclusive minifig can pull the entire value of a set upwards.”
That's why minifigs are analysed separately, with their own rules.

Under the hood · rates by minifig category

Exclusive (1 retired set)     :  g_min = 10 %, g_max = 15 %  → avg. 12,5 %/yr
Rare     (2–3 retired sets)   :  g_min =  8 %, g_max = 13 %  → avg. 10,5 %/yr
Common   (4+ sets)            :  g_min =  0,5 %, g_max =  1,5 % → avg. 1 %/yr
In production                 :  g_min =  1 %, g_max =  3 %  → avg.  2 %/yr
Re-release announced            :  g_min = −5 %, g_max =  0 %  → stagnation

The confidence range is derived from the min–max range of each category. The rate only activates at the high level once all parent sets are retired.

Source fusion

Multiple marketplaces, one truth

Marketplaces don't speak the same language. A listing (“asking price”) is often inflated; a completed sale better reflects the market; some include fees, others don't; conditions vary.

Rather than mixing everything together, we posit that there is a single “true” value at any given moment, which each source observes through its own filter — a bias and some noise. Like several imperfect sensors which, together, give a more reliable reading than any single one.

Under the hood · the measurement model

Our engine estimates each source's bias instead of suffering it, then reconstructs the true value. eBay sales serve as the anchor; estimates from other aggregators are used only to verify our results, never to feed them.

Uncertainty

Why a range, never a single number

The market is volatile. Announcing a single price would give false precision. We display a range — low, median, high — and it widens the further out we look. We project up to 5 years: beyond that, uncertainty becomes too large to be useful.

The further the forecast, the wider the range.

Under the hood · where uncertainty comes from

We don't just account for market noise. Our range adds up four sources of uncertainty:

›uncertainty about the set's parameters (we never know its curve perfectly);
›market and theme randomness (they can turn);
›unpredictable hype;
›the re-release risk, which can cause a sudden price drop.

To combine these, we simulate thousands of possible future scenarios and read the range from their spread. The annual volatility of a set is around 25% — hence naturally wide ranges.

Collection

Valuing an entire portfolio

For an entire collection, we don't simply add up estimates. Sets from the same theme, or exposed to the same market, rise and fall together — which is exactly what our market and theme factors capture.

Our thousands of scenarios therefore move correlated sets together: the collection range accounts for this, instead of underestimating risk through a simple sum. You see the total value, its breakdown by theme, and the difference between the “on paper” value and what you would actually receive on resale, net of fees.

Honesty

What our model can't do

The secondary market is illiquid: a displayed price is not a guaranteed sale.
Volatility is high: our ranges are wide because reality is.
LEGO can re-release almost anything, which remains unpredictable.
Box condition changes everything: sealed is worth much more than open.
We project up to 5 years; beyond that, we prefer not to venture.
Parameters are global averages: theme-by-theme calibration (Star Wars more volatile than City) is in development. Current rates are starting points, not constants.
For minifigs, exclusivity is estimated from your own collection. A global BrickLink count will refine these rates in a future version.

Warning

These estimates do not constitute investment advice. They are statistical projections, useful for orientation and informed decision-making, but carrying an irreducible share of uncertainty. Do your own research before any purchase or sale.

Estimates recalculated each month from the most recent sales.
The more history a set has, the more precise the estimate; for very recent sets, we rely on the past behaviour of their theme.

How we estimate the value of your LEGO