Proprietary · EK Conviction Engine

The VC Map.

One model behind QuantEstate and Qubitline. It turns filings, momentum and financials into a 0–100 conviction score, then runs 2,000 Monte Carlo paths to forecast three-year revenue.

Conviction vs growth

045701000%50%100%150%Ridgeline Compute CampusCatskill Hydro DCVoltgrid StorageHudson Edge ColoGreenwich Hill Parcel 14Tensorline LabsMontclair 6-unitLedgerwise PaymentsPrinceton Estate LotStamford Quantum Co

Bubble size = median 3-yr revenue · gold band = Pursue (70+)

DealSectorGrowthMarginP50 yr 3ConvictionCall
Ridgeline Compute CampusData centers90%32%$77.4M88Pursue
Catskill Hydro DCData centers110%18%$13.8M80.4Pursue
Voltgrid StorageEnergy130%-15%$16.2M73.1Pursue
Hudson Edge ColoData centers55%21%$18.8M69.1Watch
Greenwich Hill Parcel 14Land12%60%$518K63.4Watch
Tensorline LabsAI infra150%-40%$6.8M59.1Watch
Montclair 6-unitMultifamily6%45%$355K58.9Watch
Ledgerwise PaymentsFintech35%8%$18.7M37.8Pass
Princeton Estate LotResidential RE0%0%$0K27.3Pass
Stamford Quantum CoAI infra60%-70%$549K5.9Pass

How the model works

  1. 01 · Signals. Public filings, liens, permits, hiring, news and founder footprint become six normalised features.
  2. 02 · Conviction. A weighted logistic model converts them to a 0–100 score: Pursue 70+, Watch 45–70, Pass below 45.
  3. 03 · Forecast. 2,000 seeded Monte Carlo revenue paths give P10/P50/P90 and the odds of doubling.
  4. 04 · Learning. Every closed or passed deal recalibrates the weights — the data moat compounds.

Current weights

  • growth+2.2
  • margin+1.4
  • momentum+1.6
  • founder+1.2
  • valuation+1
  • risk-1.8

Sample pipeline is illustrative and synthetic. Weights are a prototype prior, to be fitted on closed-deal outcomes.