My thinking was to simplify, and take a simple approach with minimal effort, that if results were positive, THEN and only Then, perhaps warrant deeper efforts/backtesting. IE: There are some IV values that are trivial to obtain (VXST, VIX, VIX3M, VXMT for SPX) (9, 30, 90, 180 Day IV) all derived similar to VIX AKA CBOE VIX White Paper. If these are acceptable for eval, then merely code exactly what you seek and view results with a TOS ThinkScript, without the additional effort of some backtesting) I am considering this, but just spotted your post and have not had time to look deeper (but seem fairly easy to identify every exception and quantify however you desire); Since you are only looking on the Call side, this seems to be hardly ever challenged, but should be able to nail it down without too much work.
I wouldn't hold to much value to VIX, as the VIX is an average of IV's across several (most) strikes. So it also incorporates skew... What would you trade if you find the 1sd up is undervalued? Butterfly?
The critical thing to remember about IV is that it represents the cost of manufacturing an option via dynamic delta-hedging. It does not represent the expected move of the underlying. So if you bought a call and delta-hedged it, the p&l loss on one side is expected to be offset by the p&l gain on the other side. That is all. The market maker does not care if the stock moves up more than apparently implied by IV of a call, only that the cost of delta-hedging that move is less than the price he/she sold the option for, on average.
your 2017-price-iv-data selection is also very biased, no? though similar to 2013-14 price data, the iv behavior is quite unique why are you focusing on POT? are you playing binary options? if so, your broker usually has a minimum of 3-5% edge over you btw, POT is just a number.. certainly not a fixed one.. in some cases, that POT can decrease even when the UL price increase.. how is that happened? how much is the reward, say if the strike touched after 1/3 or 2/3 time spent? what you should care about is the average expectancy.. probability and r/r you need to map this out objectively.. not only when you are winning.. :]
IMHO: Observations of SPX IV surface over time do not support the notion that skew is a common error component of VIX. Sharp VIX events do have this impact you reference, however, the occurrence is seldom excessive (2018 Feb 5th & 6th during portions of the day did show significant Skew influence). Stated a bit differently, for "government work", using VIX instead of "skew adjusted" volatility for a 30 Day SPX IV seems adequate for most tasks, especially for the one posted here. Once you discount time effects and underlying price movements, the Vol surface is amazingly smooth and uniform across all strikes and time.
@OP and to his original question: "[What is the] Accuracy of IV for Expected Move?". Short answer: quite accurate. I've got the historical data to prove it (paying some $1000 / year for it) and am still struggling to build a system to exploit the mis-accuracy, of which the hardest part is figuring out when IV is accurate and when it is not Working on this since 12 years already. Think like going from kindergarten-level of options knowledge to high school graduate level. It's not university (college) yet, which probably explains why I haven't made money yet. Yet another yet, as retarded as I myself realize that I am compared to the vastness of knowledge required to advance this stuff (beat the market), I also realize that I'm much closer to the end result than a guy who's already starting, no matter how shiny his credentials. And I started it as an university graduate with a significant competence in math and 5 years experience in programming already (as a side note, I passed the math-physics-heavy exam as the first admitted at the respective university out of =~ 300 candidates competing for the hardest section). It's not Ivy league or Red Brick but trust me, there is [intelligent] life outside those oases too.
Yeah. The only good thing about it is that it'll require much less (an order or size less, where "order" starts at 2 and you divide by it) before I'm getting to the end result of this 2n'd part of my endeavor, which is making money. Then perhaps if I'm lucky to live enough after wasting my talent, youth and health on #2, maybe I can live enough to see some of #3.
There is a lot to be said about "talent" and it's correlation to "success". The strongest predictor to success is previous success, not talent, knowledge, physical fit or what else. If your parents were wealthy, you've got some 95% chances of keeping the fortune. If they were not, like my case coming from small village gentry, then albeit we used to have some 50% of the population living in the countryside, less than 2% of their children make it past university level. And I can attest from all the friends, work colleagues, acquaintances that I know: I'm the only peasant here. With my other-than-inheritance-money capabilities, I'd probably have had 1000% more chances having been born a black woman in Brooklyn than a white male in an Eastern Europe village. (Note that I'm not even mentioning the 3'rd ingredient of success of which I have aplenty which is "smart/wise" or "beautiful/handsome", interchangeably. Well, to get a picture of it, only if you consider a zero cut hairstyle Putin having a little of both of those things ) Things are changing though. I've had one of my urban-born friends, otherwise relatively equally capable and skilled, all of them much wealthier than me who didn't have money to shit in the city center when I arrived here (going into a bar for this exact purpose takes a lot of money or some acquired skill to do it for free), tell me what I already knew. That he was much-aided by the lack of serious competition. With businesses, women, politics. He said: "it was soo much better when I had money (actually he said "everything") and you had nothing!".