Market Consensus: Mid Price vs Last Trade vs VWAP
Why consensus definition matters
Many scorecards say they benchmark against “the market”. That is meaningless unless you define:
• which price you use (last trade, mid, VWAP)
• when you sample it (timestamp or window)
• how you handle thin markets
If you do not define consensus, you get noisy baselines and accidental look ahead bias.
Option 1: last traded price
Last trade is the most recent executed price.
Pros
• simple to fetch
• reflects an actual transaction
Cons
• can be stale in low volume markets
• can be moved by one small trade
• can be far from where you can currently trade
Last trade is often the worst choice in thin markets.
Option 2: mid price
Mid price is the midpoint between best bid and best ask:
mid = (bid + ask) / 2
Pros
• reflects current market quotes
• more stable than last trade when trading is sparse
Cons
• can still be misleading if the spread is wide
• can be gamed in extremely illiquid books (quotes can be posted without intent)
Option 3: VWAP over a window
VWAP is volume weighted average price over a time window:
VWAP = sum(price * volume) / sum(volume)
Pros
• less sensitive to one small print
• can reflect where volume actually traded
Cons
• requires enough volume, otherwise it collapses back to a few prints
• window choice is a methodology parameter that must be documented
How to choose in practice
If the market is liquid
VWAP over a short window or mid price at a checkpoint are both reasonable. Pick one and document it.
If the market is thin
Mid price is usually better than last trade, but you should also flag:
• wide spreads
• low depth
• low recent volume
In very thin conditions, market consensus may be too noisy to use as a benchmark. Base rate can be a better default.
Consensus window and checkpoints
A robust approach is to define a consensus window relative to a fixed evaluation checkpoint.
Example:
• checkpoint: T-24h before settlement
• consensus: VWAP from T-30h to T-24h
This avoids late information leakage and reduces noise.
How to prevent look ahead bias
Two rules:
• consensus timestamp must be at or before the forecast time
• if you compare to a checkpoint, your forecast must be the one that existed at that checkpoint
If either rule is broken, you are using future information.
Common mistakes
Mistake: using last trade because it is easy
This is the most common source of noisy and misleading baselines.
Mistake: not reporting liquidity
If the spread is huge or volume is near zero, your market baseline is low quality. Flag it or exclude it.
Mistake: changing definitions later
If you change mid vs VWAP vs window, your benchmark series breaks. Keep methodology stable.
Takeaway
Market consensus must be defined. Last trade is often noisy. Mid price is a good default when spreads are reasonable. VWAP is strong when there is enough volume. For scorecards, combine consensus with checkpoints and liquidity flags to avoid noise and look ahead bias.
Related
• VWAP
• Choosing a Baseline: 50 50 vs Base Rate vs Market Consensus