risk-mgmt

Portfolio Analytics — using all 8 tabs without getting lost

Sharpe / Sortino / VaR / Stress / Correlation / Factors / Kelly / Pairs / Whales / Yield — one tab at a time

12 min · intermediate

What you'll have when finished

  • Read every Risk Overview metric and know what thresholds mean "good" vs "concerning"
  • Run a real stress test against historical periods (not generated scenarios)
  • Spot hidden concentration via correlation + factor exposure
  • Size positions with Kelly without blowing up
  • Use Whale Watcher as decision support, not decision generation

Before you start

  • Empty tabs are not bugs — they need real trades or open positions to compute. Paper trade first if you don't want to risk capital while learning.
  • Pairs Trading is EXPERIMENTAL and requires explicit acknowledgement. Correlations break — sometimes catastrophically.
  • Sharpe / Sortino assume normal-distribution returns. Crypto has fat tails. Treat these as relative comparisons, not absolute hedge-fund benchmarks.

Walkthrough

  1. Tab 1 — Risk Overview: the 10 metrics that matter

    Each metric is color-coded with thresholds: **Sharpe Ratio** (good >1, excellent >2) — return per unit of total volatility. Below 0.5 = your edge isn't covering risk. **Sortino Ratio** (good >1.5) — same as Sharpe but only penalizes DOWNSIDE volatility. Crypto tends to look better on Sortino than Sharpe. **Max Drawdown** (low <15%, concerning >25%) — worst peak-to-trough loss. If this is >50%, your strategy survived a 2× recovery from the bottom — which is rare. **Calmar Ratio** (good >1) — annualized return ÷ max drawdown. Penalizes deep drawdowns even if recovery happened. **VaR 95% / 99%** — max daily loss at confidence level. The 99% is the loss you should NOT exceed 99% of days, but the 1% of days are when accounts blow up. Don't bet against it. **Win Rate** — % of trades that closed in profit. 40-60% is normal for active trading. Lower than 40% is fine IF profit factor compensates. **Profit Factor** (good >1.5) — gross profit ÷ gross loss. 1.0 = break-even pre-fees. <1 = losing strategy. **Annualized Return / Volatility** — projected yearly numbers from your historical sample. Volatility >80% annualized is high even for crypto. Per-source breakdown at the bottom shows these per Smart Trade / Grid / DCA / CEX / DEX. Use it to find which surface is dragging your overall numbers.

    Success criteria: Read every metric on the page · Identified which threshold band each falls in · Spotted which per-source line is dragging the total

  2. Tab 2 — Stress Test: real historical periods, not generated scenarios

    7 pre-built scenarios run against YOUR current allocation using REAL historical price data from each period: - **BTC -30% Crash** — what would have happened to your positions if today's portfolio existed during the March 2020 crash - **ETH +40% Rally** — counterfactual gains during May 2021 alt season - **DeFi Exploit** — re-runs the Curve hack contagion - **USDT Depeg** — re-runs the May 2022 USDT wobble - **Gas Spike 500gwei** — re-runs Q4 2021 NFT mint chaos - **March 2020 Replay** — full COVID crash period - **Bull Run +100%** — Q4 2020 melt-up Per-position impact bars show which holdings would have lost most. Total portfolio change at the bottom. **Important:** these use real historical price series, not generated scenarios. If you want a custom period, use the date-range picker — it pulls real OHLCV for any window. No synthetic "what if BTC fell 50% in a day" simulations — that policy is durable. **Action rule:** if any single scenario causes >30% portfolio loss, rebalance toward lower-correlation assets.

    Success criteria: Ran all 7 pre-built scenarios · Picked one custom historical period · Identified your worst-loss scenario

  3. Tab 3 — Correlation: concentration in disguise

    N×N matrix of correlations between your holdings over a 90-day window. **> 0.85** — two assets effectively move as one position. Holding both adds no diversification, only complexity. Replace one with an uncorrelated asset. **0.5 - 0.85** — related but partially independent. Some diversification benefit. **< 0.5** — meaningfully different exposures. Real diversification. **< 0** — inversely correlated. Excellent hedge characteristic. In crypto, most majors are 0.7+ correlated to BTC. ETH/SOL/AVAX/LINK is mostly the SAME bet on "L1 risk-on" — not a portfolio. True diversification needs lower-correlation assets: stablecoins, gold-pegged tokens, or short positions. **Action rule:** if any 2 of your top-5 holdings have correlation > 0.85, replace one.

    Success criteria: Identified your highest-correlation pair · Decided whether one should be replaced

  4. Tab 4 — Factor Exposure: are you betting on one theme?

    8 factors your portfolio could be loaded on: - **Momentum** — last 30-day price action bias - **Volatility** — high-vol vs low-vol assets - **Market Cap** — large vs small cap exposure - **DeFi** — DeFi protocol tokens (UNI, AAVE, CRV, etc.) - **Layer 1** — L1 chain tokens (ETH, SOL, AVAX, etc.) - **Layer 2** — L2 chain tokens (ARB, OP, MATIC, etc.) - **Meme** — memecoin exposure (DOGE, SHIB, PEPE, etc.) - **Staking Yield** — yield-bearing tokens (stETH, rETH, mSOL, etc.) **Action rule:** > 75% exposure to any single factor = concentration risk. The market can rotate against entire factors (e.g., "DeFi summer" → "DeFi winter"). Diversify across at least 3 factors at >15% each.

    Success criteria: Identified your top-2 factor exposures · Verified at least 3 factors are above 15%

  5. Tab 5 — Kelly Criterion: math-optimal position sizing

    Computes optimal trade size given your historical win rate and avg win / avg loss. **Formula:** Kelly% = WinRate − (1 − WinRate) / (AvgWin / AvgLoss) **Full Kelly** = the math-optimal size if your historical edge persists exactly. NEVER use this live. It assumes infinite trials, no fat tails, no edge decay. **Half Kelly** = the academic standard. Reduces blow-up risk while still growing aggressively. **Quarter Kelly (recommended for crypto)** = bakes in a 4× safety margin against fat tails and edge decay. Most professional crypto traders converge here. **When Kelly is undefined** the page shows the reason: negative expectancy (lose more than you win on average), too few trades (<30 minimum for any statistical confidence), or no closed positions yet. Don't bypass these warnings. **Action rule:** size at Quarter Kelly. If Quarter Kelly says >10% of bankroll per trade, your sample is too small or your edge is illusory — be skeptical.

    Success criteria: Computed your Quarter Kelly size · Checked if Kelly is defined or undefined and why

  6. Tab 6 — Pairs Trading: EXPERIMENTAL, read the warning

    Auto-detected mean-reversion opportunities: "ETH/BTC spread is 2.3σ above its 90-day mean. If you believe the ratio reverts, short ETH and long BTC equal notional." Each opportunity shows z-score, correlation, signal direction, confidence level. Filtered to your holdings when you have positions; otherwise shows the universe. **Why EXPERIMENTAL:** correlations break. Sometimes they break because one asset has fundamentally re-rated (e.g., a chain rolls out a major upgrade). When a pair you're shorting suddenly becomes uncorrelated, BOTH legs can lose simultaneously. **Action rule:** if you take a pair trade, size it at <5% of bankroll. Set a stop-out z-score (e.g., exit if z-score goes from 2.3 to >3.5 or back to <0.5). Never average down on a pair trade. The famous "LTCM blow-up" was a pair-trading shop. Their math was right; their assumption that spreads always revert was wrong. Use this lesson.

    Success criteria: Read the EXPERIMENTAL warning · Picked a small starter position size · Set a stop-out plan in advance

  7. Tab 7 — Whale Watcher: decision support, not decision generation

    External on-chain data — works immediately, no user data required. **Smart Money Accumulation** — per-token 7-day net flow by labeled whale wallets. Direction (accumulation/distribution), whale count, signal (Strong/Mild Buy/Sell), confidence. **Top Smart Money Wallets** — top 10 most profitable wallets ranked by 30-day P&L. Their win rate, trades, top holdings, most recent action. **Recent Whale Transactions** — large trades > $500K with type, USD value, anonymized addresses (non-exchange wallets), smart money tag. **Critical caveat:** by the time you see a whale trade on-chain, the price has often already moved. Whales' real edge is being at the venue first; retail watching them is always second. **Use it as:** thesis validation/invalidation. If you're considering buying X and Whale Watcher shows smart money distributing X for 7 straight days, that's a signal to dig deeper, not to ignore. **Don't use it as:** primary signal generation. The "smart money" label is heuristic — some "smart money" wallets are actually market makers or MEV bots whose strategies aren't copyable.

    Success criteria: Looked up your top holding's smart money signal · Recognized this is decision support, not a buy signal

  8. Tab 8 — Yield: live DeFi rates

    Live APY / TVL / chain / risk-rating for major DeFi protocols: Aave, Compound, Lido, Rocket Pool, Marinade. Filterable by chain. **Risk ratings:** - **Low Risk** — major audited blue chips (Aave v3, Compound). Smart contract risk dominates; minimal economic risk. - **Medium Risk** — newer protocols, less battle-tested, or higher dependency on external systems. - **High Risk** — leveraged yield, recursive borrowing, anything with looping. **Important:** APY shown is current. It can drop the next block. Always check 7-day average separately. **Common mistakes:** - Chasing 50%+ APY on small protocols (almost always token-emission Ponzi) - Ignoring the TVL drop signal (large drops = users exiting, often before exploits) - Forgetting bridges (yields on Avalanche-Aave look great but you need to bridge USDC there + back, paying gas + slippage) **Action rule:** keep DeFi positions in protocols with TVL > $500M and at least 12 months of public on-chain history.

    Success criteria: Filtered to your active chain · Identified at least one low-risk opportunity · Read the risk rating before clicking through

What's next

Most users converge to this routine: **Weekly (5 min):** check Risk Overview Sharpe + Win Rate + Per-source breakdown. Identify drift early. **Monthly (20 min):** 1. Run Stress Test against your current allocation 2. Check Correlation matrix — has any pair drifted above 0.85? 3. Check Factor Exposure — has any single factor crossed 75%? 4. Recompute Quarter Kelly with the new month's data 5. Journal one anomalous trade per [Trading Journal](/journal) **Quarterly (60 min):** 1. Full strategy review using [AI Strategy Coach](/strategy-coach) 2. Replace correlated holdings 3. Re-stress against a new custom period if market regime shifted