Modeled 1.9x gross on a fund and collected 1.31x net. Where the other 0.6 went
2019 vintage, $60M value-add fund, 11 deals, $150k commitment from me. Final distribution landed in the spring and the fund is done at 6.5 years. Realized deal level gross multiple across the portfolio was 1.87x. My net was 1.31x. Everything I'm about to list was disclosed. I had the documents. I built the model wrong.
Management fee. 2 percent on committed capital for the five year investment period, then on invested. Called it $15,000 over five years plus another $4,400 after. That alone is 13 percent of my commitment and I had modeled it against invested capital from year one, which understated it badly because my money wasn't fully drawn until month 31.
Fund expenses. Uncapped, ran 0.85 percent of commitments a year, plus $520k of organizational costs across the fund. My share of expenses over the life came to about $9,800.
The waterfall. American, deal by deal, with an 8 pref. Deals 2, 4 and 5 exited well in 2021 and 2022 and the manager took carry on them. Deals 8 and 10 lost money in 2024. Clawback existed. It was capped at carry actually received by the GP entity net of taxes assumed at the highest marginal rate, and the manager had already distributed carry to individuals. I got back roughly 40 cents of every dollar of clawback the arithmetic said was owed.
The subscription line. They ran a credit facility and called capital late, so the reported net IRR was 16.8 percent. My multiple was 1.31x. Both numbers are accurate. Only one of them is my money.
What I'd do differently: build the model in calendar quarters on committed dollars, not on invested dollars, and put every fee and expense in the same column as the distribution it reduces. Then ask the manager for the prior fund's net to LP multiple after all fees and expenses, in writing, and refuse to accept gross deal IRRs as an answer. If they can't produce a net multiple by vintage, they know why.