volatility pairs trading new york group 13 gaurav gandhi palash kasodhan david lin
DESCRIPTION
Volatility Pairs Trading New York Group 13 Gaurav Gandhi Palash Kasodhan David Lin Michael Rehwinkel. Outline. Overview Implied Volatility Time Series Proof of Concept Trading Implementation Results Conclusion. Overview. Correlation of Implied Volatilities - PowerPoint PPT PresentationTRANSCRIPT
Volatility Pairs Trading
New York Group 13
Gaurav Gandhi
Palash Kasodhan
David Lin
Michael Rehwinkel
Outline
• Overview
• Implied Volatility Time Series
• Proof of Concept
• Trading Implementation
• Results
• Conclusion
Overview
• Correlation of Implied Volatilities
• Reversion of correlated options
• Proof of Concept
• Implementation using straddles
Implied Volatility Time Series
• Goal: Time series of implied volatility of At-the-money options for chosen names
• Dataset
– Oil Industry: SIC 13
– CRSP for equity prices
– OptionMetrics for option prices
• Time series creation
– Java Application
– Only 90 of 214 tickers remained due to missing data
– Interpolation for up to 5 days of missing data
Proof of Concept
• Pair most correlated for all 90 tickers
• Trade At-the-money implied volatility directly
• Top 10 correlated Pairs
Correlation/Trading Profit
0
0.2
0.4
0.6
0.8
1
1.2
0.88 0.89 0.9 0.91 0.92 0.93 0.94 0.95Correlation
Tra
din
g P
rofi
t
Trading Implementation
• At-the-money Straddles
– No Greek hedging
• Normalize data to calculate simple ratio
• Parameters/Thresholds:
– For entering position
– For exiting position (for profit/bailout & forced exercise)
– Tailing days for moving average
• Client exercising
Cross Validation
Results
• Formation Period in 2007 H1 and trading in 2007 H2crit 0.3
bail 0.5
extrigger 0.33
ttexp 60
norm.days 30
c Total PNL
1.5 5.08536568
2 7.434366373
2.5 3.669043944
3 1.260628469
Example Pair CNQ/DO (using cutoff =2)OpenDate CloseDate x.position x.strike x.expiry x.openVal x.closeVal x.pnl y.strike
20070709 20070719 1 70 20070922 1 1.337179487 0.337179487 105
20070802 20070815 1 70 20071222 1 1.163492063 0.163492063 100
20070824 20070904 -1 65 20071222 1 1.093137255 -0.093137255 100
20070921 20071029 -1 75 20071222 1 1.108910891 -0.108910891 115
20071102 20071114 1 80 20080119 1 1.202654867 0.202654867 110
20071120 20071130 -1 70 20080119 1 0.925531915 0.074468085 110
20071210 20071221 -1 65 20080322 1 1.068181818 -0.068181818 125
20071226 20071231 1 70 20080322 1 0.961904762 -0.038095238 140
y.Expiry y.openVal y.closeVal y.pnl position.pnl days.open no.data.close f.exercise bail reached.crit.pnl
20070922 1 0.945736434 0.054263566 0.391443053 10 0 0 0 1
20071222 1 0.778801843 0.221198157 0.38469022 13 0 0 0 1
20071222 1 1.462416107 0.462416107 0.369278852 11 0 0 0 1
20071222 1 0.873333333 -0.126666667 -0.235577558 38 0 1 0 0
20080119 1 0.872222222 0.127777778 0.330432645 12.04166667 0 0 0 1
20080119 1 1.096710526 0.096710526 0.171178611 10 0 1 0 0
20080322 1 1.702487562 0.702487562 0.634305744 11 0 0 0 1
20080322 1 1.014218009 -0.014218009 -0.052313248 5 0 0 0 0
Example Pair CNQ/DO (using cutoff =2)
Position PnL Distribution with Cutoff
cut-off = 1.5 cut-off = 2
Position PnL Distribution with Cutoff
cut-off = 2.5 cut-off = 3
Number of Days Position Open
• Volatility Pairs Trading can be profitable, as the results indicate• The cut off parameter has the maximum influence on PnL• We need a high bailout to account for 4 option positions• Bid – Ask Spreads eat into potential profits. Other transaction costs
not even considered.• Not hedged perfectly (unlike trading IV directly) as the following risk
remains– Residual delta risk– Net Theta decay as the long and short straddle wont cancel theta
completely
Conclusions
• Greek Hedging - Delta & Theta
• Synthesize Variance Swaps using Listed Options
• Dynamic Hedging vs Static Hedging
• Dispersion Trading: for e.g. buying the volatility of an index and selling the volatility of its constituents
• Use Street Events Feed – To identify when the volatility pattern could be broken by extra-ordinary news
Next Steps