Optimal Execution for the Proxy Basket: Minimizing Market Impact on Entry and Rebalance
Abstract
The proxy basket is liquid by design — $85M ADV on the ETF leg, $30M+ on each single-name leg. That does not mean the $1.7M entry is free. Lift the offer on all three legs at the open and the slippage bill runs $3,000–$5,000 on day one, before a single option is sold. Execute passively over the session and the same entry costs $1,200–$1,800. Rebalanced monthly and rolled alongside the options overlay, that friction compounds into a five-figure annual drag. This note prices each piece: entry math, VWAP versus implementation shortfall, the fill assumptions that break backtests, and what 52 rolls a year cost.
1. The Entry Math: $1.7M Across Three Legs
Start with the concrete ticket. Basket notional $1.7M:
- $1.02M (60%) — small-cap industrials ETF, ADV $85M.
- $340K (20%) — liquid single-name industrial A, ADV $32M.
- $340K (20%) — liquid single-name industrial B, ADV $30M.
Participation rates look harmless: 1.2% of ADV on the ETF, ~1.1% on each single name. That is the number that fools people into hitting “buy” in one clip.
Price it leg by leg at the touch. The ETF trades a 1–2 cent spread on a $60 handle — roughly 2–3 bps to cross. Each single name trades a 4–6 cent spread on $40–50 handles — 8–12 bps to cross. A single market-order entry therefore costs:
- ETF leg: $1.02M × 0.0003 ≈ $300
- Single-name legs: $680K × 0.0010 ≈ $680
- Walking the book on the single names at 1% of ADV adds another 5–10 bps of temporary impact: $350–$700
Total aggressive-entry bill: $1,700–$2,400, or 10–14 bps. On a wide-spread morning with chasing fills, it prints $3,000–$5,000 (18–30 bps) — 6–10% of one month’s $52,500 net carry, burned before the overlay earns anything.
The micro-cap itself is untouched — that is the point of the structure. But the proxy side still demands real execution. “Liquid” means the cost is measurable in basis points, not that it is zero.
2. VWAP Execution: The Default for Entry and Rebalance
For the initial build and each monthly rebalance, volume-weighted average price (VWAP) is the workhorse. The logic is simple: at ~1% participation, spreading the order across the session captures the spread more often than it pays it, and temporary impact decays to noise.
Practical settings: full session for entry, first half for rebalances under $400K. Cap participation at 5% of trailing 5-minute volume. Post at or inside the touch; pay the spread only in the final 15 minutes to complete. Expected cost: 7–9 bps blended, or $1,200–$1,500.
VWAP’s weakness: it guarantees average, not arrival. If the basket runs 1.5% while the algo works passively, spread savings are dwarfed by drift — and drift on a hedge leg is tracking error against the live micro-cap exposure.
3. Implementation Shortfall: When Arrival Price Is the Benchmark
Implementation-shortfall (IS) algorithms optimize against the price at order creation, not the session average. They front-load when spreads are tight and volatility is low, slow down when impact builds, and cross aggressively when price moves against the order.
Use IS instead of VWAP in three cases:
- Entry on a moving tape. Beta drift exceeds 0.15 and every hour unhedged is open risk. Paying 3–5 bps extra to finish by midday beats saving 2 bps against VWAP while the market runs 80 bps.
- Single-name legs above 2% of ADV. Binds if the basket concentrates — one name halted, the rebalance forced into two names.
- Option-linked stock tickets. Trimming the underlying against assignment is time-linked to expiry and cannot wait for a full-session VWAP.
Expected cost: 10–15 bps all-in. VWAP wins on quiet days. IS wins when delay costs more than spread.
| Execution method | Benchmark | Expected cost (this basket) | Best use |
|---|---|---|---|
| Market / marketable limit | None | 18–30 bps ($3,000–$5,000) | Never for entry; only for emergency hedge completion |
| VWAP, full session | Session VWAP | 7–9 bps ($1,200–$1,500) | Default: initial build, routine monthly rebalance |
| Implementation shortfall | Arrival price | 10–15 bps ($1,700–$2,600) | Trending tape, beta-drift urgency, expiry-linked tickets |
| Passive close / MOC split | Closing price | 5–8 bps ($850–$1,350) | Rebalances under $300K when closing-auction volume exceeds 8% of ADV |
The table is the decision rule. Default to VWAP. Switch to IS when the cost of waiting exceeds the cost of crossing. Never lift the full ticket at the open — that row exists so nobody repeats it.
4. Fill Assumptions That Invalidate Backtested Carry
Backtests of the options overlay almost always assume frictionless stock fills: mid-price, no spread, no impact, full size instantly. Three assumptions in particular quietly inflate the published 36.5% APY.
Mid-price fills. Marking entries at mid overstates returns by half the spread per side. Across entry plus twelve monthly rebalances at ~20% turnover, that is 25–50 bps a year of phantom return. Haircut gross carry by 30–50 bps or demand a bid/ask-aware backtest.
Instant full-size fills. Top-of-book on these single names is $15K–$40K — a $340K leg never prints in one trade at the touch. Real fills walk 2–4 levels. Add 3–5 bps per single-name ticket versus the idealized fill.
Zero rebalance friction. Turnover per rebalance runs 15–25% of notional ($250K–$425K). At 8–12 bps all-in, each costs $200–$500; twelve cost $2,400–$6,000 a year, invisible in any backtest that holds weights static.
The correction is mechanical: subtract 60–100 bps annualized in equity execution friction from backtested carry before quoting net yield. The structure still pays. It just pays 35.5–36.0%, not the spreadsheet’s 36.5%.
5. The Real Cost of Rolling 52 Times a Year
The equity side is the smaller bill. The larger one is the options overlay. A monthly 30Δ short call on three components plus a bi-monthly put spread is roughly 48 tickets a year; run the short calls on a 2–3 week cycle and the count clears 52 rolls.
Each roll crosses an options spread that is far wider than the equity spread:
- ETF options: 3–6 cents wide on $2.00–$4.00 of premium. Rolling 30–40 contracts costs $90–$240 in spread per leg.
- Single-name options: 8–15 cents wide on comparable premium. Rolling 10–15 contracts per name costs $80–$225 per leg.
- Put spreads (two legs): double the ticket count, $150–$400 per roll on the ETF hedge.
Annualized at 52 rolls: $12,000–$22,000 a year in options spread and impact — 0.7–1.3% of the basket. Add brokerage and OCC fees (~4,000–6,000 contracts a year: $2,000–$4,000) plus 2–4 assignment-driven stock tickets ($500–$1,500 friction each).
Three rules keep the bill inside budget. Roll on limit and work the order — makers compete inside the quote on polite multi-leg tickets. Stagger the three components by 3–5 days instead of rolling everything into month-end expiry. Skip the roll when bid-ask exceeds 10% of premium.
Total friction, equity plus options: $17,000–$32,000 a year, ~1.0–1.9% of notional — a 3.5–6.5% haircut to net carry. Planned and survivable, but only if it is in the model from day one.