DeFi Trading Engine

Autonomous DeFi trading bot with self-improving review system. Scans for opportunities, executes trades, logs performance, and learns from mistakes.

When to Use

Apply this skill when:

• - Setting up automated crypto trading on Base or other EVM chains
• Building a self-improving trading system
• Implementing systematic DeFi trading strategies
• Executing DCA, momentum, or mean reversion strategies
• Reviewing and optimizing trading performance
• Managing trading risk (position sizing, drawdown limits)
• Integrating with Bankr CLI or other DEX tools

Architecture

Self-Improvement Loop:

CODEBLOCK0

Components:

1. 1. Token Scanner (scan-tokens.py) — Finds trading opportunities
2. Risk Manager (risk-manager.py) — Enforces position limits and risk rules
3. Trade Executor (trade-executor.py) — Executes trades via Bankr CLI
4. Daily Review (daily-review.py) — Analyzes performance and suggests improvements
5. Config File (trading-config.json) — Central configuration for all parameters

Quick Start

1. Setup

Create workspace:

CODEBLOCK1

Copy skill scripts:

CODEBLOCK2

2. Configure

Create trading-config.json:

CODEBLOCK3

3. Setup Bankr (if needed)

See references/bankr-setup.md for Bankr CLI setup.

4. Run the Trading Loop

Manual execution:

CODEBLOCK4

Automated loop (cron):

CODEBLOCK5

Core Scripts

scan-tokens.py

Scans for trading opportunities using free APIs.

Data Sources:

• - CoinGecko trending coins
• Volume spikes (24h volume vs 7d average)
• Price momentum (1h, 4h, 24h trends)
• Liquidity and market cap filters

Output (candidates.json):
CODEBLOCK6

Usage:

python3 scan-tokens.py --output candidates.json --min-score 7.0

---

risk-manager.py

Enforces risk limits before every trade. Acts as the gatekeeper.

Checks:

• - Position size within limit
• Max active positions not exceeded
• Daily trade limit not exceeded
• Cooldown period respected
• Max drawdown not breached

Usage:
CODEBLOCK8

Exit Codes:

• - 0 — Trade approved
• INLINECODE9 — Trade denied (prints reason)

Example Output:

✅ Risk check passed
  - Position size: $40 (limit: $40)
  - Active positions: 3 (limit: 5)
  - Daily trades: 5 (limit: 8)
  - Cooldown: OK (35 minutes since last trade)
  - Drawdown: 8.5% (limit: 15%)

---

trade-executor.py

Executes trades via Bankr CLI (or generic DEX interface).

Supported Actions:

• - buy — Market buy
• INLINECODE11 — Market sell
• INLINECODE12 — Limit order buy
• INLINECODE13 — Limit order sell
• INLINECODE14 — Stop-loss order
• INLINECODE15 — Take-profit order

Usage:
CODEBLOCK10

Trade Log (trades/YYYY-MM-DD.json):

[
  {
    "timestamp": "2026-03-13T15:45:00Z",
    "symbol": "SOL",
    "action": "buy",
    "amount_usd": 40,
    "price": 145.5,
    "quantity": 0.275,
    "tx_hash": "0xabc123...",
    "status": "success",
    "take_profit_price": 151.32,
    "stop_loss_price": 133.86
  }
]

---

daily-review.py

Analyzes trade history, calculates P&L, identifies weaknesses, and suggests parameter adjustments.

Metrics Calculated:

• - Total P&L (realized + unrealized)
• Win rate (% of profitable trades)
• Average win vs average loss
• Sharpe ratio (if enough data)
• Max drawdown
• Best/worst trades

Output (reviews/review-YYYY-MM-DD.md):

INLINECODE18

Usage:

python3 daily-review.py --start-date 2026-03-01 --end-date 2026-03-13

---

Configuration Reference

Risk Parameters

Parameter	Default	Purpose
INLINECODE19	40	Max $ per trade
INLINECODE20

4 | Exit when +4% gain | | stop_loss_pct | 8 | Exit when -8% loss | | max_active_positions | 5 | Max concurrent positions | | max_daily_trades | 8 | Max trades per day | | cooldown_minutes | 30 | Wait time between trades | | max_drawdown_pct | 15 | Stop trading if down 15% |

Strategy Parameters

Parameter	Options	Purpose
INLINECODE26	INLINECODE27, mean_reversion, dca, INLINECODE30	Strategy type
INLINECODE31

volume_spike_and_price_up, oversold, breakout | Entry condition | | exit_signal | take_profit_or_stop_loss, reversal, time_based | Exit condition | | timeframe | 5min, 15min, 1h, 4h | Trading timeframe |

Bankr Integration

Parameter	Default	Purpose
INLINECODE44	INLINECODE45	EVM chain (base, ethereum, polygon)
INLINECODE46

trading-wallet | Bankr wallet name | | slippage_pct | 1.5 | Max acceptable slippage |

---

Strategy Templates

See references/strategies.md for detailed strategy implementations:

1. 1. DCA (Dollar-Cost Averaging) — Buy fixed amount on schedule
2. Momentum Swing — Ride short-term momentum with tight stops
3. Mean Reversion — Buy dips, sell rallies
4. Asymmetric Bets — Small positions on high-upside opportunities

---

Risk Management Rules

The risk manager enforces these rules:

Position Sizing

CODEBLOCK14

Active Position Limit

CODEBLOCK15

Daily Trade Limit

CODEBLOCK16

Cooldown Period

CODEBLOCK17

Max Drawdown Circuit Breaker

CODEBLOCK18

When max drawdown is hit, all trading stops until manually reset.

---

Self-Improvement Process

The bot learns from performance:

1. Daily Review
Run daily-review.py to analyze trades.

2. Pattern Recognition
Identify which setups worked:

• - Entry conditions with >70% win rate
• Tokens with consistent performance
• Timeframes with best risk/reward

3. Parameter Adjustment
Update trading-config.json based on findings:

• - Tighten filters if win rate < 60%
• Adjust position size if drawdown too high
• Change timeframe if signals lag

4. Backtest Changes
Test new parameters on historical data (manual or automated).

5. Monitor
Run new parameters for 7 days, then review again.

Cycle: Weekly reviews → Parameter tweaks → Monitor → Repeat

---

Safety Features

✅ DO:

• - Start with small position sizes ($40 default)
• Use stop-losses on every trade
• Respect cooldown periods (avoid overtrading)
• Run daily reviews to catch bad patterns early
• Keep max drawdown limit low (15% default)
• Paper trade first (simulate without real funds)

❌ DON'T:

• - Disable risk manager checks
• Increase position size without testing
• Remove stop-losses ("this time is different")
• Trade during network congestion (high gas fees)
• Ignore max drawdown signals
• Use leverage (this engine is spot-only by design)

---

Monitoring & Alerts

Track bot health:

Check active positions:
CODEBLOCK19

Check today's P&L:
CODEBLOCK20

Alert on max drawdown:

# Add to cron (every hour)
python3 risk-manager.py --action check_drawdown && echo "Trading halted: max drawdown exceeded"

---

Troubleshooting

Problem: Risk manager denies all trades

Solution: Check trading-config.json limits. May have hit daily trade limit or max drawdown.

---

Problem: Trades execute but P&L is negative

Solution: Run daily-review.py to identify losing patterns. Tighten entry filters or adjust stop-loss.

---

Problem: Bankr CLI errors

Solution: Check wallet balance, network connection, and gas fees. See references/bankr-setup.md.

---

Problem: Scanner returns no candidates

Solution: Lower min_score threshold or relax liquidity filters.

---

Advanced Features

Paper Trading Mode

Test strategies without real funds:

CODEBLOCK22

All trades simulate execution, no real transactions.

Multi-Strategy Support

Run multiple strategies in parallel:

CODEBLOCK23

Backtesting

Test parameters on historical data (requires historical price data):

CODEBLOCK24

(Backtest script not included — implement based on your data source)

---

Resources

• - Bankr Setup: INLINECODE56
• Strategy Templates: INLINECODE57
• CoinGecko API: https://www.coingecko.com/en/api/documentation
• Base Chain Docs: https://docs.base.org

---

Version: 1.0 Last Updated: 2026-03-13 Security Note: Store API keys and wallet private keys securely. Never commit to Git.