The real inspiration for this presently brief series of reflections on indicators, telling stats etc. was inspired of course by the seeming irrelevance of value at risk/variation based analysis and the like.
When judging a portfolio people will usually consider Value at Risk because its a simple and malleable metric. How much could I be likely to lose over a specific period of time. Are you 95% sure, rather be 99% sure, the language surrounding the variable is borrowed from statistics, and so is the calculation, so its a nice thing to tell would be customers, etc. People with large amounts to invest are undoubtedly happy to hear, "I can say with 99% confidence that you can't possibly lose more than XXXX.XX over the follower ____ months/days/years." Kind of seductive isn't it? However,this is misleading for several reasons, the first of which being quite simply the overlap in lexicon, occurring regarding statistical confidence, and qualitative(subjective) confidence per se, but either way this rhetorical discrepancy isn't really the source of the intrigue behind this musing.
Seeing as how, this market is like a wave, moving above and below certain points in price on an almost daily basis as the past few days have show, perhaps Value at Risk is actually skewed negatively, and if not then at least perhaps the volatility of returns which is often taken into account as a sort of risk, is skewed and somewhat misleading in a negative way. This of course being the case, because when looking at the probability of a security etc.'s price being below a baseline price, one is going to get a probability of around half, which is somewhat misleading given that said price, base price(mean for this scenario) relationship will give us a positive relationship(residual) the other half of the time.
For example, in recent days the stock market has been like a snake coiling around a raised cog, and extending down to a lower cog, and then extending up to a higher cog and then repeating this process.
Is this variability really a risk? Variability is usually considered a risk, because it represents a deviation from the norm, an "unknown". However, if we can in essence count on a little bit of variability around a sort of pivot point, on a daily basis, is it really an "unknown"?
The situation is perhaps reminiscent of Donald Rumsfeld's "Known knowns and Known unknowns", vs "unknown unknowns" discussion, which seems like a crock at first, but on reflection starts to make a little more sense. Is foreseen variability a hurdle an investor or investment manager has to mention or reflect on if it is very likely going to occur in a patterned kind of way, which is easy to understand. Perhaps the supposedly simplifying metrics are, as opposed to their creators' intents' obscuring our views of the markets, and portfolios in this specific investment climate, as opposed to illuminating them.
Considering this, one may thus want to try to figure ways to reconsider variance. Perhaps as opposed to using raw data per se, it would actually be better to consult or analyze brief period moving averages in regards to data points for considering variance or whatnot, because even though these sorts of metrics would undoubtedly lead to smoothing, and larger correlatory patterns emerging between individual securities etc, and the market in general, perhaps this method would lead to more telling analytical tools, that would actually reflect, more succinctly the underlying market conditions, and security price movements which these sorts of metrics are designed to analogize.