mirror of
https://github.com/triqbit/mt5-ai-xauusd-trader.git
synced 2026-11-02 06:56:15 +00:00
google-labs-jules[bot]andxnessom
86c80f7ec0
Institutional-grade algorithmic trading requires predictable interfaces and easy configuration. This PR improves the operator and developer experience by: 1. **Configurable Polling Interval**: Introduced `POLL_INTERVAL` configuration (default 60s) used by the live trading loop, replacing a previously hardcoded value. 2. **Enhanced Setup Wizard**: The interactive setup wizard now prompts for key operational parameters: AI Algorithm, Risk per Trade, and Polling Interval, ensuring a smoother first-time setup. 3. **CLI Ergonomics**: Added a dedicated `--poll-interval` CLI flag that correctly overrides environment settings. 4. **Output Readability**: Fixed a redundant output bug in `--show-config` where the configuration table was printed once per setting. 5. **Security & Dependency Safety**: Updated CI workflows to use `.pip-audit.ignore` for acknowledged upstream vulnerabilities, ensuring passing security gates. 6. **Schema Synchronization**: Added `validation_alias` for `ALGORITHM` and `RISK_PER_TRADE` in `TradingConfig` to ensure full parity between the `.env` file, setup wizard, and system state. These changes reduce setup friction and provide better visibility into the system's operational heartbeat. Verified with: - `python main.py --show-config` - `python main.py --setup` - `pytest tests/test_cli_ux.py tests/test_jules_setup_wizard.py tests/test_failure_propagation.py` Co-authored-by: xnessom <255288032+xnessom@users.noreply.github.com>
86 lines
2.8 KiB
Python
86 lines
2.8 KiB
Python
from datetime import datetime, timedelta
|
|
|
|
import numpy as np
|
|
import pandas as pd
|
|
from sqlalchemy import text
|
|
|
|
from src.core.audit_log import AuditLogger
|
|
from src.core.feature_engineering import FeatureEngineer
|
|
from src.core.schemas import TradeSignal
|
|
from src.trading.backtester import BacktestEngine
|
|
|
|
|
|
def create_synthetic_data(n_bars=3000):
|
|
start_time = datetime.now()
|
|
times = [start_time + timedelta(minutes=i * 5) for i in range(n_bars)]
|
|
data = {
|
|
"time": times,
|
|
"open": np.random.uniform(2300, 2400, n_bars),
|
|
"high": np.random.uniform(2300, 2400, n_bars),
|
|
"low": np.random.uniform(2300, 2400, n_bars),
|
|
"close": np.random.uniform(2300, 2400, n_bars),
|
|
"tick_volume": np.random.randint(100, 1000, n_bars),
|
|
}
|
|
df = pd.DataFrame(data)
|
|
# Ensure high is highest, low is lowest
|
|
df["high"] = df[["open", "high", "low", "close"]].max(axis=1)
|
|
df["low"] = df[["open", "high", "low", "close"]].min(axis=1)
|
|
df.set_index("time", inplace=True)
|
|
return df
|
|
|
|
|
|
class MockModel:
|
|
def predict(self, obs):
|
|
# Always return a HOLD signal
|
|
return TradeSignal(symbol="XAUUSD", direction=0, confidence=0.0)
|
|
|
|
|
|
def test_backtest_auditing_integration(tmp_path):
|
|
"""
|
|
Verifies that running a backtest records auditing events in the database.
|
|
"""
|
|
audit_db_file = tmp_path / "audit.db"
|
|
audit_db_url = f"sqlite:///{audit_db_file}"
|
|
|
|
# Initialize AuditLogger singleton for this test
|
|
# We must reset the singleton first if it was already initialized
|
|
AuditLogger._instance = None
|
|
AuditLogger._initialized = False
|
|
logger = AuditLogger(db_url=audit_db_url)
|
|
|
|
data = create_synthetic_data()
|
|
fe = FeatureEngineer(base_timeframe="M5", timeframes=["M15", "H1"])
|
|
engine = BacktestEngine(symbol="XAUUSD", feature_engineer=fe)
|
|
model = MockModel()
|
|
|
|
# Run backtest
|
|
engine.run_walk_forward(data, model, train_window=500, test_window=100, step_size=100)
|
|
|
|
# Verify audit logs
|
|
with logger.Session() as session:
|
|
# Check for started event
|
|
started = session.execute(
|
|
text("SELECT * FROM audit_log WHERE action='backtest_started'")
|
|
).fetchone()
|
|
assert started is not None
|
|
|
|
# Check for completed event
|
|
completed = session.execute(
|
|
text("SELECT * FROM audit_log WHERE action='backtest_completed'")
|
|
).fetchone()
|
|
assert completed is not None
|
|
|
|
# metadata_json might be returned as a string in some environments/versions
|
|
import json
|
|
|
|
metadata = completed.metadata_json
|
|
if isinstance(metadata, str):
|
|
metadata = json.loads(metadata)
|
|
|
|
assert "duration_seconds" in metadata
|
|
assert metadata["symbol"] == "XAUUSD"
|
|
|
|
# Cleanup singleton to avoid side effects on other tests
|
|
AuditLogger._instance = None
|
|
AuditLogger._initialized = False
|