Echo
"The session commits at the open. By 15:20, that commitment is spent. I trade what comes after."
Echo
"The session commits at the open. By 15:20, that commitment is spent. I trade what comes after."
Every trading session tells a two-part story. The first part: the morning establishes direction. Big institutions come in with a program — maybe a macro catalyst, maybe a gap, maybe an index rebalance — and they spend the morning executing it. Price moves in their direction. Volume is elevated. ADX climbs. The trend is real.
The second part: by 3:20 PM ET, that program is 90%+ complete. The institutions that drove the morning move are nearly done. What remains isn't continuation — it's the aftermath. Sellers who couldn't get clean fills against the trend. MOC order supply from completing programs. Options hedging against the day's extreme. The pressure flips.
Echo enters at 3:30 PM ET in the direction opposite to the morning signal and exits at 3:58 PM ET. Twenty-eight minutes to capture the exhaustion reversion before the close.
Imagine the morning trend is a powerful shout into a deep canyon. The institutions shout with immense volume and strength, sending sound waves surging forward. But sound waves cannot travel forever; as they hit the canyon's far wall, they run out of forward energy. The shout dies, and the reverse sound waves — the echo — begin to bounce back through the air.
Echo is the mountain nymph, cursed to only repeat the words of others in reverse resonance. She doesn't yell into the void herself; she waits for the morning's great shout to hit its absolute limit of exhaustion, then rides the returning sound waves as they bounce back in the opposite direction.
Echo doesn't bet against the trend; she bets that the trend's shout has reached the canyon wall by 3:20 PM, leaving only the return echo to trade.
The original intuition for Echo was a continuation trade: enter at 3:30 in the same direction as the morning, ride the close-leg institutional completion. This is actually backed by academic research — Gao et al. (2018) in the Journal of Finance documented that the first-half-hour return predicts the last-half-hour return. It's real. Many traders use it.
But the data from Crucible's own backtests told a different story. When Echo added filters — VWAP persistence, elevated volume, ADX confirmation — the continuation signal got weaker, not stronger. The filtered sample (days where the morning trend was most confirmed and committed) actually reversed more often in the close window.
The academic finding applies broadly. But when you filter for days with maximum institutional confirmation — the sessions where the morning program was largest and most committed — those are precisely the sessions where the institutional programs are most complete by 3:20. The close window sees the exhaust, not the continuation. The very features that look like "strong trend" are actually "trend nearly complete."
Echo evaluates four conditions at precisely 3:20 PM ET. All four must be true. If any one fails, Echo is flat all day — no trade, no exceptions. The gates do two jobs simultaneously: confirm the morning program was real, and confirm it was large enough to be exhausted.
First-half-hour return ≥ ±0.25%. The morning had a real directional commitment. Below 0.25% — it's noise, not a program. No signal, no trade.
Price must still be on the correct side of session VWAP at 3:20 PM. If the morning signal was up, price must still be above VWAP. If it flipped, the program reversed.
Trend strength must confirm the session was actually trending, not ranging. ADX below 20 = the morning "signal" was chop, not institutional commitment.
Session cumulative volume (9:30–3:20 PM) must be at least 20% above the 20-session rolling average. Below-average volume means a thin session — the institutional programs weren't large enough to be meaningfully exhausted.
Passing all four gates qualifies a session for a potential trade. But within those qualified sessions, some are more exhausted than others. Echo uses an exhaustion score to filter out the weakest qualifying trades — the ones where the morning program was technically large enough to qualify but not large enough to reliably exhaust.
The exhaustion score multiplies two independent measures of program size: how much bigger-than-usual volume was (the institutional footprint) and how far the morning signal moved (the program's directional commitment).
Think of the exhaustion score like the acoustics of a canyon. Two factors determine how powerful the return echo will be: how loud you shouted (the morning signal return) and how wide the canyon walls are to catch the sound (the institutional volume). A quiet whisper in a narrow crevice creates no audible echo. A loud shout in a grand canyon creates a resonant, clear return echo.
Echo only trades when the acoustic conditions score at least 0.80 — ensuring a loud shout in a wide canyon, guaranteeing a strong return echo.
Echo doesn't limit itself to one instrument. Each session, it evaluates three instruments independently. Any instrument that passes all four gates and meets the exhaustion score floor enters a trade at 3:30 PM.
If multiple instruments qualify simultaneously, they all enter — but the total session risk is capped at 2.0% of equity. Each instrument's position is sized at 0.75% risk normally, but if the combined risk would exceed the 2.0% budget, all positions scale down proportionally.
Market order at 3:30 PM ET precisely. The entry is not triggered by price action at 3:30 — it's triggered by the Phase 2 confirmation (all gates passed at 3:20 PM). The direction is opposite to the morning signal: if the morning was bullish, Echo enters short. If the morning was bearish, Echo enters long.
Why 3:30 and not 3:20? The 10-minute gap between gate evaluation (3:20) and entry (3:30) is intentional. It allows the gate checks to complete cleanly without entering into the late-afternoon's most uncertain window. The last two minutes before 4:00 PM are dominated by MOC order imbalances — these are excluded by the hard exit at 3:58.
All positions closed at market at 3:58 PM. No exceptions. The close window's last 2 minutes (3:58–4:00 PM) are dominated by market-on-close order imbalances and settlement mechanics unrelated to the exhaustion thesis. Echo never holds into it.
Stop-loss: 1.5× ATR(14) on 5-minute bars placed at entry. If hit before 3:58 PM, exit immediately — the exhaustion thesis was wrong on this session. No re-entry after a stop-out.
At exactly 10:00 AM, compute the first-half-hour return on SPY, XLE, and XBI. Any instrument with |return| ≥ 0.25% has a morning signal. Note the direction. Do not trade.
Echo watches the session unfold. Monitors VWAP side and general session character. Does not trade at any point during the morning or midday.
For each instrument with a morning signal: check VWAP persistence, ADX ≥ 20, vol_ratio ≥ 1.2, and compute exhaustion_score. Any instrument that passes all four gates AND scores ≥ 0.80 gets a trade. If none qualify, no trade today.
Market order on all qualifying instruments, direction opposite to morning signal. Stop placed at 1.5× ATR immediately at fill.
Market order to close everything. Done for the session. The last 2 minutes before 4:00 PM are not Echo's territory.
Echo is in Phase 3 — paper trading active since 2026-06-20. TEMPER cleared strategy v6, which introduced the exhaustion_score floor (T=0.80) to restore win rate after the v5 multi-instrument expansion diluted quality with low-exhaustion sessions.
Development set (2021–2023, 94 trades): 56.4% WR, profit factor 2.14, max drawdown 2.7%. Holdout set (2024, 26 trades): 73.1% WR, profit factor 4.25, max drawdown 1.1%. All benchmarks cleared. Key caution: 2022 bear-market year was 52.4% WR / PF 1.07 — the primary ongoing risk in a sustained bear regime.
Exhaustion_score logged per trade. After 40+ live trades: confirm live score median ≥ 1.0 (in-sample median was 1.952). XBI win rate monitored separately — below 50% over 15+ trades triggers XBI exclusion review. The T=0.80 threshold must not be re-optimized in production.