Methodology

For any artwork at any point in time, the final LAE is the sum of these three components. The result is a current estimated price with a confidence range — and, because the components are time-aware, an estimate for any historical point.

Most artworks transact once or not at all. A model that needs many observations per artwork has nothing to learn from. Pairs analysis on repeat sales sidesteps this.

Artist significance, retrospective inclusion, condition nuance — these matter to price but aren't in the data. Differencing techniques cancel out these latent factors when comparing repeat sales.

A Basquiat painting and a Basquiat drawing don't trend together. A linear model can't capture that. Tree-based models can — but only if the question they're solving is narrow enough.

Where we report year-by-year returns — such as an artist’s Annual Returns — we use the Compound Annual Return (CAR): the annualized, compounded rate of return realized across repeat sales of the same work. CAR is the same calculation as the compound annual growth rate (CAGR); we say “return” because it is measured from actual buy–sell pairs. Because it compounds rather than averaging each year’s change, it neither overstates nor understates performance across volatile years.

For an artist — or, for a held work, that work's artist cohort — the score reads a five-year window of auction activity together with the artist's trailing momentum: sell-through rate, buy-in rate, the split of lots selling above / within / below estimate, the sample size (lots offered), and 12-month price momentum. At the holding level it also reads compound annual return and the position's weight in the portfolio. Missing inputs lower the confidence rather than breaking the score.

The demand-and-liquidity axis is a weighted blend of whatever signals are present, re-normalised by the weights that actually fire: sell-through (weight 0.5, mapped 50%→90%), share sold above estimate (0.4, mapped 20%→60%), a below-estimate penalty (0.2), and a buy-in penalty (0.3, mapped 10%→40%). The question it answers: can this work be sold, and is demand beating the trade's own pricing?

Twelve-month momentum sets the trend: rising above +5%, falling below −5%, otherwise flat. A separate “peaking” flag fires above +25% — the recent-run-up-equals-correction-risk signal from the art-finance literature.

Market strength is crossed with trend, deliberately, so that “sell” and “watch” are not the same recommendation:

• Weak market (below 45) → Watch — illiquid; exit is hard, so monitor rather than act.
• Strong market (65+), peaking or falling → Sell — realise value while the work is still liquid.
• Strong market, rising → Accumulate.
• Strong market, flat → Hold.
• Moderate market (45–64): falling → Watch (don't sell into weakness); peaking → Sell; otherwise → Hold.

The crucial distinction: a strong but falling market says Sell (exit while you can); a moderate, falling market says Watch (selling now would be into weakness). That is why a deep, liquid name with flat momentum reads Hold, while a thinner one with falling momentum reads Watch.

The 0–100 composite used to rank holdings is an equal blend of market strength and a momentum-derived trend score (50 at flat, 100 at +20%, 0 at −20%), less a penalty for positions above 25% of portfolio value. Confidence is High when market data and momentum are both present over at least 30 lots, Medium above 10 lots, otherwise Low.

Every rule that fires emits a structured reason — a code, a severity, and the value that triggered it — rendered in plain language and sorted so cautions appear first. Because the thresholds are fixed and published, the “why” is always reconstructable from the inputs; nothing is generated by a language model. An optional AI-written narrative may be layered on top in future, but never in place of the rules.

The thresholds are a defensible first pass, not yet back-tested against realised sale outcomes. The artist-level read uses the artist's whole market; the per-work and per-holding versions narrow to the work's cohort and add return and concentration. The Historical View is a market read for professionals, not personalised investment advice.

We do not publish opinions in prose. The unit is a structured signal: a single sourced fact about an artist's market, tagged with its direction (bull or bear), how material it is to price, the horizon over which it plays out, a confidence level, how much of it is already reflected in past sale prices, how fresh the evidence is, and a one-line rationale linking the fact to value. A fact that does not map to a direction, a magnitude, and a horizon is treated as context, not signal. The result is a ledger in which every line can be audited on its own.

Signals fall into two families that are scored separately and never mixed. Directional signals move the bull/bear balance — the view on where price is heading. Structural signals describe price-discovery quality — whether the marked price can be trusted at all — and they only ever lower it; a favourable market structure is never counted as a reason to be bullish. The separation is deliberate. Illiquidity, concentrated ownership, or a contested authority are reasons to trust an observed price less, not reasons to be bearish on the artist — and conflating the two is the most common way an art-market model misleads.

Twelve axes carry the directional view. Each is bidirectional: depending on the evidence, the same axis can post a bull or a bear signal.

• Institutional validation — Major museum exhibitions, acquisitions, and biennial selection — the slow, durable signal that an artist is entering the canon.
• Gallery representation — The tier of gallery placing the work and any recent upgrade or downgrade in representation — a forward read on primary-market support, for living artists and active estates.
• Supply & estate — Estate discipline, foundation oversight, and catalogue raisonné status, weighed against supply overhang or estate dumping.
• Authentication & title — Whether a work is authenticated, catalogued, and free of attribution, title, or restitution disputes.
• Narrative momentum — Shifts in critical and media attention — rediscovery and of-the-moment heat versus fading coverage.
• Speculation risk — Signs of overheating, flipping, and unsustainable run-ups that historically precede corrections.
• Provenance — Distinguished, single-owner, or celebrity ownership history versus thin, murky, or gapped provenance.
• Primary–secondary gap — The relationship between primary-market asking prices and secondary-market results — a read on genuine underlying demand.
• Critical & scholarly standing — Monographs, serious scholarship, and canonical inclusion — the academic foundation beneath a market.
• Awards & honours — Major prizes, national-pavilion selection, and formal recognition.
• Reputational risk — Scandal, legal exposure, or controversy capable of moving a market sharply.
• Zeitgeist & cohort — Whether the artist sits in a category with structural tailwinds or headwinds — rediscovered women, regional cohorts, out-of-favour movements.

Eighteen axes describe ways the observed price itself becomes less trustworthy. They are impairments only — each can lower price-discovery quality but never raise it. They were identified one market archetype at a time, and a candidate is adopted only when it is a genuinely distinct mechanism that generalises beyond a single artist.

• Ownership concentration — A few holders control and defend supply, so observed prices may reflect a managed float rather than a broad market.
• Authentication regime — A contested, absent, or punitive authentication authority where the certifier's identity is itself price-determining.
• Market maturity — A young, thin, or volatile secondary market formed inside a single speculative window, where comparables are unreliable.
• Oeuvre heterogeneity — Output that splits into distinct media or series sub-markets, so an artist-level average misleads on any single work.
• Liquidity float — Canonical works are largely museum-locked, leaving a near-zero private float and decades-stale comparables.
• Geographic concentration — Demand and sale venues concentrated in one region, adding currency, regulatory, and regional-cycle risk and a narrow bidder pool.
• Settlement reliability — Reported hammer prices that may not correspond to settled, paid transactions, so headline results overstate liquid value.
• Brand & commercial dependence — A price level underwritten by an active commercial brand or hype engine rather than autonomous demand.
• Ideological provenance — Documented politics or history that permanently caps the buyer and display pool — a one-directional ceiling.
• Conservation fragility — Perishable, organic, or kinetic works that decay and raise a replacement-identity question when parts are remade.
• Attribution stability — An accepted corpus that itself shifts as scholarship and connoisseurship evolve, moving status and value by orders of magnitude.
• Mortality overhang — A known-ill or elderly living artist's supply-discontinuity uncertainty — distinct from the death event itself.
• Site-specificity & title — Works fused to non-owned property, where title is contested and value can be destroyed on removal.
• Signature-motif dependence — A market resting on one endlessly-repeated icon, raising fungibility and self-dilution risk.
• Authentication-destruction risk — An authority that can physically seize or destroy a rejected work, not merely withhold a certificate.
• Cast & edition authenticity — In reproducible cast or edition media, multiple legitimate originals coexist on a lifetime-to-posthumous hierarchy, so which cast dominates value.
• Diminished-capacity authorship — Genuine doubt about the artist's agency at the time of making, splitting an authentic-by-hand body into contested tiers.
• Digital-custody dependence — For digital-native work, value resting on off-the-artwork substrates — token binding, key custody, hosting, and file permanence.

Alongside the AI-gathered forward signals, the bull/bear case surfaces LiveArt's own market-structure indicators — the share of lots sold above or below estimate, the bought-in rate, sell-through, 12-month momentum and long-run CAGR — drawn directly from our settled-sales record over a five-year window. They appear as primary-source cards, badged "LiveArt market data", interleaved into and leading each bull/bear column. They are evidence only: because these same metrics already feed the quantitative disposition that anchors the Outlook, they are deliberately excluded from the Forward View tilt to avoid double-counting.

The structural impairments compound: each independent impairment multiplies the others down, so several mild impairments together can pull an artist's price-discovery quality well below half even when no single one is disqualifying. This one reading, expressed as a percentage, becomes the Structural Integrity dimension of the top-line below. A low reading is itself the flag — it tells you to weight the observed price, and any trend drawn from it, accordingly.

Two disciplines keep the layer additive and honest. First, a signal already absorbed into past sale prices is discounted, so the layer contributes only what the backward-looking quant has not already priced in — forward, not-yet-realised catalysts carry the most weight. Second, low-confidence signals are shown but excluded from scoring, and a catalyst whose date has passed is dropped from the forward view until its outcome is confirmed. Monetary figures appear only when sourced; unproven claims are explicitly hedged; market manipulation is never asserted as fact.

Direction blends the quant score with the net Forward View tilt into a single reading, then bands it:

• Accumulate — the weight of evidence favours adding to a position.
• Hold — the read is balanced or stable; no action is indicated.
• Reduce — the weight of evidence favours trimming.
• Monitor — conviction is too low to take a position; shown with the direction it leans, this is the honest “not yet known.”

These four labels are the fused top-line. They take the Historical View's own accumulate / hold / sell / watch as one of their inputs, alongside the Forward View tilt — so the top-line can lean against the raw quant signal when the forward evidence warrants it. On a single artwork the same Direction is expressed in the single-asset register — Buy / Hold / Sell — since you can't accumulate or reduce one object.

Conviction is how much trust to place in the Direction. It blends two things: Evidence — how deep and reliable the underlying market data is — and Agreement — how closely the backward-looking and forward-looking reads point the same way. We combine them so a shortfall in either tempers the result, and we surface disagreement rather than averaging it away: when the past and the forward view conflict, conviction falls even if each read is individually clear. Low conviction is what turns a Direction into a Monitor.

Structural Integrity asks a different question: can you actually act on the call? It is the price-discovery-quality reading from the qualitative layer, surfaced as its own dimension and deliberately kept out of Conviction. A museum-locked masterpiece can be a perfectly high-conviction Accumulate that is nonetheless hard to execute; folding executability into conviction would wrongly mute the view. When structural integrity is impaired, the top-line names the reason in plain language — museum-locked float, concentrated ownership, contested authentication, fragmented sub-markets, and the like.

The three dimensions resolve into one sentence — for example, “Hold · Moderate conviction · limited structural integrity (museum-locked float).” Where the forward qualitative read disagrees with the backward quant read, the divergence is shown explicitly as the headline insight, because a forward signal that contradicts the price chart is often the most decision-useful output the system produces.

Most artists do not yet carry a qualitative ledger. For them the top-line reads off the Historical View alone — no qualitative adjustment, full structural integrity assumed, conviction set by the strength of the market data — with no false precision added. The qualitative layer only ever sharpens a read where genuine, sourced signals exist.

Mean absolute error reported by artist tier, price bucket, medium, and region — not just a global headline number.

We check that an ±8% range actually contains roughly 80% of realized prices. Miscalibrated intervals are worse than wide ones.

Each model carries a version. Material changes ship with a changelog noting what shifted and why. Enterprise clients receive segment-level performance reports.