4The Asset Manager and the Fund

On 3 June 2019 a UK equity income fund suspended dealing. It had been worth more than £10.1 billion two years earlier and was worth £3.6 billion in the run-up to the suspension; its investors had kept asking for their money back, and its managers had kept selling what was easy to sell. The UK regulator later found that at the suspension only 8% of the fund’s investments could be sold within seven days, while its investors were entitled to their money within four. A fund’s liquidity terms are a promise about its assets, written in the documents of a vehicle; this chapter reads the vehicle, its fees and its promises as the manager must run them.

4.1 Structures: manager, fund and vehicle

An asset manager (One Quant Book 1, chapter 1) does not own the money it manages. It runs a fund, a separate legal vehicle owned by its investors, under a mandate (Book 1, chapter 3), and is paid fees out of it. The separation is what protects investors if the manager fails, and what lets one strategy be sold in several wrappers.

4.2 Service providers: administrator, custodian, auditor, prime broker

The service providers are the fund’s controls. An administrator that prices the portfolio independently stops a manager marking its own book; a custodian that holds the assets stops them being used for anything but the fund; an auditor tests both once a year. A prime broker (Book 1, chapter 6) finances the positions and often holds the assets it lends against, which is why a hedge fund’s documents say what may be rehypothecated (chapter 18). A manager chooses these firms, but investors’ due diligence (chapter 25) checks each of them.

4.3 Fees in practice: classes, hurdles, crystallisation and equalisation

A fund’s fees are the management and performance fees of Book 1, chapter 3, with a high-water mark; in practice they are set per share class, and three mechanisms decide who pays what.

The more often fees crystallise, the more the manager earns on the same returns: a gain crystallised in March is not given back when April loses. On five years of monthly returns with a mean of 8% a year and a volatility of 12%, a 20% performance fee crystallised every month takes 11.54% of the starting capital on average over 1 000 paths, against 11.26% quarterly and 10.78% annually.

4.4 Liquidity terms and the liquidity mismatch

Form PF, the confidential report large US hedge-fund advisers file, gives the aggregate picture: investors in qualifying hedge funds could withdraw 9.0% of their capital within seven days in the third quarter of 2025, while the funds reported that they could sell 51.8% of their portfolios within seven days (Figure 4.2). At every horizon the portfolios were more liquid than the investors’ terms: in aggregate the hedge-fund industry is not mismatched. The fund suspended in 2019 was mismatched by a factor of more than twelve at seven days: its investors could have all of their money within four days, and it could sell 8% of its assets within seven.

Selling liquid first minimises the cost of today’s redemption and hands the remaining investors a less liquid fund; the next redemption is then more expensive, and redeeming early becomes worth more than staying. The model of firm.fundterms (Listing 4.1) meets a 30% redemption within a seven-day notice period for two ladders, the aggregate hedge-fund portfolio of Form PF and the illiquid fund, with illustrative selling costs that double for assets sold faster than their horizon:

def redeem(ladder, request: float, notice_days: int, policy: str = "liquid_first", gate: float | None = None,
           swing: bool = False, fire: float = 3.0) -> dict:
    """Meet a redemption of `request` (share of NAV) within notice_days. Buckets slower than the notice period
    can be sold only at `fire` times their cost. The gate caps the payment; the rest is deferred. Without swing
    pricing redeemers are paid at a NAV that ignores the selling cost, so the investors who stay bear all of it
    (their dilution); with it, the redeemers bear it. Returns shares of the starting NAV."""
    pay = request if gate is None else min(request, gate)
    order = sorted(ladder, key=lambda b: b.days)
    sold = {b: 0.0 for b in order}
    if policy == "liquid_first":
        need = pay
        for b in order:
            x = min(b.share, need)
            sold[b] = x
            need -= x
    elif policy == "pro_rata":
        for b in order:
            sold[b] = b.share * pay
    else:
        raise ValueError(policy)
    cost = sum(x * b.cost * (fire if b.days > notice_days else 1.0) for b, x in sold.items())
    stayers = 1.0 - pay
    # redeemers are paid at a NAV that ignores the cost of selling: without swing pricing the stayers bear all of it
    dilution = 0.0 if swing else cost / max(stayers, 1e-12)
    left = sum(b.share - sold[b] for b in order)
    after = [Bucket(b.days, (b.share - sold[b]) / max(left, 1e-12), b.cost) for b in order]
    return {"paid": pay, "deferred": request - pay, "cost": cost, "dilution_stayers": dilution, "ladder_after": after}

The pro rata slice keeps the fund’s shape but costs the stayers 5.4% on the hedge-fund ladder and 8.8% on the illiquid one, because a slice of every bucket must be sold within a week. Swing pricing moves the cost to the redeemers and leaves the liquidity problem where it is. A gate buys time: it pays 10% now, defers 20%, and leaves the manager weeks to sell the slower assets at a normal cost. The illiquid fund has no good option: whatever it does, its first 8% of redemptions exhaust everything it can sell in a week. The largest redemption it can meet within seven days at a cost below 1% is 8% of its NAV; the Form PF ladder’s is 51.8%.

4.5 What investors ask for: side letters and managed accounts

A side letter that gives one investor better liquidity creates a mismatch among investors: it redeems first and leaves the others with a less liquid fund, which is why regulators and consultants ask whether any investor has preferential liquidity. Large investors that want full transparency, their own liquidity and no exposure to other investors’ flows use a separately managed account instead: the manager trades the investor’s own account on the same model, and the investor’s custodian holds the assets. The manager then runs several copies of one strategy, and must allocate trades among them fairly, which its compliance function tests (chapter 16).

4.6 Tutorial: one fund, two investors and a bad week

What to change next. Run two redemption weeks in a row and watch the liquid-first fund’s second week; give one investor a side letter with daily liquidity in a fund with monthly dealing.

4.7 Build: the fund’s terms

4.8 Exercises

4.9 Problem: The Friday Queue

4.10 Interview questions