QUANTITATIVE ANALYTICS

Quantitative Analytics Suite

Every metric institutional desks use. P&L attribution, alpha decomposition, Kelly sizing, risk matrices, Monte Carlo forward simulation, and execution quality -- all computed in real time.

P&L ATTRIBUTION

Know Where Every Dollar Came From

Decompose your returns by strategy, market category, time period, and individual agent. Instantly see which strategies carry the portfolio and which drag it down. Attribution updates tick-by-tick as positions resolve.

  • Strategy-level P&L with daily, weekly, monthly rollups
  • Market category breakdown (politics, sports, crypto, macro)
  • Per-agent contribution with win rate and edge metrics
pnl_attribution.json LIVE
// Portfolio P&L — Last 30 Days
{
  "total_pnl": +$4,281.40,
  "by_strategy": {
    "momentum":   +$2,140.20  // 50.0%
    "sentiment":  +$1,830.50  // 42.8%
    "contrarian": +$ 620.70  // 14.5%
    "mean_revert": -$ 310.00  // -7.2%
  },
  "by_market": {
    "politics": +$2,450.00,
    "crypto":   +$1,120.30,
    "sports":   +$ 890.10,
    "weather":  -$ 179.00
  }
}
alpha_report.out COMPUTED
$ yorn analytics alpha --window 90d

BENCHMARK COMPARISON
─────────────────────────────────────
Portfolio Return:    +18.4%
Kalshi Avg Return:   +6.2%
Alpha (excess):      +12.2%
Information Ratio:   2.14

SIGNAL SCORING (top 5)
─────────────────────────────────────
1. news_catalyst     Score: 0.84  Decay: 4.2h
2. momentum_12h      Score: 0.76  Decay: 8.1h
3. poll_divergence    Score: 0.71  Decay: 18h
4. whale_tracking     Score: 0.68  Decay: 2.1h
5. twitter_sentiment  Score: 0.42  Decay: 0.8h

DECAY DETECTION
─────────────────────────────────────
WARNING: twitter_sentiment alpha
decayed 61% over past 14d.
Recommendation: reduce weight or retire.
ALPHA GENERATION

Measure the Edge That Matters

Compare returns against Kalshi market averages, random walk baselines, and custom benchmarks. Signal decay detection monitors each alpha source over time, alerting you when a previously profitable signal is losing predictive power.

  • Benchmark comparison: market avg, random walk, custom
  • Signal scoring with predictive power ranking
  • Automatic decay detection with retirement alerts
2.14 Information Ratio -- 2.14x return per unit of tracking error vs. benchmark
KELLY CRITERION SIZING

Position Sizes You Can Audit

Full Kelly, fractional Kelly, and drawdown-adjusted Kelly -- computed in real time for each eligible trade. Ruin probability monitoring and drawdown adjustments help keep sizing decisions inside your configured risk limits.

  • Full Kelly for maximum geometric growth rate
  • Fractional Kelly (quarter, half, three-quarter)
  • Drawdown-adjusted sizing with ruin probability guard
kelly_sizing.json LIVE
// Position Sizing — Active Trade
{
  "market": "Will Fed cut rates in Q2?",
  "estimated_edge": 0.142,
  "market_price": 0.58,
  "model_prob": 0.722,
  "kelly_full": {
    "fraction": 0.246,
    "position": $246.00,
    "growth_rate": 1.74%
  },
  "kelly_half": {
    "fraction": 0.123,
    "position": $123.00,
    "growth_rate": 1.31%
  },
  "drawdown_adjusted": {
    "current_dd": -8.4%,
    "dd_multiplier": 0.72,
    "adj_position": $89.00
  },
  "ruin_probability": {
    "at_full_kelly": 4.2%,
    "at_half_kelly": 0.3%,
    "at_dd_adjusted": 0.1%
  }
}
risk_dashboard.out LIVE
MetricValueLimitStatus
Daily VaR (95%)-$312-$500OK
Daily CVaR (99%)-$487-$750OK
Max Drawdown-8.4%-15%OK
Current Drawdown-3.1%-15%OK
Correlation Max0.820.70WARN
RegimeTRENDING (0.74 conf)--
correlation_matrix 4x4
POL SPT CRY MAC
POL    1.00   0.12   0.31   0.82
SPT    0.12   1.00   -0.08  0.15
CRY    0.31   -0.08  1.00   0.44
MAC    0.82   0.15   0.44   1.00

ALERT: POL-MAC correlation 0.82 exceeds
0.70 threshold. Consider reducing overlap.
RISK ANALYTICS

Institutional-Grade Risk Management

Correlation matrices detect hidden concentration risk. Regime detection identifies trending, mean-reverting, or crisis markets and adjusts risk parameters automatically. VaR and CVaR at multiple confidence levels with drawdown circuit breakers.

  • Correlation matrices with auto concentration alerts
  • Regime detection: trending, mean-revert, crisis
  • VaR/CVaR at 95% and 99% with real-time updates
  • Drawdown circuit breakers with automatic de-risking
BACKTESTING & MONTE CARLO

Replay the Past, Simulate the Future

Historical backtesting replays your strategy against every resolved market -- no lookahead bias, no survivorship bias. Monte Carlo runs 10,000 forward paths with confidence intervals on returns, drawdowns, and probability of hitting your target.

  • Historical replay on all resolved markets
  • 10,000 forward Monte Carlo simulations
  • Confidence intervals at 50th, 75th, 95th, 99th percentiles
monte_carlo_10k.out COMPLETE
$ yorn backtest --strategy momentum_v3 \
  --start 2024-01-01 --monte-carlo 10000

HISTORICAL BACKTEST (24 months)
─────────────────────────────────────
Total Trades:    1,847   Win Rate: 58.3%
Avg Return:      +2.31%  Sharpe:   1.89
Max Drawdown:    -14.2%  Recovery: 18d

MONTE CARLO FORWARD (10K paths, 90d)
─────────────────────────────────────
Percentile   Return    MaxDD
───────────────────────────────
1st         -22.4%    -31.0%
5th         -8.1%     -18.7%
25th        +4.2%     -9.4%
50th        +12.8%    -6.1%
75th        +21.6%    -4.2%
95th        +38.4%    -2.8%
99th        +52.1%    -1.9%

Prob(profit in 90d): 71.4%
Prob(ruin):         0.4%
execution_quality.json LIVE
MetricThis MonthAvgImpact
Total Trades342----
Avg Slippage-0.8c-1.2c-$27.36
Platform Fees$171.00$158.40-3.99%
Fill Rate97.4%95.8%IMPROVED
Net Edge After Costs+8.2%+7.6%IMPROVED
optimization_hints 3 SUGGESTIONS
1. Batch small positions (<$15) to reduce
   per-trade fee ratio. Est. savings: $24/mo

2. Shift 18% of volume to limit orders.
   Est. slippage reduction: -0.4c avg

3. Avoid first 30s after market open --
   slippage is 3.2x higher. Saves: $18/mo
SLIPPAGE & FEE IMPACT

Every Cent Leaking From Your Edge

Track execution quality across every trade. Measure slippage between signal price and fill price, compute fee drag as a percentage of gross P&L, and get concrete optimization suggestions with estimated dollar savings.

  • Per-trade slippage measurement (signal vs. fill)
  • Fee drag computation with net-of-cost edge
  • Automated optimization suggestions with $ estimates
$42/mo Estimated savings from execution optimization suggestions
SESSION COMPARISON

Compare Everything Against Everything

Period-over-period analysis, strategy A/B testing with statistical significance, and regime-conditional comparisons. See how strategies behave differently in trending vs. mean-reverting vs. crisis environments.

  • Period vs. period with delta highlighting
  • Strategy A/B testing with significance tests
  • Regime-conditional performance breakdowns
session_compare.out COMPUTED
MetricFeb 2026Mar 2026Delta
Net P&L+$1,840+$2,441+32.7%
Win Rate55.1%59.8%+4.7pp
Avg Edge8.4c11.2c+2.8c
Max DD-12.1%-8.4%+3.7pp
Sharpe1.421.89+0.47
ab_test_result.json SIGNIFICANT
// A/B Test: momentum_v2 vs momentum_v3
{
  "variant_a": "momentum_v2",
  "variant_b": "momentum_v3",
  "return_a": +9.8%,
  "return_b": +14.2%,
  "p_value": 0.023,
  "significant": true,
  "regime_breakdown": {
    "trending":    v3 +6.1% vs v2 +3.8%,
    "mean_revert": v3 +1.2% vs v2 +2.4%,
    "crisis":      v3 -2.1% vs v2 -5.8%
  }
}

Beyond basic backtesting

Institutional-grade analysis tools that separate signal from noise.

Walk-Forward Analysis

5 rolling windows, 70/30 train-test split. Measures degradation between in-sample and out-of-sample performance to detect overfitting before it costs you money.

Correlation Matrix

Strategy-to-strategy correlation heatmap. Auto-alerts when hidden concentration risk exceeds thresholds. Diversification scoring for your agent fleet.

6-Factor Attribution

Decompose returns by time-of-day, spread, volume, price level, momentum, and liquidity. Know exactly which factors drive your alpha and which are noise.

4-Regime Analysis

High-vol, normal, low-vol, and trending regimes. Per-regime win rate, PnL, and trade count so you know when your strategy thrives and when to sit out.

Parameter Optimization

Automated parameter sweep with suggested ranges for lookback periods, thresholds, and position sizing. Impact scoring tells you which params matter most.

Fee Impact Breakdown

Gross vs. net PnL waterfall. Exchange fees, spread costs, and settlement fees separated. See your fee-adjusted Sharpe ratio and break-even trade count.

Seven Pillars of Quantitative Analysis

Everything you need to run a data-driven prediction market portfolio.

P&L
Attribution by strategy, market, period
Alpha
Benchmark comparison, signal scoring
Kelly
Optimal sizing with ruin guards
Risk
VaR, CVaR, correlations, regimes

Stop Guessing. Start Measuring.

Portfolio analytics, risk metrics, execution quality, and agent comparisons without needing to maintain a separate quant stack.