guardrailed-llm-agent — Step 5: The LLM Risk Manager: Building the Policy Table (snippet 1)

step_5_the_llm_risk_manager_building_the_1.py

def state_stats(train_df):
    tmp = train_df.copy()
    tmp['state_lag'] = tmp['state'].shift(1)   # use yesterday's state
    tmp['ret_fwd']   = tmp['ret']              # to predict today's return
    tmp = tmp.dropna(subset=['state_lag', 'ret_fwd'])

    g = tmp.groupby('state_lag')['ret_fwd']
    stats = pd.DataFrame({
        'count':      g.size(),
        'mean':       g.mean(),
        'std':        g.std(ddof=0),
    })
    # Sharpe-like: annualized signal-to-noise ratio per state
    stats['sharpe_like'] = (
        stats['mean'] / (stats['std'] + 1e-12)
    ) * math.sqrt(252)
    return stats.sort_values('count', ascending=False)