Strategy Agent¶

Entry scanning, technical evaluation, signal generation, and trade state management

January 2026 Status

ACTIVE - Trend following strategy with sentiment integration

Feature	Status	Details
Entry Scanning	✅ Active	MA20, RSI, volume criteria
Sentiment Integration	✅ Active	Veto at -0.6, boost at +0.5
Trade State Machine	✅ Active	SETUP → FILLED → HOLDING → CLOSED
Catalyst Scoring	✅ Active	0-100 points for position sizing
Exit Monitoring	✅ Active	Stop/target detection

Overview¶

The Strategy Agent scans for entry candidates, evaluates technical setups, generates trading signals, and manages trade state machines.

src/agents/trading/strategy_agent.py

Responsibilities: - Scan for entry candidates (stocks meeting criteria) - Evaluate trend following setup (MA20, RSI, volume) - Generate entry/exit signals with sentiment weighting - Manage trade state machines - Monitor open positions for exit conditions - Score catalysts for position sizing

Strategy: Trend Following (3:1 R/R)¶

Entry Criteria¶

Criterion	Requirement	Points
Price > MA20	Uptrend confirmation	Required
RSI 50-70	Momentum without overbought	1/3
Volume > 1.5x avg	Confirmation of interest	1/3
ADX > 25	Strong trend (alternative to RSI)	1/3

Setup Valid: At least 2 of 3 criteria met

Exit Targets¶

Exit Type	Calculation	Description
Stop Loss	Entry × 0.95	5% below entry
Profit Target	Entry × 1.15	15% above entry
Risk/Reward	3:1	Minimum ratio

Event Architecture¶

Events Published¶

Event	Description	Data
`strategy.candidate_found`	Potential entry detected	symbol, criteria_met
`strategy.entry_signal`	Entry criteria met	symbol, entry_price, stop_price, target_price, sentiment_data
`strategy.exit_signal`	Exit condition met	symbol, reason, pnl_pct
`market_data.get_quote`	Request quote data	symbol
`market_data.get_bars`	Request historical bars	symbol, timeframe, limit

Events Subscribed¶

Event	Handler	Description
`strategy.scan`	`handle_strategy_scan`	Scan universe for candidates
`strategy.evaluate`	`handle_strategy_evaluate`	Evaluate specific symbol
`position.opened`	`handle_position_opened`	Track new position
`market_data.quote.updated`	`handle_market_data_quote_updated`	Monitor for exits
`sentiment.score`	`handle_sentiment_score`	Receive sentiment data

Sentiment Integration¶

How Sentiment Affects Trading¶

Sentiment Score	Action	Impact
< -0.6 (bearish)	VETO entry	Trade blocked
-0.6 to +0.5	Normal	No adjustment
> +0.5 (bullish)	BOOST size	+25% position

Combined Score Calculation¶

technical_score = (criteria_met / 3) * 100  # 0-100
sentiment_adjustment = sentiment_score * 20 * confidence  # -20 to +20
combined_score = technical_score + sentiment_adjustment

Configuration¶

self.sentiment_veto_threshold = -0.6    # Block entries below this
self.sentiment_boost_threshold = 0.5    # Boost size above this
self.sentiment_boost_multiplier = 1.25  # 25% larger position
self.sentiment_cache_ttl_seconds = 300  # 5 minute cache

Trade State Machine¶

States¶

stateDiagram-v2
    [*] --> SETUP_DETECTED: detect_setup()
    SETUP_DETECTED --> ENTRY_APPROVED: approve_entry()
    ENTRY_APPROVED --> ORDER_PLACED: place_order()
    ORDER_PLACED --> FILLED: filled()
    FILLED --> HOLDING: set_holding()
    HOLDING --> EXIT_TRIGGERED: trigger_exit()
    EXIT_TRIGGERED --> CLOSED: close()
    CLOSED --> [*]

    SETUP_DETECTED --> [*]: invalid_setup()
    ENTRY_APPROVED --> [*]: order_rejected()
    ORDER_PLACED --> [*]: order_cancelled()

State Transitions¶

From	To	Trigger	Data
SETUP_DETECTED	ENTRY_APPROVED	`approve_entry()`	quantity
ENTRY_APPROVED	ORDER_PLACED	`place_order()`	order_id
ORDER_PLACED	FILLED	`filled()`	fill_price
FILLED	HOLDING	`set_holding()`	-
HOLDING	EXIT_TRIGGERED	`trigger_exit()`	reason
EXIT_TRIGGERED	CLOSED	`close()`	exit_price

Catalyst Scoring System¶

Score tickers based on multiple catalysts (0-100 points).

Catalyst Points¶

Catalyst	Points	Criteria
FDA Approval	25	Within 90 days, >50% probability
Earnings Beat	25	30-60 days out, >10% surprise expected
M&A Activity	20	13D filings, rumors, strategic buyers
Product Launch	15	Announced within 60 days
Insider Buying	15	>10% shares bought in 90 days
Analyst Upgrade	15	Initiation, upgrade, or target raise >20%
Technical Breakout	10	52-week high, volume surge

Usage¶

score = agent.score_catalyst_stack("AAPL", {
    "earnings_date": datetime(2025, 1, 30),
    "earnings_surprise_est": 0.15,  # 15% expected beat
    "has_13d_filing": False,
    "has_ma_rumors": True,
    "analyst_upgrades": 2,
    "current_price": 175.50,
    "price_52w_high": 180.00,
    "volume_ratio": 3.5
})

# score = 60 (M&A + Analyst + Technical)

Entry Requirement¶

Minimum 2-3 catalysts (score > 50)
Higher scores = larger position sizes

Error Handling¶

Sentiment API Failures¶

Error Type	Fallback Behavior	Timeout
API timeout	Use cached sentiment	30s
API unavailable	Default to neutral (0.0)	3 retries
Parse error	Log error, use neutral	None
Rate limit	Use cached data	60s backoff

def get_sentiment_with_fallback(self, symbol):
    try:
        sentiment = self.sentiment_api.get(symbol, timeout=30)
        self.sentiment_cache[symbol] = sentiment
        return sentiment
    except (TimeoutError, ConnectionError):
        cached = self.sentiment_cache.get(symbol)
        if cached and not self._is_cache_expired(cached):
            return cached
        return {"score": 0.0, "confidence": 0.0}  # Neutral fallback

Market Data Failures¶

Error Type	Action	Recovery
Quote unavailable	Skip symbol in scan	Retry next cycle
Historical data missing	Use available bars	Continue with partial data
Price feed delay	Use last known price	Mark as stale

Performance Metrics¶

Strategy Performance (Backtest 2024)¶

Metric	Value	Benchmark
Win Rate	68%	>60% target
Avg R/R	2.8:1	3:1 target
Max Drawdown	-12%	<15% limit
Sharpe Ratio	1.45	>1.0 target
Total Return	+34%	SPY: +26%

Real-Time Monitoring¶

# Performance tracking
self.metrics = {
    "signals_generated": 0,
    "signals_executed": 0,
    "sentiment_vetos": 0,
    "avg_processing_time_ms": 0,
    "cache_hit_rate": 0.0
}

Alerts Configuration¶

Alert	Threshold	Action
Processing time > 5s	Critical	Notify DevOps
Cache hit rate < 80%	Warning	Check Redis
Sentiment API failures > 10%	Warning	Check API status

Usage Examples¶

Initialize Agent¶

from src.agents.trading.strategy_agent import StrategyAgent

agent = StrategyAgent(
    redis_host="172.200.3.164",
    redis_port=6379,
    redis_db=0,
    sentiment_api_url="https://api.sentiment.corbie.ai",
    cache_ttl=300
)

agent.start()

Scan Universe¶

agent.publish_event(
    event="strategy.scan",
    data={
        "symbols": ["AAPL", "MSFT", "GOOGL", "NVDA", "TSLA"]
    }
)

Evaluate Specific Symbol¶

agent.publish_event(
    event="strategy.evaluate",
    data={
        "symbol": "AAPL",
        "current_price": 175.50,
        "ma20": 172.00,
        "rsi": 58,
        "volume": 85000000,
        "avg_volume": 50000000
    }
)

# If criteria met, publishes:
# strategy.entry_signal with:
# - entry_price: 175.50
# - stop_price: 166.73 (5% below)
# - target_price: 201.83 (15% above)
# - risk_reward_ratio: 3.0
# - sentiment_score, sentiment_confidence
# - position_size_multiplier

Get Active Trades¶

trades = agent.get_active_trades()
# Returns Dict[str, TradeStateMachine] for HOLDING/EXIT_TRIGGERED states

for symbol, trade in trades.items():
    print(f"{symbol}: {trade.state.value}")
    print(f"  Entry: ${trade.data.entry_price:.2f}")
    print(f"  Stop: ${trade.data.stop_price:.2f}")

Entry Signal Flow¶

graph TD
    A[strategy.evaluate] --> B{Price > MA20?}
    B -->|No| C[Reject: Below MA]
    B -->|Yes| D{RSI 50-70?}
    D --> E{Volume > 1.5x?}
    E --> F{2+ Criteria Met?}
    F -->|No| G[Reject: Insufficient Criteria]
    F -->|Yes| H{Get Sentiment}
    H --> I{API Available?}
    I -->|No| J[Use Cache/Neutral]
    I -->|Yes| K{Score < -0.6?}
    J --> K
    K -->|Yes| L[VETO: Bearish Sentiment]
    K -->|No| M{Score > +0.5?}
    M -->|Yes| N[Apply 1.25x Multiplier]
    M -->|No| O[Normal Size]
    N --> P[Publish Entry Signal]
    O --> P

Exit Signal Monitoring¶

The agent monitors positions for:

Condition	Action
Price >= Target	Publish `strategy.exit_signal` with reason `TARGET_HIT`
Price <= Stop	Publish `strategy.exit_signal` with reason `STOP_HIT`

def handle_market_data_quote_updated(self, message):
    if last_price >= target_price:
        trade.trigger_exit(reason="TARGET_HIT")
        self.publish_event("strategy.exit_signal", {...})

    elif last_price <= stop_price:
        trade.trigger_exit(reason="STOP_HIT")
        self.publish_event("strategy.exit_signal", {...})

Configuration Reference¶

Parameter	Type	Default	Description
`redis_host`	str	"172.200.3.164"	Redis host for message bus
`redis_port`	int	6379	Redis port
`redis_db`	int	0	Redis database number
`sentiment_api_url`	str	None	Sentiment service endpoint
`cache_ttl`	int	300	Cache TTL in seconds
`max_concurrent_evaluations`	int	10	Parallel evaluation limit
`processing_timeout_ms`	int	5000	Max processing time

Strategy Parameters¶

Parameter	Default	Description
`stop_loss_pct`	0.05	5% stop loss
`profit_target_pct`	0.15	15% profit target
`target_risk_reward`	3.0	Minimum R/R ratio
`rsi_min`	50	Minimum RSI for entry
`rsi_max`	70	Maximum RSI for entry
`adx_min`	25	Minimum ADX for entry
`volume_multiplier`	1.5	Volume vs average threshold

Sentiment Parameters¶

Parameter	Default	Description
`sentiment_veto_threshold`	-0.6	Block entries below this
`sentiment_boost_threshold`	0.5	Boost position above this
`sentiment_boost_multiplier`	1.25	25% position increase
`sentiment_cache_ttl_seconds`	300	5 minute cache
`sentiment_timeout_seconds`	30	API call timeout
`sentiment_retry_attempts`	3	Max retry attempts

Troubleshooting¶

Common Issues¶

Issue	Symptoms	Solution
No signals generated	Agent running but silent	Check market data feed
High sentiment API failures	Warnings in logs	Verify API endpoint and auth
Redis connection errors	Agent crashes	Check Redis connectivity
Slow processing	Timeouts in logs	Scale Redis or reduce scan universe

Debug Commands¶

# Check agent health
agent.get_health_status()

# View sentiment cache
agent.get_sentiment_cache_stats()

# Monitor active trades
agent.get_trade_summary()

# Performance metrics
agent.get_performance_metrics()

Log Analysis¶

# Strategy agent logs
grep "strategy_agent" /var/log/corbie/agents.log

# Sentiment failures
grep "sentiment.*error" /var/log/corbie/agents.log

# Performance issues
grep "timeout\|slow" /var/log/corbie/agents.log

Integration Testing¶

With Risk Management Agent¶

def test_integration_risk_management():
    # Strategy generates signal
    strategy_agent.publish_event("strategy.entry_signal", signal_data)

    # Risk management validates
    assert risk_agent.last_validation_result["approved"] == True

    # Integration verified
    assert strategy_agent.pending_signals == 0

With Market Data Agent¶

def test_market_data_integration():
    # Request market data
    strategy_agent.request_quote("AAPL")

    # Verify data received
    assert market_data_agent.quote_requests > 0
    assert strategy_agent.last_quote_received is not None

Logging Examples¶

Entry Signal¶

INFO Entry signal: AAPL @ $175.50 | Stop: $166.73, Target: $201.83 | R/R: 3.00:1 | Sentiment: +0.35

Sentiment Veto¶

WARNING SENTIMENT VETO: AAPL entry blocked due to bearish sentiment (-0.72, 85% confidence)

Sentiment Boost¶

INFO SENTIMENT BOOST: AAPL position size +25% due to bullish sentiment (+0.65)

Target Hit¶

INFO Target hit: AAPL @ $201.83 (target: $201.83)

Stop Hit¶

INFO Stop hit: AAPL @ $166.73 (stop: $166.73)

Error Handling¶

ERROR Sentiment API timeout for AAPL, using cached data (5 minutes old)
WARNING High processing time: 4.2s for NVDA evaluation (limit: 5s)

Strategy Agent¶

Overview¶

Strategy: Trend Following (3:1 R/R)¶

Entry Criteria¶

Exit Targets¶

Event Architecture¶

Events Published¶

Events Subscribed¶

Sentiment Integration¶

How Sentiment Affects Trading¶

Combined Score Calculation¶

Configuration¶

Trade State Machine¶

States¶

State Transitions¶

Catalyst Scoring System¶

Catalyst Points¶

Usage¶

Entry Requirement¶

Error Handling¶

Sentiment API Failures¶

Market Data Failures¶

Performance Metrics¶

Strategy Performance (Backtest 2024)¶

Real-Time Monitoring¶

Alerts Configuration¶

Usage Examples¶

Initialize Agent¶

Scan Universe¶

Evaluate Specific Symbol¶

Get Active Trades¶

Entry Signal Flow¶

Exit Signal Monitoring¶

Configuration Reference¶

Strategy Parameters¶

Sentiment Parameters¶

Troubleshooting¶

Common Issues¶

Debug Commands¶

Log Analysis¶

Integration Testing¶

With Risk Management Agent¶

With Market Data Agent¶

Logging Examples¶

Entry Signal¶

Sentiment Veto¶

Sentiment Boost¶

Target Hit¶

Stop Hit¶

Error Handling¶

See Also¶