Insights

The industry-standard age-curve model looks reasonable in-sample -- and fails the same overfitting test quant funds run before trusting a strategy with real capital.

One aggregate verdict hides real variation. A published, fund-group breakdown of held-out accuracy, and why two specific groups deserve extra caution.

TVPI and IRR alone don't say whether a fund beat public markets over the same dated cashflows. KS-PME and Direct Alpha do -- here's what they are and how we compute them.

Our published Buyout accuracy blends ages 4-9. Split out just the youngest slice -- the age band with the least realized track record, and the one most secondaries deals actually happen in -- and both models get measurably harder to trust.

Secondaries intermediaries publish an average Buyout clearing price twice a year. We priced the same snapshot at the youngest ages and lined our two models up against it.

A FTSE 250 PE vehicle disclosed the exact price it sold young Buyout fund stakes for. Here's how that real transaction compares to the published survey and to our own models.

There isn't one accuracy number -- there are four, one per pricer, each earned a different way. Out-of-sample verdicts, AEPC, Welch's t-tests on whether one model is actually statistically more accurate than another (per-fund and portfolio-level), and the calibration story behind Equisect Ledger.

A real pricing edge and a well-shaped projection are different claims. Four path- and distribution-fidelity checks, scored across all five models on the same held-out funds -- and no model wins on all four.