About the data

This tool does not use data from, and will not produce accurate results for certain portfolios and fee structures including growth equity funds, funds that focus on fast distributions to LPs, and fund-of-funds that invest in many different venture funds such as AngelList's broadly indexed managed funds. More recent funds with an effective duration of shorter than three years may see substantial near-term volatility in their performance percentile.

Data was last updated and is valid up to September 10, 2024.

Learn more on the AngelList blog
Start investing in venture funds
Host your fund on AngelList

Methodology

Formally, effective duration is defined as the amount of time a given IRR would need to be compounded to produce the given TVPI.

The fund percentile calculator is a monotone bilinear interpolation, consisting of a two-dimensional matrix of interpolating knots. Creating the function is equivalent to fitting the set of interpolating knots with monotone values.

Our input points were benchmark funds and fund-equivalents, i.e., the combined performance net to LPs for AngelList syndicate leads that have made at least ten investments but do not operate microfunds. Median benchmarks are created from the most recent finalized PitchBook venture-capital benchmark report by taking the median net IRR and median net TVPI from relevant fund-vintage cohorts (for all funds reporting both values). A final benchmark point is the performance of the AngelList platform. For the funds, we removed any fund with effective duration of less than one year or greater than eight years. We also removed certain external funds at our discretion that appeared to have inaccurately high self-reported IRR or TVPI numbers. This left us with 1125 total input points: 1113 funds and fund-equivalents and 12 benchmark points.

We began by calculating the rolling median and standard deviation of fund performance for a uniform window of six months before and after. Based on these values we used a normal distribution (centered at the median) to produce initial percentiles for each input point. We then projected this solution onto our space of monotone interpolating knots by assigning values to the set of interpolation knots to minimize the sum of squared error from the implied percentile scores to their initial percentile scores. For our benchmark points, we set their target percentile at 50 and weighted projection error 50 times higher than for our fund points. Finally, in our optimization we required our interpolating knots be monotone non-decreasing in net TVPI and monotone non-increasing in effective duration to produce an overall monotone function.

”Top-Tier VCs” and ”Other VCs” are defined by AngelList based on internal assessment of each fund's industry reputation. Full list available upon request (NDA required).