Execution Intelligence

Feedback loops that measure and improve execution quality over time.

Execution Intelligence provides a feedback loop: measure execution quality, identify problems, and automatically adjust strategy parameters to improve.

The Feedback Loop

1. Execute Order

   Submit order to the exchange via the strategy pipeline.

2. Measure Quality

   Track fill rate, slippage, adverse selection, and market impact.

3. Diagnose Issues

   Identify root causes: spread too tight, size too large, bad timing.

4. Adjust Parameters

   Auto-tune spread width, order size, and timing based on diagnostics.

5. Execute Again

   Run the next cycle with improved parameters. Loop continues.

Most trading systems are open-loop: they execute and move on. Execution Intelligence closes the loop by feeding results back into the strategy.

Execution Quality Metrics

Metric	What It Measures	Why It Matters
Fill rate	% of orders that get filled	Low fill rate = quoting too aggressively
Slippage	Difference between expected and actual fill price	High slippage = poor execution or thin books
Adverse selection	P&L immediately after fill	Negative = we’re getting picked off by informed traders
Market impact	Price movement caused by our orders	High impact = we’re too large for this market
Timing cost	Cost of delayed execution	Relevant for TWAP/VWAP algos

# Measure execution quality
quality = execution_quality(
    strategy="political_mm",
    period_days=7,
)
# {
#     "fill_rate": 0.73,
#     "avg_slippage_bps": 12.5,
#     "adverse_selection_5s": -0.003,
#     "market_impact_bps": 8.2,
# }

Adaptive Execution

Based on quality metrics, automatically adjust strategy parameters.

Spread Auto-Tuning

High adverse selection --> widen spread
Low fill rate --> tighten spread
High market impact --> reduce order size

Signal	Parameter	Direction
Adverse selection > threshold	spread_width	Increase by 10%
Fill rate < 50%	spread_width	Decrease by 5%
Market impact > 10bps	order_size	Decrease by 20%
Fill rate > 90% and low adverse selection	order_size	Increase by 10%

Time-of-Day Patterns

Markets have liquidity patterns. Execution Intelligence learns when to be more/less aggressive:

# Analyze time-of-day patterns
patterns = execution_patterns(strategy="crypto_mm", group_by="hour")
# {
#     "best_hours": [14, 15, 16],  # UTC - US market hours
#     "worst_hours": [3, 4, 5],     # Low liquidity
#     "recommendation": "Reduce size 50% during 03:00-06:00 UTC"
# }

Cross-Exchange Execution

When the same market exists on multiple exchanges, route to the best venue.

Smart Routing

Factor	Weight	Description
Price	40%	Best bid/ask across venues
Liquidity	30%	Depth available at target price
Fees	20%	Maker/taker fee structure
Latency	10%	Historical fill time

# Smart order routing
route = smart_route(
    market="will-btc-100k",
    side="buy",
    size=100,
    venues=["polymarket", "kalshi"],
)
# {"venue": "polymarket", "expected_price": 0.62, "expected_slippage": 0.001}

Exchange Health Monitoring

Detect degraded exchange performance and route around it:

• API latency spikes
• Increased error rates
• Orderbook thinning
• Fill rate drops

MCP Tools

Tool	Description
execution_quality	Quality metrics for a strategy or the fund
execution_patterns	Time-of-day and market-specific patterns
route_analysis	Best venue analysis for a given trade
exchange_health	Current health status of all connected exchanges