Frequently Asked Questions

Dr K VIX Volatility Model
How do you decide whether to introduce a new change into your model?

Since July, the S&P 500 has traded in the shallowest band in its entire 59-year trading history. It has traded in an even more flat manner than in 1976 when it typically corrected no more than 4-5%, an unprecedented period which lasted 7 1/2 months. I remember Bill O'Neil telling me that the 1976 market was one of the most grueling, difficult periods to trade. Well since July, 2016 tops 1976.

Fortunately, such periods *always* come to an end since the only constant is change when it comes to markets. Patience has never been more a virtue. Further, buying the ebb in stocks, ie, constructive weakness, and selling into the flow, ie contextual strength, in these markets has yielded profits as our members have seen during certain target rich periods each year including this year [2016].

It is also interesting to note that the ETF XIV which tracks the S&P 500 VIX short-term futures, can have an upward bias, so even though the S&P 500 has been trading flat since July 2016, XIV yielded a +17.2% profit on the July 15 sell signal which lasted nearly two months. This signal peaked near a +40% profit at which time I emailed members suggesting partial profits could be taken: http://www.virtueofselfishinvesting.com/reports/view/market-lab-report-fear-is-a-four-letter-word-vix-volatility-model-vvm-important-update-9-8-16 

Over this record flat period, the model's self-learning algorithms have, as always, been accounting for the change in the way markets trade so any necessary adjustments have been made accordingly. This includes a simple rule which allows the model to buy back a position if certain criteria are met to minimize the changes the model will get sidelined as it did after its fail-safe kicked in on 9-16-16 forcing it back into cash. With the new rule in place, the position would have been bought back on 9-16-16 then sold on 10-21-16 for an +18.5% profit when the model switched to a buy signal.

Since then (as of this writing on 10-26-16), the model has had two fail-safes kick in on 10-21-16 and 10-25-16 which is totally normal in terms of how the model tends to have many small losses along the way.

Incidentally, all adjustments must not just improve the profit/loss profile when accounting for any changes in market behavior, but must also prove out over the entire duration of the backtests. Naturally, most potential changes are never made since most changes may improve recent results but at the expense of prior results. The key is that any rare change introduced into any model must be backed by inherent logic, then proven out historically then under fire, in real-time trading.


Published: Oct 26 2016, Modified: Nov 20 2016