Real-World Use Cases

# Compare different market periods
crisis_results <- run_annem_analysis(
  symbols = c("SPY", "VIX", "GLD"),
  n_agents = 500,
  n_steps = 100
)

# Analyze agent performance during high volatility
performance_analysis <- crisis_results$agent_performance
high_volatility_agents <- performance_analysis[performance_analysis$sharpe_ratio > 0.5, ]

# Multi-asset portfolio analysis
portfolio_symbols <- c("AAPL", "MSFT", "GOOGL", "TSLA", "NVDA", "JPM", "JNJ")
portfolio_results <- run_annem_analysis(
  symbols = portfolio_symbols,
  n_agents = 1000,
  n_steps = 250
)

# Identify best performing agent types for allocation
best_agents <- portfolio_results$agent_performance %>%
  group_by(agent_type) %>%
  summarise(avg_return = mean(total_return), avg_sharpe = mean(sharpe_ratio))

# Study social influence vs. fundamental analysis
social_heavy <- list(
  social_network = 0.6,      # High social influence
  fundamentalist_ml = 0.2,
  neural_momentum = 0.1,
  contrarian_ai = 0.1
)

fundamental_heavy <- list(
  fundamentalist_ml = 0.6,   # High fundamental focus
  social_network = 0.1,
  neural_momentum = 0.2,
  contrarian_ai = 0.1
)

# Compare outcomes
social_results <- run_annem_analysis(agent_distribution = social_heavy)
fundamental_results <- run_annem_analysis(agent_distribution = fundamental_heavy)

# Study network evolution patterns
market <- create_annem_market(n_agents = 200)
results <- market$run_simulation(n_steps = 500)

# Analyze network metrics over time
network_evolution <- market$analyze_network_evolution()
network_plots <- plot_network_evolution(network_evolution)

# Calculate information flow efficiency
network_metrics <- calculate_network_metrics(market$network)

# Stress test with different market conditions
stress_symbols <- c("SPY", "TLT", "GLD", "VIX")
stress_results <- run_annem_analysis(
  symbols = stress_symbols,
  n_agents = 1000,
  n_steps = 100
)

# Calculate risk metrics
risk_metrics <- calculate_performance_metrics(stress_results$simulation_results$actual_returns)
max_drawdown <- risk_metrics$max_drawdown
var_95 <- quantile(stress_results$simulation_results$actual_returns, 0.05)

# Develop trading strategy based on meta-learning agents
results <- run_annem_analysis(
  symbols = c("AAPL", "MSFT"),
  agent_distribution = list(meta_learning = 1.0),  # Only meta-learning agents
  n_agents = 100
)

# Extract decision patterns for strategy development
meta_decisions <- results$simulation_results$agent_decisions
strategy_signals <- apply(meta_decisions, 1, mean)  # Aggregate decisions

# Comprehensive model comparison
symbols <- c("SPY", "EFA", "EEM", "AGG")
results <- run_annem_analysis(symbols = symbols, n_agents = 500)

# Get benchmark comparison
comparison <- results$benchmark_comparison
print(comparison)

# Calculate improvement metrics
annem_mse <- comparison[comparison$Model == "ANNEM", "MSE"]
var_mse <- comparison[comparison$Model == "VAR", "MSE"]
improvement <- (var_mse - annem_mse) / var_mse * 100
cat("ANNEM improves MSE by", round(improvement, 1), "% vs VAR")