Value-at-Risk
Published Last updated
Key Takeaway
Risk metric quantifying the maximum expected loss at a specified confidence level over a defined time period, answering how much capital you can lose with specified probability.
Learn These First
What Is Value-at-Risk?
Risk metric quantifying the maximum expected loss at a specified confidence level over a defined time period, answering how much capital you can lose with specified probability.
How Value-at-Risk Works
Frequently Asked Questions
What does a 95% VaR of $50,000 actually mean?
It means that 95% of the time, your portfolio losses won't exceed $50,000 over the specified period (daily, weekly, or monthly). Conversely, 5% of the time, losses might exceed $50,000. For a 95% daily VaR of $50,000, expect that approximately one trading day per 20 (5% of 20 days), losses might exceed $50,000. For monthly 95% VaR of $50,000, approximately once per year, monthly losses might exceed $50,000. It's a probability statement: the loss threshold you can be 95% confident won't be exceeded. It doesn't describe what happens when threshold is exceeded; it only identifies the threshold itself. If losses do exceed threshold, they might be $51,000 or $500,000—VaR doesn't specify.
How should I use VaR to size positions in my crypto portfolio?
Calculate your portfolio's daily 95% VaR, determining acceptable maximum daily loss. Then size positions such that this VaR represents your maximum acceptable daily risk. Example: $100,000 account, acceptable daily risk 2% ($2,000), calculate portfolio 95% daily VaR; if VaR is $5,000, reduce position sizes until VaR declines to $2,000. This ensures daily losses you can expect at 95% confidence remain survivable. Additionally, many traders set intra-day stop losses below VaR thresholds, accepting small losses automatically to prevent larger VaR-level losses. Use VaR as position-sizing constraint: calculate VaR, reduce positions if VaR exceeds comfort level, maintain positions if VaR remains acceptable.
Why is VaR insufficient for crypto risk management?
Crypto exhibits fat-tail distributions—extreme losses occur more frequently than normal distribution predicts. A 95% VaR calculated assuming normal distribution dramatically underestimates actual tail risk. Additionally, crypto experiences unprecedented events (exchange collapses, regulatory shocks, technological crises) that exceed any historical VaR threshold. Finally, VaR only describes normal-market conditions; under stressed conditions (market crashes, liquidity disappearance), losses frequently exceed VaR significantly. Professional crypto risk managers supplement VaR with stress testing (calculate losses if Bitcoin declines 50%), scenario analysis (losses from extreme regulatory action), and stress scenarios capturing extreme possibilities. VaR provides useful baseline; it's insufficient as sole risk measure. Treat VaR as floor, not ceiling, for expected losses in crypto.
Common Misconceptions About Value-at-Risk
My 95% VaR is $50,000, so my maximum possible loss is $50,000.
VaR represents the threshold at 95% confidence, not absolute maximum loss. The 5% of time that losses exceed VaR might experience losses exceeding $50,000 substantially. Crypto has experienced 80-90% account drawdowns during extreme events—far exceeding calculated VaR thresholds. VaR is floor, not ceiling, for expected losses. Treating VaR as loss ceiling creates false confidence, leading to excessive leverage under assumption 'VaR prevents catastrophe.' Cryptographic tail events (exchange hacks, regulatory shutdowns) can produce losses far exceeding VaR. Supplement VaR with stress testing (calculate losses under extreme scenarios) to reveal true maximum loss potential.
VaR accurately measures crypto portfolio risk because it's statistically rigorous.
Statistical rigor in VaR calculation doesn't prevent misapplication to crypto. VaR assumes past distributions persist—an invalid assumption for crypto experiencing regime shifts and unprecedented events. Crypto's fat-tail distributions mean actual extreme losses occur more frequently than normal-distribution VaR predicts. Parametric VaR (assuming normal distribution) particularly underestimates crypto tail risk. Historical-simulation VaR is better but still limited to historical experience; unprecedented events exceed historical VaR. Sophisticated mathematics doesn't prevent fundamental limitation: VaR measures past, not future. Use VaR as useful baseline risk metric while acknowledging its limitations for crypto where unexpected events frequently occur.
If I maintain portfolio VaR below my risk tolerance threshold, I'm adequately protected.
VaR compliance provides baseline discipline but insufficient protection. Extreme events (tail scenarios outside VaR confidence interval) frequently strike crypto despite VaR-compliant sizing. Additionally, VaR depends on historical correlations remaining stable; during crises, correlations increase, amplifying losses beyond VaR calculations. Smart traders use VaR as starting point, not endpoint: maintain VaR compliance while also implementing stress testing, position limits, kill switches, and diversification. VaR plus comprehensive risk management provides real protection; VaR alone creates false confidence leading to excessive risk-taking.