The hedge fund in an ETF — are hedge-fund returns just rentable factor beta?
Hasanhodzic & Lo's question, run on the investable HF-replication ETFs: QAI (multi-strategy), DBMF
(managed futures), MNA (merger arb), BTAL (anti-beta). Bootstrap each hedge-fund strategy from liquid
factors — equity, rates, credit, commodity, dollar, momentum — and ask what's replicable beta and what,
if anything, is left. Alpaca SIP daily, total-return; each fund regressed over its own max window.
CONDITIONAL — mostly rentable factor beta, with one real exception. The multi-strategy hedge fund
clone is ~75% replicable with zero residual alpha and delivers cash-like returns that lag a 60/40 —
confirming Lo: 2-and-20 ate the rest. The honest exception is managed futures (DBMF), a genuinely
low-correlation crisis diversifier the linear clone can't capture. Rent that; skip the complex.
| fund | strategy | R² on liquid factors | residual α/yr | residual Sharpe |
|---|---|---|---|---|
| QAI | multi-strategy | 75% | −0.3% | ~0 |
| MNA | merger arb | 27% | +0.4% | ~0 |
| BTAL | anti-beta | 29% | +5.2% | ~0 |
| DBMF | managed futures | 18% | +5.2% | ~0 |
The multi-strat clone is three-quarters rentable beta with no residual alpha. Managed futures and anti-beta have low R² — but that's because they're nonlinear (trend, convexity), not because they hide linear alpha; the residual Sharpe is ~0 in every case.
| book | Sharpe | CAGR | maxDD | corr→SPY |
|---|---|---|---|---|
| QAI (multi-strat) | +0.67 | +4.5% | −15% | +0.82 |
| DBMF (managed futures) | +0.65 | +7.9% | −21% | +0.11 |
| MNA (merger arb) | +0.37 | +2.4% | −17% | +0.52 |
| BTAL (anti-beta) | −0.21 | −5.7% | −53% | −0.53 |
| SPY | +0.86 | +15.9% | −34% | +1.00 |
| 60/40 (SPY/IEF) | +0.88 | +10.1% | −21% | +0.97 |
The "alternatives" delivered cash-to-mediocre returns and lagged a 60/40 — the net-of-fee hedge-fund
experience in investable form. QAI is equity beta (corr 0.82) at a fraction of SPY's return; BTAL
bled (anti-beta is a structural short). The standout is DBMF: a respectable +7.9% at 0.11
correlation to equities — the one thing here that actually diversifies.
An equal-weight basket of the four is market-neutral: β→SPY +0.01, corr→SPY +0.03, with +2.8%/yr over SPY-beta and Sharpe +0.53. Genuinely diversifying — and genuinely low-return. That's the honest shape of "liquid alternatives": uncorrelated, not high-octane.
bootstrap is a diversifier finding, not an aggressive engine — the aggressive capstone is
belligerent. The usable conclusion: rent managed
futures (DBMF) as a low-correlation diversifier (it pairs with boulder/bridgewater), and don't
pay for the rest of the HF complex, which is rentable beta the clone already proves. For the family's
books, DBMF is a sleeve ingredient; QAI/MNA/BTAL are a lesson.
The ETF clones are themselves fee'd replicas. The real test is raw HFRI / manager-level data: does top-decile manager selection (not the index) add alpha beyond the rentable beta — the one place the "2-and-20 is worth it" claim could still be true. That needs manager databases, not ETFs.
export ALPACA_KEY_ID=... ALPACA_SECRET_KEY=...
python3 research/bootstrap_1.pyBlaque Baux — a Carter Warrens research initiative. Research only; not investment advice. The verdict falls where the data puts it — most of the hedge fund was rentable; the part that wasn't is the part that diversifies.