#!/bin/bash
set -e

echo "========================================="
echo "PPO Hyperopt Deployment"
echo "========================================="
echo ""

# Set PYTHONPATH
export PYTHONPATH="/home/jgrusewski/Work/foxhunt:$PYTHONPATH"

# Activate venv
source .venv/bin/activate

# Deploy PPO hyperopt
# Note: hyperopt_ppo_demo already supports separate policy_learning_rate and value_learning_rate
python3 scripts/runpod_deploy.py \
    --gpu-type "RTX A4000" \
    --image "jgrusewski/foxhunt-hyperopt:latest" \
    --command "hyperopt_ppo_demo --parquet-file /runpod-volume/test_data/ES_FUT_180d.parquet --trials 50 --episodes 2000 --base-dir /runpod-volume/ml_training/ppo_hyperopt --early-stopping-min-epochs 50"

echo ""
echo "✅ PPO hyperopt deployment initiated"
echo "Monitor logs: python3 scripts/python/runpod/monitor_logs.py <pod_id>"
echo ""
echo "Previous Results (2025-11-01):"
echo "  Duration: 14.3 minutes (99.8% faster than 18-24 hour estimate)"
echo "  Cost: \$0.06 (98.7% cheaper than \$4.50-\$6.00 estimate)"
echo "  Trials: 63 completed (26% bonus over 50 target)"
echo "  Best Trial #1 (objective: 2.4023):"
echo "    • policy_learning_rate: 1.0e-06 (ultra-conservative)"
echo "    • value_learning_rate: 0.001 (aggressive)"
echo "    • clip_epsilon: 0.1126"
echo "    • entropy_coeff: 0.006142"
echo "    • value_loss_coeff: 0.5"
echo ""
echo "Expected duration: 10-20 minutes (historical)"
echo "Expected cost: \$0.04-\$0.08 @ \$0.25/hr (RTX A4000)"
echo ""
echo "Success Criteria:"
echo "  - Policy loss decreasing over episodes"
echo "  - Separate policy_lr and value_lr optimized by Optuna"
echo "  - Entropy coefficient balanced (0.006-0.01 range)"
echo "  - Final backtest: > 10% return, Sharpe > 1.5"