§ Validated record · Three cases

Three real zones. Frozen scores. Twelve months of recorded outcomes.

You read predictions against history, not against trust. Each card below: the score at a past as_of, the four signals that drove it, and what the zone actually did over the next year compared to its commune.

Case 01 · Lyon Métropole

Bron

South-east urban fringe, Lycée Jean-Paul-Sartre catchment

As ofSeptember 2023ObservedSeptember 2024H3 id881f902535fffff
Score
71
Strong outperformer
/ 100 calibrated · isotonic
12-mo excess
+13.8pp
Zone +6.7 % / commune -7.1 %
What the model saw
  • Above-commune household income (+€2,300 mean per consumption unit)
  • Six schools within 800 m, three of them private — family-buyer magnet
  • Apartment median still 18 % below the Bron commune mean — room to compress
  • PLU zone class U with cadastre footprint share 41 % — actual headroom for in-fill
Top SHAP drivers
  • Income vs commune+0.14
  • Schools within 1 km+0.08
  • Apt price below commune+0.06
  • Building stock older-0.04
What it would save you

Most family-office RE managers read Lyon Métropole's −7 % YoY headline and stayed away. The Plotix score said Bron was the exception — twelve months later, +13.8 pp of out-performance compounding inside the same metro.

Case 02 · Métropole d'Aix-Marseille-Provence

Marseille 1er

Saint-Charles station corridor, around place Sadi-Carnot

As ofMarch 2024ObservedMarch 2025H3 id8839680643fffff
Score
64
Above commune
/ 100 calibrated · isotonic
12-mo excess
+9.2pp
Zone +4.1 % / commune -5.1 %
What the model saw
  • Transit density: 17 stops < 500 m (metro, tram, regional rail) — top-quartile in AMP
  • Encadrement des loyers active since August 2022 — predictable rental yield ceiling
  • 47 % cell area in U-zone PLU, 0 % in A or N — every parcel is buildable
  • 230 k+ active businesses indexed within Marseille — economic anchor
Top SHAP drivers
  • Transit accessibility vs commune+0.12
  • PLU buildable share+0.07
  • BPE health / commerce density+0.05
  • Cat-Nat risk (sea-flood proximity)-0.03
What it would save you

Marseille's headline market dropped ~5 % YoY across the AMP. Plotix flagged that Saint-Charles still had the transit + encadrement-predictable-yield combo investors actually pay for. A buyer's agent could have closed two mandates with this single screen.

Case 03 · Toulouse Métropole

Toulouse

Capitole — historical centre, around place Wilson

As ofJune 2024ObservedJune 2025H3 id8839601959fffff
Score
59
Above commune
/ 100 calibrated · isotonic
12-mo excess
+5.7pp
Zone +8.4 % / commune +2.7 %
What the model saw
  • 11 km of road network, 76 major-axis segments — peak central accessibility
  • Density: 717 active businesses per H3 cell, 717 m² mean building footprint
  • U-zone share 91 %, AU 0 %, A/N 9 % — full urban consolidation, no more land
  • DPE coverage 95 % with 38 % F+G share — regulatory pressure builds on owners
Top SHAP drivers
  • Road network density vs commune+0.11
  • Income + 30-44 cohort share+0.09
  • Energy F+G share-0.05
What it would save you

A premium-agency seller could defend the Capitole's pricing within Toulouse without hand-waving. A buyer's agent could explain why the F+G share is the alpha here — owners under pressure to renovate or sell into a constrained-supply zone.

§ Method

How we built these.

How are these cases validated?
Each case is a real H3 cell. We scored it at an as_of in the past, then froze the score. Twelve months of DVF transactions later, we read the realised median €/m² for that hex AND the commune it sits in. Excess return is zone YoY minus commune YoY. No revisions, no cherry-picked windows.
Are these cases hand-picked?
Yes — one per metro, chosen to span the score range (71 / 64 / 59). The model produces ranked picks for every commune; we are not claiming every top-30 % pick lands +13 pp. The aggregate backtest is on the methodology page — 65.6 % hit rate, +3.52 pp realised lift.
Can I see the live score for each of these?
Open the live map and search the address — Bron / Saint-Charles / Capitole. The current score reflects today's data and may have drifted from the case-study as_of. That is the point: every score is point-in-time, and history is recorded.
§ Begin

See if your next deal sits in the top 30 % of its commune.

Full Team plan free for fifteen days. No card on file. Decide on day 16.

Full Team plan · No card · No auto-renew · Cancel any day