Tick-Level Sense (v2.6+)

Reactive vs Anticipatory

The classic sense() uses volume spikes as a liquidity proxy:

Volume spike (current vol >> avg vol) → elevated liquidity_above → trap_risk rises

Problem: volume is reactive — it is recorded after the bar closes, meaning the signal arrives too late.

With bid/ask data, STRATA uses spread widening instead:

Spread widening (current spread >> avg spread) → elevated liquidity_above → trap_risk rises

Spread is anticipatory — market makers widen spreads before executing large orders, giving STRATA a leading indicator.

Auto-Detection

sense() automatically detects bid/ask — no code changes needed:

from strata import sense

# OHLCV only — uses volume proxy (reactive, as before)
candle = {"open": 150.0, "high": 151.2, "low": 149.5, "close": 150.8, "volume": 1_200_000}
signals = sense(window)
# signals["source"] == "volume"

# Add bid/ask — automatically uses spread (anticipatory)
candle["bid"] = 150.75
candle["ask"] = 150.85
signals = sense(window)
# signals["source"] == "spread"
# signals["liquidity_above"] reflects spread z-score, not volume

Spread Signal Formula

spread_ratio   = (ask - bid) / mid_price          # relative spread
avg_spread     = mean(spread_ratios[-vol_period:]) # rolling average
std_spread     = std(spread_ratios[-vol_period:])  # rolling std

z = (current_spread - avg_spread) / std_spread

liquidity_above = 1 - 1 / (1 + max(0, z)^1.5)

Mapping: - z = 0 (normal spread) → liquidity_above = 0.0 - z = 2 (2σ above avg) → liquidity_above ≈ 0.73 - z = 4 (4σ above avg) → liquidity_above ≈ 0.98

Extended: sense_tick()

For full Level-2 data (bid/ask + order book sizes):

from strata import sense_tick

candle = {
    "open": 150.0, "high": 150.5, "low": 149.8, "close": 150.2,
    "volume": 1_000_000,
    "bid": 150.18, "ask": 150.22,
    "bid_size": 2500,   # volume at best bid
    "ask_size": 800,    # volume at best ask
}
signals = sense_tick([...window..., candle])

Returns all standard signals plus:

Signal	Range	Meaning
spread_pressure	[0, 1]	Current spread vs historical norm
side_imbalance	[-1, 1]	Order book pressure: +1=sell, -1=buy

# bid-heavy (bid_size >> ask_size) → buy pressure → negative imbalance
signals["side_imbalance"] = -0.51   # buy pressure dominant

# ask-heavy (ask_size >> bid_size) → sell pressure → positive imbalance
signals["side_imbalance"] = +0.74   # sell pressure dominant

Effect on State Machine

Wide spread → elevated liquidity_above → higher trap_risk in STRATA-CORE:

spread_spike detected (z > 2σ)
    → liquidity_above = 0.73
    → update_state(): trap_risk increases
    → GUARD: may block LONG/SHORT if trap_risk > threshold
    → Final action: HOLD (conservative, avoiding the trap)

This happens before the price moves — anticipatory protection.