This is the link for an abstract. The research is very similar to what I have been conducting with my Fair Volatility Estimate model in that VIX Exchange Trade Products trading has shown tremendous profit potential. His research is conducted in a much more organized manner than mine.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2255327
Upon closer reading, some strategies presented in the paper would not make viable "trading systems". One of the strategies has not shown a trade since late 2011. Such few trades does not a trading
system make... Good read, though...
(this post was adjusted on August 22nd to correct for some errors)
You read the article inaccurately. It is his system #3 (his second best) that hasn't had a trade since 10/10/11. His #4 VRP system has had 10 switches since then. Also, why would a low number of trades negate it as a trading system??? Sounds good to me.
ReplyDeleteAlso, to answer your question above, I believe any model or trading strategy would have to stand the test of time AND # of observations. Otherwise, the rules runs the risk of applying hindsight to an uncertain and changing future with possible outcomes being dramatically different than past results.
DeleteJohn, I think we are communicating back and forth on seekingalpha.com so I will just repost what I commented there.
ReplyDeleteYes, you are right. What I am meaning to say is that I would take the trading rules & backtested results with heavy dose of skepticism. I believe Tony recognizes this as well by attaching the Grim Reaper on backtested results. The research is spot on.
If I may give one example, I have one "trading rule" that if I take out 1 short VIX futures trade entered before the May 6, 2010 flash crash, it's CAGR would have been 182.9% with max drawdown of 39.9% since Jan 2008. This is inclusive of $8.3 million of slippage & commission costs. A hypothetical $100k account would have returned $36 million in 5.66 years. I could argue that a flash crash like the one in 2010 would not happen again to that extent, but the fact is that it did happen. Including the 1 trade, the numbers change to 156.1% CAGR with 77% max dd. A hypothetical $100k account shows ending balance of $20.5 million in this particular simulation.
Perhaps you are knowledgeable enough to understand that a small change in parameters like # of days used in a moving average or even slightly lower volatility risk premiums would dramatically change the performance when dealing with compounded growth #s.
I am skeptical of my own research as it pertains to simulated results (and my simulated results are superior to ones provided in the abstract), therefore, I cannot help but be skeptical of the simulated results in the abstract.
The abstract organizes the research in a more disciplined manner and points out the risks and opportunities with great clarity. This is why I have posted the abstract in my blog.
John, if you are the same person in SA, I thank you for introducing the abstract.