Beat VWAP, Still Pay More: Calculate Implementation Shortfall Including Unfilled Orders

A buy order can beat the market’s volume-weighted average price and still leave the portfolio worse off than the original investment decision implied. There is no contradiction: the fill-price comparison covers the shares bought, while the decision also includes shares that never traded. Implementation shortfall measures that wider gap. The useful question is not whether one benchmark makes the trade look good, but whether the full cash-and-position ledger reconciles. The stock, prices and fees below are hypothetical.

Fix the decision, quantity and endpoint before measuring cost

At 10:00, an investor decides to buy 1,200 shares at a recorded reference price of $50.00. The order reaches the execution process at 10:02, when the arrival reference is $50.08. Two lots fill; the remaining 500 shares are canceled at a predeclared 15:00 valuation cutoff, when the reference price is $50.40.

The 700 executed shares cost $35,088, giving a volume-weighted fill price of $50.125714. Add an assumed $7 in explicit fees. No financing interest, tax, currency movement or dividend is included. Those items would need separate treatment if present.

The decision price and arrival price refer to different timestamps in this example. A midpoint or other reference is a measurement convention, not proof that all 1,200 shares were executable there. Record the chosen quote source, order size and capacity assumptions; a benchmark built on impossible liquidity cannot become evidence of guaranteed attainable performance.

Reconcile the shortfall to cash and positions

Start both portfolios with $60,000. The paper portfolio buys all 1,200 shares at $50.00 and ends with $60,480 at the cutoff, a $480 gain. The actual portfolio spends $35,095 including fees, retains $24,905 cash and holds 700 shares worth $35,280. Its ending value is $60,185, a $185 gain.

Implementation shortfall = $60,480 − $60,185 = $295. This is a cost relative to the paper decision, not an actual cash loss of $295: the actual portfolio still gained $185 over the stated window.

The same result comes from the order ledger. Executed-price slippage is 400 × ($50.10 − $50.00) + 300 × ($50.16 − $50.00) = $88. The unfilled opportunity component is 500 × ($50.40 − $50.00) = $200. Add $7 fees to reach $295.

Expressed against the intended decision notional, the cost is $295 ÷ $60,000 × 10,000 = 49.17 basis points. A filled-only report instead shows $95 ÷ $35,000 × 10,000 = 27.14 basis points. That figure is not arithmetically wrong, but it covers a different quantity and denominator. It cannot stand in for the full-decision cost.

Split delay and execution without counting waiting twice

One consistent attribution measures delay only on shares that eventually filled, then measures the unfilled shares over the entire decision-to-cutoff period. With arrival at $50.08, the breakdown is:

The sum is $295. Another valid convention assigns delay to all intended shares: 1,200 × 0.08 = $96. If so, the remaining-share component must start at arrival, becoming 500 × (50.40 − 50.08) = $160. The total remains $96 + $32 + $160 + $7 = $295.

Combining the $96 delay with the original $200 opportunity term produces $335, overstating cost by $40. It counts the first $0.08 movement on 500 unfilled shares twice. Different attribution labels are acceptable; an unreconciled ledger is not. Also, the $32 execution term is not a clean estimate of the order’s causal market impact: unrelated news and price movements can contribute.

Why a VWAP win does not settle the question

Suppose market VWAP for the specified execution window is $50.18. Buying the executed 700 shares at that benchmark would cost $35,126. The actual pre-fee spend of $35,088 is $38 better, or $31 better after the $7 fee. The trader has beaten that VWAP comparison while the full decision still shows $295 shortfall.

VWAP answers a useful but narrower question about prices during a selected interval. State the window, venues, eligible prints, auction treatment and whether the investor’s trades enter the benchmark. A benchmark influenced by one’s own order is not an independent control. Nor should an attractive fill-only score hide a low completion rate.

This does not make VWAP intrinsically bad. A mandate seeking participation through a session may value it; a time-sensitive decision may care more about pre-trade prices and completion. Choose the benchmark to fit the purpose before execution, rather than choose the most flattering benchmark afterward.

Selling reverses the price sign, not the fee sign

Let Q be intended quantity, qᵢ each fill, pᵢ its price, p₀ the decision reference and pT the fixed endpoint price. Use s=+1 for a buy and s=−1 for a sell. With the same-currency and no-financing assumptions above:

Shortfall = s × [Σ qᵢ(pᵢ−p₀) + (Q−Σ qᵢ)(pT−p₀)] + net explicit costs.

For a mirrored sale, take decision price $50.00, arrival $49.92, fills of 400 at $49.90 and 300 at $49.84, cutoff $49.60 and fees $7. Selling below the decision reference costs $88; the unsold 500 shares lose $200 relative to the paper sale. Fees add $7, again giving $295. Do not multiply the fee by the side sign.

Opportunity cost can also be negative. Keep the original buy fills and fees, but consider a separately specified scenario with cutoff price $49.80: the unfilled component is −$100, and total shortfall is −$5. That is a benefit relative to the benchmark, not proof that the portfolio made money or that waiting is always superior. Changing the endpoint after seeing the price would invalidate the comparison.

What a defensible execution review should contain

Keep the parent decision, intended quantity, timestamps, order revisions, fills, fees and final valuation in one audit trail. Cancellation does not erase the unfilled quantity from the original decision. A genuine change of investment view should be recorded when it happens, with a new decision segment where appropriate, rather than rewritten after the outcome. Reconcile cancellation acknowledgments and late fills so that replacement orders are not counted twice.

The mechanics behind incomplete orders are explained in limit prices, queue priority and partial fills. A price cap may reduce the price paid on fills while leaving exposure unimplemented. Faster trading may improve completion but raise spread or impact costs; neither speed nor patience is universally optimal.

Finally, compare like with like: order size relative to liquidity, urgency, volatility, trading window and completion requirements. Report completion rates and the distribution of shortfall across comparable orders, not just a single average. A one-order calculation can expose a bookkeeping mistake; it cannot by itself establish misconduct, identify causal impact or rank a broker’s skill.