VWAP (volume-weighted average price)

Definition

VWAP is the volume-weighted average price of all trading in an instrument since a session anchor — the average price actually paid today. It has two distinct lives: an EXECUTION BENCHMARK (institutional orders are graded against it — its original, best-grounded use) and an INTRADAY REFERENCE LEVEL (price above/ below VWAP as session bias; distance from VWAP as stretched/ compressed — vwap_dist_pct).

How it works / structure

• Formula: VWAP_t = Σ(P_i × V_i) / Σ(V_i) over all trades since the anchor (session open by default; anchored variants start at events — earnings bars, swing extremes).
• Parameters (engine-executable): anchor (session vs custom), price convention (trade prints vs bar typical price × volume — the bar approximation is standard at replay granularity), band construction (±kσ of VWAP-distance), session definition for futures (ms-sessions-auctions).
• Why it attracts flow: execution algorithms slice orders to track VWAP (Berkowitz et al established the benchmark; Madhavan surveys the strategy class) — making VWAP partly self-referential intraday: it is watched because it is watched, and institutional slicing clusters around it.

When it applies

Execution-quality accounting (fills vs session VWAP — the platform’s slippage grading uses it, ms-slippage-friction); intraday mean-reversion and trend conventions (strategy-day-trading-styles VWAP-fade and VWAP-hold setups); anchored-VWAP as a “average holder basis since event” estimate for post-event behavior hypotheses.

Risk profile & failure modes

• Resets daily: session VWAP carries no information across sessions; overnight gaps orphan yesterday’s level entirely.
• Late-session inertia: as cumulative volume grows, VWAP freezes — afternoon VWAP distance measures the morning, not the present; early-session VWAP is noise on thin volume.
• Level lore: “VWAP acts as support/resistance” is folklore in its strong form — flow clustering near VWAP is real (execution algos), a reliable bounce is not; replay decides per convention.
• Data granularity: tick-true VWAP and bar-approximated VWAP diverge on volatile bars; the engine pins the approximation and grades claims at that fidelity.

Evidence & limits

Berkowitz-Logue-Noser (1988) established VWAP as the execution- cost benchmark; Madhavan (2002) documents the VWAP-tracking execution industry. That institutional flow anchors to VWAP is documented fact; that price systematically reverts to or bounces from VWAP is a trading convention requiring per-setup replay. The execution-benchmark use is the evidenced one; the signal use is hypothesis.

Falsifiable-thesis examples

Illustrations only, not signals:

• “X, stretched 2σ above session VWAP in the first hour on a no-news day, will touch VWAP before the close” — falsified by the session’s path.
• “This program’s fills averaged within 5bp of interval VWAP this month” — falsified by the execution accounting.

Cross-references

• Execution grading: ms-slippage-friction, ms-liquidity
• Intraday setups built on it: strategy-day-trading-styles
• Distribution sibling: indicator-volume-profile (where volume traded by PRICE rather than time)
• Platform bindings: vwap_dist_pct, rel_volume