Intelligence
Portfolio Analytics
10 min read
What it is
Institutional-grade analytics for your crypto portfolio: Sharpe / Sortino / VaR / Drawdown, stress testing, correlation matrix, factor exposure, Kelly Criterion sizing, pairs trading signals, whale watcher, and yield opportunities. 8 tabs, every metric computed from real fills or real on-chain data — no fabricated numbers.
How to think about it
Tabs that need user trading data (Risk Overview, Stress Test, Correlation, Factor Exposure, Kelly, Pairs) show an empty-state card with paper-trading + bot-setup CTAs until you have enough fills. Whale Watcher + Yield are external-data tabs and work immediately. Pairs Trading is EXPERIMENTAL — requires an explicit acknowledgement.
Step-by-step
-
Risk Overview tab — 10 metrics + AI commentary
Sharpe (return per unit of risk, good >1), Sortino (downside-only, good >1.5), Max Drawdown (worst peak-to-trough, low <15%), Calmar (return ÷ DD, good >1), VaR 95%/99% (max daily loss at confidence level), Win Rate, Profit Factor (gross profit ÷ gross loss, good >1.5), Annualized Return + Volatility. Plus per-source breakdown (Smart Trade / Grid / DCA / CEX / DEX / Futures / Algo).
-
Stress Test tab — real historical scenarios
7 pre-built scenarios run against your current allocation: BTC -30% Crash, ETH +40% Rally, DeFi Exploit, USDT Depeg, Gas Spike 500gwei, March 2020 Replay, Bull Run +100%. Per-position impact bars + total portfolio change. The scenarios use REAL historical data — no synthetic "what if" generation (durable rule from the audit).
-
Correlation tab — concentration in disguise
N×N matrix of correlations between your holdings. Pairs > 0.85 are flagged as concentration risk — they don't add diversification. Pairs 0.5–0.85 = partial diversification. < 0.5 = meaningfully independent. < 0 = inversely correlated (rare in crypto majors).
-
Factor Exposure tab
8 factors: Momentum, Volatility, Market Cap, DeFi, Layer 1, Layer 2, Meme, Staking Yield. Bar chart shows your exposure to each. > 75% in one factor triggers a concentration warning. Use to spot when your "diversified" portfolio is actually one big bet on a single theme.
-
Kelly Criterion tab
Computes optimal position size from your real win rate + avg win/avg loss. Shows Full Kelly (aggressive — historical only, NEVER use live), Half Kelly (academic standard), Quarter Kelly (recommended for crypto). The page also surfaces honest caveats when Kelly is undefined (negative expectancy, too few trades, etc).
-
Pairs Trading tab — EXPERIMENTAL
Auto-detected mean-reversion opportunities (e.g., "ETH/BTC spread is 2.3σ above mean"). Z-score, correlation, signal, confidence level. Requires explicit warning checkbox before use — substantial losses possible if correlation breaks during your hold period.
-
Whale Watcher tab — external on-chain data
Smart money accumulation per token (7-day net flow), top 10 most profitable wallets with their P&L and recent actions, recent whale transactions > $500K with smart money tagging. Always live (no user-data dependency).
-
Yield tab — live DeFi rates
Live APY / TVL / chain / risk rating for opportunities across Aave, Compound, Lido, Rocket Pool, Marinade. Filterable by chain. Always live — pulled from each protocol's on-chain state.
Tips & pitfalls
- Empty tabs are NOT a bug — they need real trades to compute. Paper trade first if you don't want to use real capital while you learn the metrics.
- AI Portfolio Commentary appears at the top of Risk Overview once you have enough data — auto-generated narrative about your performance with best/worst movers and risk flags.
- Sharpe and Sortino assume normal-distribution returns. Crypto returns have fat tails, so these metrics over-state risk-adjusted returns. Treat them as relative comparisons across your own strategies, not absolute benchmarks.
- VaR (95%/99%) is a per-day loss estimate. The 99% number is the loss your portfolio should NOT exceed on 99% of days — but it WILL be exceeded on 1% of days. Don't bet the farm against it.
- Pairs trading correlation can break — sometimes spectacularly. The "ETH/BTC spread 2σ above mean" trade lost money for everyone during the May 2022 LUNA collapse. Use small size and accept it can be wrong.
- Kelly Criterion is the math-optimal size IF your historical edge persists. In crypto it usually doesn't. Quarter Kelly bakes in a 4× safety margin and is the right starting point.
- Stress test uses REAL historical periods, not synthetic generated scenarios — see [[no-synthetic-scenarios]] for the philosophy. Want a custom period? Use the date-range picker in Stress Test.