Global UI: Price vs EWMA VWAP Z-Score Common floor: > 2.0

Price vs EWMA VWAP Z-Score

Statistical rarity of the current EWMA VWAP deviation, normalized per instrument. One bar behind the live reading by design.

Definition

#
Price vs EWMA VWAP (%) gives the raw deviation. This metric asks whether that deviation is unusual for this instrument. The baseline uses completed bars only (n−1 history), so the z-score trails the in-progress bar by one. That lag keeps the current bar from distorting its own reference before it closes.

Formula & calculation

#
Baseline: completed 1-minute EWMA VWAP deviations [price_vs_ewma_vwap(bar_1), …, price_vs_ewma_vwap(bar_n−1)]
Z-Score: (live_price_vs_ewma_vwap − mean(history)) / stddev(history)

Units & range

Z-score (standard deviations). Positive = extended above EWMA VWAP relative to this instrument's historical norm. A threshold of 2.0 carries the same statistical meaning across instruments because normalization is per-symbol.

Interpretation

#
Above +2.0: price has moved outside roughly the top 2% of EWMA VWAP deviations seen on this instrument. The threshold carries the same statistical weight on BTC and a small-cap altcoin.
Below −2.0: the same rarity on the downside.
The one-bar lag preserves the reference distribution's integrity. Remove it and the current bar teaches the baseline that its own extreme is normal.

Practical usage

#
Use in cross-symbol rules instead of the raw percentage. A threshold of 2.0 means the same rarity on BTC as on any altcoin.
For fade setups, flag the stretched deviation and wait for volume to compress or taker imbalance to reverse:
price_vs_ewma_vwap_z > 2.5 AND anomaly_5m.sell_vol_z rising AND anomaly_5m.current_move weakening
For breakout confirmation, pair with volume z-scores and taker imbalance. A z-score spike on thin volume is often a thin-market artifact.

Common mistakes

#

Frequent interpretation traps and misuse patterns to avoid when applying this metric.

  • Reading a high z-score as bullish or bearish. It measures extension, not direction.
  • Expecting the z-score to update mid-bar. It updates when the in-progress bar closes.
  • Acting on extension without volume or taker flow. Extension on thin participation is noise.