I am not in disagreement with you. I wanted to make sure the OP understood that a successful backtest is not necessarily the road to riches, and he/she needs to do more work, and in particular understand the risks involved, validity of assumptions in backtest, etc, etc. Its like when I see backtests done on a non tradeable instrument by some Muppet, like a cash index. It's over before it even began, I don't even bother critiquing their "stragegy" further after such an elementary mistake, but I digress...
You mean if I run a back test and it shows that I will make 200% per month that I will not make 200% per month going forward? Ahhh.... now you ruined my day and my right to make lots and lots of money... Seriously though, I completely agree with you a back test has to be taken with a grain of salt. A misused backtest is a great way to be called, "fresh meat" for the market.
First, nice job on the trading - congrats. Second, the only rule we really follow on the floor is not to follow a rule. Use your indicators as the signal for an "opportunity" but don't follow blindly. If you were to have purchased "cheap IV" relative to HV in early/mid 2009 you would have lost everything. VIX and vols in general have been trending to trends - just be aware of that tendency recently b/c you can go broke on the turn.
VIX - yikes... What to say that hasn't already been said. I have several issues with the VIX as a method of predictive statistical volatility. 1. While they have changed the model, it is certainly not reflective of the actual increase in skew. It is possible, to see an increase in the skew while the mean stays the same. The market would be indicating higher volatility, yet the VIX could very well be unchanged. 2. The model also assumes a normal distribution curve, but we all know that is not the case - hence part of the skew adjustments try to make up for this difference. My theory (as whacked as it may sound) is that volatility is in two states, like in physics - the matter vs. anti-matter. Volatility has a net total amount - which is partially reflected in implied prices - but the rest of is in an anti state. The velocity, acceleration, and jerk from the anti-state to the real state is based on how far volatility is over or under invested in. Fair value in this case is the perfect measurement between implied expectations vs. statistical outcome. It is WE the traders (investors) that over and under vest in volatility - through fear and greed (for lack of a better terms). The book (market maker) adjusts accordingly to order flow. Being able to measure the two states of volatility and the 3 derivatives of position - I would assume - give us a better gauge of over and under vest value vs. "fair" value. It is something I have been f'n around in our think tank - I welcome any thoughts...
I implemented the VIX and have to say the same thing as you. yikes... I think the biggest issue with the VIX or the VIX calculation is that it does mask the skew. When the VIX goes up or down, you don't exactly know why since the calls and puts are mushed together. Another interesting aspect on the VIX is that in options expiry week the VIX automatically jumps ahead a cycle. Thus in options expiry week you basically can't do any math on the current options since those vols are not used in the basis of the VIX calculation. I can understand why they are ignored, (due to jumps) but it is still something you need to be aware of. Mapping the statistical to implied is an interesting way to go since you are looking both forward and backwards. Christian
lionfish42 - that is true for the old VIX, however the new VIX does not assume normal distribution. Also VIX is volatility index - it is constructed specifically to measure volatility while removing the skew component. If you're looking for an indicator to measure skew - you can calculate skew index, or some skew-adjusted measure. If you're looking to bet on skew and have access to OTC you can trade gamma-swaps, corridor variance swaps, and others that are a blend of skew and volatility.
Sorry - I do stand corrected as to the old vs. new. As well, I am aware of the including of the skew in the new model. However, the distribution in not indicative of statistical distribution. Of course that has always been the fault of every theoretical model - it is only a guide and not suppose to be a predictive tool. I think we all wish it was - wink wink. I should clarify that I am not looking to trade skew (OTC or otherwise). I'll state at this time only that we are working on a model that I (hope) believe will have better resolution on volatility than current methods. Of course I could be chasing my tail, but what if I could tell you if Schrodinger's Cat was dead or alive?
That's what the cat said when he was put into the box! I guess you missed my point on the 3 derivatives of motion..... never-mind - You should see the group I work with - it hearkens back to Xerox Parc. I don't know if that puts the proper image to mind - without getting too explicit.