Bayesian Private Equity Index from Capital Cash Flows

Everysk

Finance & Insurance
PyMC

The Challenge

Private equity and venture capital sit in a measurement blind spot. Unlike public stocks, there's no ticker, no daily close, no transaction trail to follow. Funds report returns infrequently, and the only hard data available are the capital flows themselves — money in, money out, and long silences in between.

Everysk wanted to answer a deceptively simple question: can you reconstruct what a PE or VC portfolio actually returned, using nothing but those cash flows? And could the answer hold up against the benchmarks that institutional investors already trust?

Our Approach

We started where any good analysis starts — reading the academic literature and studying the data Everysk had already assembled. A week of structured exploration revealed the signal buried in capital flow timing and magnitude.

From there, we built a model that treats unobserved fund returns as a statistical inference problem. The model captures how investment exposure and value creation shift over time, estimating a full return series from the pattern of contributions and distributions alone. We went beyond the baseline methods in the literature, refining the model's structure until the estimates were both stable and interpretable — a hard requirement when your audience is investment professionals, not statisticians.

The work was deeply collaborative. Everysk's team brought domain expertise on fund mechanics; we brought the statistical framework. Each round of results was interpreted together, with the model evolving through that dialogue.

Results

The final model produced a venture capital return index estimated entirely from capital flow data.

When we laid it against the Cambridge Associates VC Index — the industry's most widely cited benchmark — the two tracked closely, validating that the approach was capturing real signal rather than fitting noise. The cumulative return series now sits alongside US public equity returns, giving Everysk a defensible, transparent view into a market that has historically resisted measurement.

PyMC Labs Team

  • Ravin Kumar

More Work in Finance & Insurance

VisualVest

Developed a probabilistic model to forecast customer lifetime value for a robo-investing platform, where off-the-shelf approaches couldn't account for variable, AUM-based revenue.

Customer Lifetime ValueShifted Beta-GeometricSurvival ModelsBayesian GLM

Alva Labs

Built psychometric and validity models that substantially outperformed Alva Labs' existing assessment system, giving them both better trait estimates and hard evidence that their assessments predict job performance.

Item Response TheoryPsychometric ModelingHierarchical Bayesian ModelingOrdinal Regression

HelloFresh

Rebuilt HelloFresh's marketing measurement and testing infrastructure, cutting model runtimes from 20 minutes to 2 and overnight test pipelines from 6 hours to 6 minutes across 15+ global markets.

MMMGaussian ProcessesHSGPA/B Testing

Let's Chat, We Respond Fast

Tell us about your problem. We typically respond within 24 hours.

Schedule a Consultation