A High-Water Mark, Worked to the Penny
The Fluxy Team
-A High-Water Mark, Worked to the Penny
We wrote before about why high-water marks and hurdles exist. This post is the arithmetic — one investor, four periods, every number shown. It's the example we wish every emerging manager worked through before charging their first performance fee, because the failure mode isn't dishonesty. It's a spreadsheet that quietly overcharges through a recovery.
The setup
An investor deposits $100,000 into a unitized book at a NAV-per-unit of 1.0000, receiving 100,000 units. Terms: 20% performance fee above the high-water mark (HWM), crystallized annually, no hurdle (we'll add one at the end). The HWM starts at their entry NAV: 1.0000.
Period 1: a good year
The book returns +30% gross. NAV-per-unit: 1.3000. Their holding: $130,000.
Fee math: NAV above HWM by 0.3000 per unit. Fee = 20% × 0.3000 × 100,000 units = $6,000, paid by redeeming units at crystallization NAV: 6,000 / 1.3 = 4,615.38 units. The investor now holds 95,384.62 units worth $124,000 — exactly the +30% year net of a 20% fee on gains.
The HWM ratchets to 1.3000. That ratchet is the whole institution: it records the highest level on which fees have already been paid.
Period 2: the drawdown
The book loses 20%. NAV: 1.0400. Holding: $99,200. Performance fee: zero — NAV is below the 1.3000 mark. Nothing else happens, and that's the point: no clawback, no negative fee, just a mark that doesn't move down. (Management fees, if any, still accrue — they're a separate meter.)
Period 3: the recovery — where spreadsheets go wrong
The book gains 15%. NAV: 1.1960. Still below 1.3000. Correct fee: zero.
Here is the classic mistake: a naive spreadsheet computes "this period's return was +15%, fee = 20% of the gain" and charges roughly $2,975. That charges the investor twice for the same ground — they already paid on everything up to 1.3000 in Period 1. Recovering lost ground is not performance; the mark exists precisely to encode that. If your fee engine keys on per-period returns instead of a persistent per-investor mark, it makes this mistake automatically, every drawdown cycle.
Period 4: a new high
The book gains 20%. NAV: 1.4352. Now fees apply — but only above the mark: 1.4352 − 1.3000 = 0.1352 per unit. Fee = 20% × 0.1352 × 95,384.62 units = $2,579.20. HWM ratchets to 1.4352.
Total performance fees across the whole ride: $8,579 — charged only on the two occasions the investor's wealth reached ground it had never reached before.
Three complications real books hit
- Mid-cycle deposits. A second deposit buys units at the current NAV, and those units get their own entry mark. One investor can carry multiple lots with different marks; averaging them silently over- or under-charges one lot. (Unitized accounting makes this tractable — each cashflow is its own units-at-a-NAV event.)
- Hurdles. With a 5% hurdle, Period 4's fee applies only to the gain beyond mark × 1.05. Order of operations matters and belongs in writing.
- Crystallization timing. Fees charged on paper gains that later evaporate are gone; annual crystallization (or on-redemption) is the investor-friendly standard, and an explicit, auditable crystallization event beats continuous silent skimming.
This entire calculation — per-investor marks, per-lot entries, hurdle ordering, crystallization as an explicit event — is what our accounting engine runs automatically for every investor in a book. Not because the math is hard, but because it's exactly the kind of math a tired human with a spreadsheet gets wrong in the investor's disfavor, once, and never gets to explain away.
Per-investor high-water marks, hurdles, and crystallization — computed by the engine, never by hand. See investor accounting →
Related reading