Dr K VIX Volatility Timing Model (VVM)
A great deal of work has been put into VVM to debug and improve its reliability. This work showed a back-tested result of +177.03% at its peak in 2016. In addition, profit/loss was improved for each year going back to 2009.
- You will receive an email whenever the model switches signals.
- Choose from the list of suggested ETFs. Buy the ETF.
- Await the next change in signal.
Overview of VVM:
The VIX Volatility Model (VVM) seeks to turn human emotions into profits by capitalizing on emotion-driven events. These events carry predictive value. Price/volume charts are used as a guide, after all, charts are human emotions on parade.
The VIX Volatility Model uses price/volume action of leading stocks, a dynamic list that is updated as needed, and major averages to capitalize on changes in volatility in the market. It uses regression to the mean as well as trending strategies depending on market conditions. That said, it does not require an uptrending or downtrending market to be profitable.
- Not A Black Box
Developing a seasoned, contextual "chart eye" over my 25+ year career has been essential. It allows me, and thus the VVM, to remain fluid with changing market conditions. Indeed, there is no such thing as a static "black box" strategy. Any system of value should in principle be self-learning thus self-evolving. Such a system learns from new data each day. This is the way I have always operated my strategies since the 1990s. It was an early, manual incarnation of Google's cutting edge "deep learning/machine learning" approach.
- Excels in Up, Down, and Sideways Markets
The VVM incorporates a number of strategies which can profit from a broad spectrum of market behaviors. For example, the VVM excels in trendless or downtrending markets where volatility is amplified. It also knows how to ride uptrending markets which tend to be less volatile. Its returns have well outperformed the major averages in backtests and in real-time trading.
That said, I would like each member to read this FAQ on potential weaknesses of the model.
The model's focus is to analyze in real-time whether there is sufficient buying or selling pressure to warrant a change in signal. Live signals shown in the table start on 6/27/16. Backtested results in the table are shown from 12/23/15 to 6/24/16.
The model has gone through a series of evolutions or "growing pains" similar to software that updates, making improvements and removing old bugs. Indeed, the challenging markets that have become ever-more challenging with each passing year have been a gift. The model is that much stronger for it.
All signals now incorporate the following:
- The UVXY profit taking rules activate should profits exceed a certain threshold. This acts as a trigger to let profits run but stops are kept very tight. The threshold is contextual to market conditions but seeks at least a 12-18% profit. The numerous boldfaced entries in the results table show when the profit taking rules were activated.
- No override or "gaming" of the model's signals.
- No override or "gaming" of the fail-safes.
- Integration of the VIX Spiking strategy discussed here.
NO DOWN YEARS in backtests and numerous triple digit percentage (>100%) years:
I believe the risk/reward of this strategy is by far the best I've achieved so far in my 25+ year trading career. But that is no reason to oversize your positions. Make sure you understand the risks involved in trading this model before risking any capital.
Not shown: Backtests using VXX start 2/2/09. Prior to 2/2/09, backtests start 3/10/99 using QQQ.
When the model issues a sell signal, it is selling volatility thus anticipating rising markets. When the model issues a buy signal, it is buying volatility thus anticipating falling markets.
You will receive an email whenever the model switches signals. Choose from the list of suggested ETFs. Buy the ETF. Await the next change in signal.
All results starting on 11-8-16 are in real-time. Prior to the 11-8-16 buy signal, results shown do not reflect debugging which improved the algorithm's profit/loss. Drawdowns would thus be contained to -27.9% in 2016 and net profits for 2016 would still be +61.8%.
Some members chose to take profits as we suggested taking profits on 9-9-16 just after the market opened as shown here in this email sent to members.
Thus for those who took profits at that time, a gain of nearly 40% would have been realized at that time on that one 7-15-16 sell signal alone!
Further, the debugging of VVM helped reduce drawdowns not just after 9-9-16 but also in the entire run which dates back to early 2009.
The boldfaced entries in the table show that the profit taking strategies have made a material difference to profits. Also notice that the fail-safes built into the model can cause strings of small losses, multiple times a day in some cases. This serves to keep risk under control. One must lose to win. Profits have always well outweighed smaller losses in all backtests in any given year.
Note that July - October 2016 was the toughest period to make progress both in stocks and in market timing for both models. The S&P 500 traded in the shallowest band in its entire 59-year trading history. Fortunately, such rare periods always come to an end. Indeed, the Trump victory seems to have dislodged the markets out of their almost non-tradeable doldrums.
UVXY is used for buy signals, XIV is used for sell signals. UVXY moves up or down twice that of XIV. If XIV is up +5%, UVXY would be down roughly -10%. More conservative investors may opt to use VXX or VIXY.
|Date||Signal||% gain / loss||$10,000 becomes|
|There are no earlier results|