Data sourced from Dakota Private Markets, the private fund performance platform powered by Dakota. Learn More | Request Access
A fund can land in the top quartile of its vintage-year cohort and still trail most of the ten funds its investors actually weigh it against. The cohort number is accurate. It just answers a different question than the one on the table in an investment committee meeting.
That gap is the case for building a custom peer group. Instead of accepting a pre-built, anonymized cohort, you define the comparison yourself and see exactly which named funds sit inside it. Dakota Private Markets covers 18,000+ named funds, so the peer group can be as broad or as narrow as the question requires.
Why an anonymized cohort answers the wrong question
Pooled benchmarks from anonymized providers answer a general question: where does this fund sit relative to its vintage-year cohort? That is a useful directional signal.
It cannot answer the specific question: where does this fund sit relative to the 10 named funds I consider its real competitive set? You cannot see who is in the pool, so you cannot test whether the comparison is fair, and you cannot adjust it when it is not.
An aggregate median creates a second problem. It collapses a whole group into one number, which hides whether a fund is clustered with its peers or sitting far from the rest.
Step 1: Start with the question, not the filters
Before touching a filter, write down what you are trying to decide. A consultant underwriting a re-up, an allocator building a shortlist, and a GP preparing for an LP meeting each need a different peer group, even for the same fund.
A useful test: could you name five funds you would expect to see in the group? If you can, you already have the intuition the filters need to encode. If you cannot, start with a broader screen and narrow from there.
A custom benchmark on Dakota Private Markets is defined by four filters, used in any combination:
|
Dimension
|
What it controls for
|
|
Strategy
|
Return profiles and risk differ by strategy, so buyout is not compared with credit or growth
|
|
Vintage year
|
Funds raised into the same market conditions share a comparable period
|
|
Geography
|
Regional deal markets and exit environments shape outcomes
|
|
Fund size
|
A small fund and a large fund of the same strategy face different opportunity sets
|
Each filter you add tightens the group, and each one you remove widens it. That is the point of building it yourself: the benchmark is only as broad or as narrow as your question.
Step 3: Review the named funds before the numbers
This step is the one an anonymized cohort cannot offer. Because every fund in the database is individual and named, you can read the actual list of funds in your peer group before looking at a single return figure.
Check that the group passes a sanity test. Are the funds ones a reasonable investor would compare? Is anything missing that you would expect to see? Is there an outlier that does not belong? Adjust the filters until the list reflects the real competitive set.
Step 4: Compare fund by fund on IRR, TVPI, and DPI
With the peer group set, compare your target fund against it on Net IRR, TVPI, and DPI. Each metric tells you something different. IRR captures the pace of returns, TVPI captures total value relative to capital paid in, and DPI shows how much has actually been returned to investors in cash.
Read the comparison fund by fund, not against an aggregate median. You can see which peers sit above the target, which sit below, and by how much. A fund that looks middle of the pack against a median may turn out to be close to the top of a tightly defined group, or the reverse.
Step 5: Re-run it on a schedule
A benchmark run once at the point of commitment goes stale. Because the peer group is a set of saved criteria rather than a static report, you can re-run the same logic quarterly or annually as new performance data and vintages are added.
For a multi-manager portfolio, that means re-running the custom benchmark for each fund and flagging any whose relative position has moved meaningfully since the last refresh. The same peer group logic gets reused, not rebuilt.
Who this helps
Consultants and allocators can test a manager’s performance claims against a peer group they built themselves, instead of one supplied by the manager. GPs and IR teams can show investors exactly which funds make up the comparison and why, which makes the conversation about the group’s construction rather than the number.
The underlying dataset also connects each fund to its GP, portfolio companies, and transactions, so the benchmark sits alongside the research that explains it.
Dakota Private Markets for Benchmarking
Dakota Private Markets, Powered by Dakota, gives you Net IRR, TVPI, DPI, and RVPI for more than 18,000 named funds. You build a custom peer group by strategy, vintage year, geography, and fund size, then compare a target fund against each named peer, not an aggregate median. It is available for 9,500 USD per year for up to five users.
Request Access and build your first custom peer group on live, named fund data.
