Drawdown
Quick Reference
Perdita cumulativa dal picco precedente. Misura quanto capitale perdi durante losing streaks.
Definizione
Drawdown = (Peak Value - Current Value) / Peak Value
Maximum Drawdown = Peggiore drawdown mai sperimentato in un periodo.
Esempio
Capital progression:
$100,000 → $120,000 (new peak) → $90,000 (drawdown) → $130,000 (new peak)
Drawdown = ($120,000 - $90,000) / $120,000 = 25%
Quando raggiungi $130,000, drawdown torna a 0%.
Tipi di Drawdown
Maximum Drawdown (Max DD)
Peggiore calo peak-to-trough nella storia:
Max DD = Worst (Peak - Trough) / Peak
Esempio: Strategy con max DD 40% ha perso 40% del capital nel worst period.
Average Drawdown
Media di tutti i drawdown periods: - Più rappresentativo di typical experience - Max DD può essere outlier
Current Drawdown
Drawdown attuale dal most recent peak: - 0% = sei a new high - >0% = sei underwater
Drawdown vs Volatility
Non sono la stessa cosa!
Volatility: Misura quanto returns variano Drawdown: Misura cumulative losses
Esempio: - Alta volatilità + positive drift = drawdown contenuti - Bassa volatilità + negative skew = drawdown catastrofici
Expected Drawdowns
Per Volatility Target
Con SR 0.5, zero skew, target volatility:
| Vol Target | Typical Max DD | 10% Prob DD |
|---|---|---|
| 12.5% | -15% | -9% |
| 25% | -30% | -19% |
| 50% | -60% | -38% |
| 100% | -90% | -62% |
Lesson: Higher vol target = much deeper drawdowns!
Per Skew Type
Stesso SR 0.5, vol 25%:
Zero skew: - Typical max DD: -30% - Smooth recovery
Negative skew: - Typical max DD: -35% - Sharp drops, slow recovery
Positive skew: - Typical max DD: -35% - Gradual decline, sharp recovery
Recovery Time
Time to new high dopo drawdown:
Formula approssimata:
Recovery Time ≈ DD% / (SR × Vol Target)
Esempio (30% DD, SR 0.5, Vol 25%):
Recovery ≈ 30% / (0.5 × 25%) = 2.4 anni
Long time!
Psychological Impact
Typical Reactions
10% DD: Uncomfortable ma manageable 20% DD: Dubbi sulla strategia 30% DD: Strong urge to quit 40%+ DD: Panic, likely abandon system
This is normal human behavior!
Managing Psychology
- Know expected DD beforehand: No surprises
- Size positions appropriately: Lower vol = smaller DD
- Track average DD, not just max
- Plan response in advance: "At -30% I will review, not panic-exit"
Drawdown e Sharpe Ratio
Higher SR non garantisce lower DD:
Strategy A: SR 0.8, negative skew - Average DD: -20% - Max DD: -60% (rare tail event)
Strategy B: SR 0.4, positive skew - Average DD: -25% - Max DD: -35%
Strategy B psychologically easier despite lower SR!
Underwater Period
Time spent below previous peak:
Anche con profitable system: - 30-50% del tempo underwater typical - Long periods senza new highs normal
Don't panic se sei underwater per mesi.
Drawdown Limits
Hard stops (not recommended): - Exit system at -X% DD - Problem: Spesso esci al bottom, miss recovery
Soft monitoring: - Review system at -X% DD - Check se qualcosa broken - But: Don't auto-exit based on DD alone
Reducing Drawdowns
Lower Volatility Target
Vol 50% → 25% = Halve expected drawdowns
Trade-off: Halve expected returns too!
Positive Skew Strategies
Trend following, breakout: - Smoother DD profiles - Easier psychologically
Diversification
Multiple uncorrelated strategies: - Reduce DD substantially - Best risk/return improvement
Drawdown Control
Dynamic position sizing può aiutare:
Bad idea: Cut size when in DD (lock in losses) Good idea: Cut size when volatility spikes (prevents future DD)
Difference crucial!
Monitoring Drawdown
Track: 1. Current DD 2. Time underwater 3. Average DD (rolling) 4. Max DD to date
Compare to expected DD from backtests.
Red flag: Actual DD >> Expected DD suggerisce problema.
Drawdown Examples
2008 Financial Crisis
Typical quant fund: - Max DD: -40% to -60% - Recovery: 2-4 years
Lesson: Even good systems have severe DD.
COVID March 2020
Equity long-only: - Max DD: -35% - Recovery: 6 months (V-shaped)
Lesson: Fast DD can have fast recovery.
Calmar Ratio
Alternative performance metric:
Calmar = Annual Return / Max DD
Focuses on DD instead of volatility.
Problema: Based su single worst event (noisy).
Errori Comuni
- Equating DD to risk: Vol è better continuous measure
- Panic at expected DD: -30% DD può essere normale per strategy
- Comparing only max DD: Average DD more representative
- Ignoring time underwater: Long periods normale
- Stop-loss based on DD: Usually exit at worst time
- Assume small vol = small DD: Negative skew can cause large DD anche con low vol
Concetti Correlati
- [[Skew]] - negative skew aumenta DD risk
- [[Volatility Targeting]] - riduce DD volatility
- [[Kelly Criterion]] - optimal sizing considera DD probability
- [[Risk Adjusted Returns]] - DD alternative metric