Backtesting delta-hedged volatility strategies

A

ajensen

To harvest the volatility risk premium in SPY, you could sell 1-month ATF SPY straddles, delta hedge daily at the close during the month, and repeat the process monthly. I have done such analyses in the past, and it's a lot of work, because you need to

(1) buy a set of option data
(2) find the the closest-to-the-forward options on the trade date
(3) find those option prices daily in the data file
(4) compute the delta daily and compute the number of SPY shares to use in hedging
(5) compute the profit of the hedged position.

Is there a web site that lets you type in a few parameters, such as
(1) underlying symbol
(2) tenor
(3) moneyness
(4) option structure (call, put, or straddle)
(5) delta hedging frequency

and does the simulation for you? It would be worth paying for. Ultimately you may want to do the analysis yourself, but having a 3rd party simulator could save you time researching strategies that don't work.
 
Very very hard to do. There was a futures guy (W. Gallacher?)who did it years back and put results in a book by selling at ATM crude oil straddle every day to (dis)prove the myth that sellers have an edge over buyer . The end result was that he proved that the edge is to seller BUT very slight. Even if you were able to simulate that.. at best it is a lone candle in the dark approach. Better than complete darkness but a single candle nonetheless. The reason is your hedging interval.Another reason is the ATM straddle moves! One missed hedge and your returns would differ markedly . I am sure the susquehannas of the world can do it but that is w extreme quant resources. I can do it thru the sale of "strikeless straddles" which hold spot price as the strike so if SPY is at $235.75, I can back into the Feb 235.75 straddle but it is a HUGE project - gather the 2 ATM quotes, map into strikeless straddles, save into text file, load into backtester.OMG.!! susquehanna budget... Better surrogate? sell an ATM fly daily and let it expire since it is a fixed loss spread (ie no adjustments which is so so "lecture circuit" .. don't get me started...lol ) , let your terminal distro approximate your pnl then you are good to go. .. for now..
 
EdgeQuest said:
Very very hard to do. There was a futures guy (W. Gallacher?)who did it years back and put results in a book by selling at ATM crude oil straddle every day to (dis)prove the myth that sellers have an edge over buyer . The end result was that he proved that the edge is to seller BUT very slight.
More...

Yes, I read it. The book was "The Options Edge: Winning the Volatility Game with Options On Futures".
 
An amazing piece of work.. book out of print .. I had to buy mine from a public library in indiana! BUT if you noticed he did not delta hedge.. his pnl was the ending distro , hence he had some huge winner and losses there I think. this is what inspired me to backtest a "butterfly garden" to , using your words, harvest.. This method+ John Bender's piece on trading the prob distro COMBINED is very powerful...
 
EdgeQuest said:
Lastly the purchase of the 2 wings will obviate the need for delta hedging which greatly simplifies your BTest. Instead on focusing on the hedges, focus on the technical setups that "predict" quiet markets so you can harvest away. Check out this page on my site.It has the exact thing you seek.. I think...

http://edgequestllc.com/case-study-a-market-wizard-s-approach-to-trading-options
More...
It's a different trade, dude. Delta hedged fixed strike has a very different return profile then a butterfly. It's especially true in the world where wings are generally trading at a premium to the ATM.
 
I know SLE.. I did both for a decade on the floor where I was delta hedging neg gamma on certain months and isolating flies out of the mish mash of positions. The OP wanted to find out simulation of that strategy and IMHO, it is very hard to do vs. applying a surrogate strategy of using flies which are easier to simulate..far from perfect..even if he can simulate the delta strategy, the simulations will still have tons of drift bec of hedge frequency.
 
if your main goal is to test the vol risk premium, then the first order approximation is to test it on a variance swap (instead of straddles). and if you do backtests on equity indices, most data are available for free
 
spec77 said:
if your main goal is to test the vol risk premium, then the first order approximation is to test it on a variance swap (instead of straddles). and if you do backtests on equity indices, most data are available for free
More...
Why do you think variance swap is a better proxy for checking volatility risk premium? I would say it's not true.

PS. really, equity index variance swap data is available for free? Unless you are talking about the few VIX-like indices out there, it's hard to find reliable variance swap data even if you are willing to pay for it.
 
You must log in or register to reply here.
Back
Top