Saturday, November 29, 2008

Trading and Asset Allocation in Stable and Unstable VIX Pattern

As I have said, my work on using the VIX as an integral part of an asset allocation strategy is a work in progress. This discussion assumes familiarity with at least the major points made in all of these earlier posts:


Since we are currently in an unstable VIX pattern, I have been focusing my attention on what to do during the bear market. To summarize some of my earlier points, the model required a reduction in stocks in October and December 2007. Until the VIX returns to below 20 and stays below 20 for 3 months, the cash raised in 2007 must remain in cash. I actually started to raise cash before October in addition to the forced reductions required by the model.  

My first attempt to develop a trade during the bear market focused on a swing trade using the double long and short ETFs for the S & P 500, SSO and SDS respectively. It may be questionable to use the SSO, a double long S & P ETF during a bear market, but it does work most of the time in a swing trade during an unstable VIX market pattern. I have not mustered the nerve yet to buy it during a bear market.

Under the model, the SSO would be added during a peak period of market volatility during the bear, meaning a spike to 35 to 40 and then taken off when the VIX returned to 20 or below. The other part of the swing trade was to buy SDS when SSO was sold and vice versa. This strategy would have worked most of the time until this October, when the VIX went to the 35 to 40 level and then spiked in October and November to 60 to 80, which would have caused a failure of the SSO trade and a premature exit from SDS. This is why I have modified this part of the swing trade to do a mental stop loss on the SSO part of the trade when the VIX spikes above 40.

The SSO is a better trade during the stable VIX pattern, which I define to mean 3 consecutive months below 20, and keeping the trade until the start of the unstable VIX pattern, a whipsaw up to over 30 and back down to below 20. In prior Transition Phases from a bull to a bear market, the trigger for disposing of SSO was the first spike out of the stable bull pattern to over 30 and then a sell of SSO when the market returned to below 20. 

A venturesome investor may wait for the return of the VIX to below 20 from a spike to over 30 before unloading the SSO trade while a more conservative investor could unload it at the first sign of the VIX falling out of the stable pattern by moving at least several days to the mid 20s. Another modification that I have made is to keep some SDS in my portfolio as a hedge throughout the unstable VIX pattern which I have not done during this bear market, having sold my entire position during spikes upward in the VIX, and then buying it back when the VIX fell back.  

The SDS, TWM, QID, and SKF trades this year have helped me a good deal. 

But I realize that further refinement is needed and I am still working on it. This is only for those very experienced and comfortable with the inverse ETFs. I found that I had comfort with all of them except for SKF which just spooked me.  

I have allowed for some new investments during an unstable VIX period by investing cash flow and the proceeds from any sale.  

I did not have any VIX data to back up continued investing during a bear market.  By bear market, I am not using the traditional way of measuring it, a decline of 20% in the S & P 500.  I am referring instead to an unstable VIX pattern (sometimes instability will be a bear market as now and in 2001-2002 and sometimes it may be a Transition Phase from the bull to a bear or something very close to a 20% fall in the S & P 500)  

An unstable pattern starts with a clear break in a stable pattern of below 20 with a spike into the the high 20s and low 30s for the VIX. Technically even with this spike, the bull market may continue. 

Talking heads may be saying that it was a correction with the market recovering and the bull resuming its charge to 34,000 or whatever. The VIX will then fall into a Transition phase, marked by a whipsaw pattern up to 30 and down to 20 or below, then back up and then back down. 

In prior transition phases, there could be two of these patterns for the S & P 500 VIX formed with  the bull recovering from the blows which caused the spike in the VIX. But for my asset allocation model this represents the time to reduce allocations to stocks. Until the unstable VIX is resolved by the formation of a stable pattern, I am in my bear market mode, let other people put labels on what they think has happened or happening, calling it a bull, bear, a correction or whatever. 

The 3 month period was established to keep me out with the cash raised by the forced reductions until stability is more than a passing fancy which does happen in bear markets. That is why the SSO trade can sometimes work in an unstable VIX pattern (and admittedly would be the most problematic of the two trades in an unstable VIX pattern). 

It would have worked in the prior bear market when the VIX spiked to 30 in October 2001 when the S & P was around 1060 and would have been taken off in February 2002 when the S & P hit 1106 with the VIX back down to 17.4. Then SDS would be put on and kept until July 2002 when the VIX spiked to 32 and the S & P 500 average fell to 815 which would cause the sell of SDS and a buy of SSO. The next trade would be more than a year later when the VIX fell to 16 in October 2003 with the S & P at 1150. We are now moving into a bull market but I do not know it yet.  If the swing trade was kept, then I would have sold SSO and bought SDS again. Everything is working fine up to  this point. By February 2004, the VIX has been in a stable pattern for 3 months and the model would then require the sale of SDS, probably for a small loss when the S & P was still in the 1150 area and the VIX had fallen to around 15 and a long term buy of SSO kept on until the unstable VIX pattern returned. 

For me, I would be allocating possibly $5,000 to SSO at that point and another $5,000 when it enters a stable bull phase 2 pattern which is not major for me. I would sell out completely, no matter what I thought about it, with a move to the mid 20s lasting more than a few days. 


That trade has worked very well using historical data and hypothetical trades after I developed the trading model based on VIX in 2007.  

There is certainly no guarantee that it will work as well next time but I limit my potential damage from a failed traded by pre-setting my exit point. In other words, I know exactly how I will exit the trade before I make it and will not permit a variation in that decision.  In short, SSO is far more likely to work during a stable bull market pattern compared to its use in the swing trade in the unstable pattern. As I said earlier, it would have failed miserably during the major spike in the VIX this fall which caused me to modify the trade for the future with a stop loss built into it. 

I have seen some writing about VIX and asset allocation but nothing like I have developed for my own use. I am not sure what some people are looking at when they say something about asset allocation and the VIX. To follow what I am saying, I would pull up a long term chart of VIX at yahoo which goes back to 1990 and then another chart of the S & P 500, using the interactive features. Then scroll over the VIX chart during a sustained period of below 20 and then check the S & P chart to see what was happening. Then follow my descriptions of the Transition Phases and the elevated unstable VIX patterns and check the charts. It is clear to me that- so far at least- the VIX can be used as a basis for asset allocations models.  

I read today a report prepared by Bill Swerbenski from Charles Schwab. VIX Up, Market Down?  He claims that the VIX is not a predictor of market moves and up/down movement in the VIX may be more relevant for shorter term trading strategies. I must be looking at different charts than him. When I first looked at the long term VIX chart last year, using monthly closing figures first, it was obvious to me at least-and I am not a CFA or a technician but a humble individual investor-that there was a clear model that jumped off the page for asset allocation lasting for several years before requiring a change. 

Mr. Swerbenski does however provide some useful information for me. He has a chart that shows the market returns one year after major spikes in the VIX. This chart justifies to me at least my current strategy of continuing to invest my cash flow. So now I have some hard data to justify what I am currently doing. 


Test of Vix Model for the Period 1986 to 1988 using CBOE Data for  ^VXO:

Prior to the VIX chart, the CBOE had used the S & P 100 and that chart goes back to 1986 (^VXO).^VXO: Summary for CBOE OEX VOLATILITY INDEX - Yahoo! Finance  I would just briefly describe some of that data. Starting on 1/2/1986^VXO: Historical Prices for CBOE OEX VOLATILITY INDEX - Yahoo! Finance, the VXO was at 18.07 and meandered in the 17 to 21 range until late February when it started to creep up to 22. March and April were spent mostly in a higher range of 22 to 26 which would cause me concern but would not cause a shift in asset allocation. It would have triggered an SSO sale if owned by me but not for the wild and slightly crazed ones (and of course it was not then in existence being a new product). 

By early May 1986, the VXO fell below 20 and stayed there until July when it peeked into the low 20 area before returning to a below 20 range and staying there with a few small exceptions until mid September when it moved back into the mid 20 range. But in October 1986, the VXO moved back into a predominant below 20 range until January 1987 when it returned to mid to low 20s. I have not seen anything yet to cause a trigger on asset allocation but this movement would deserve daily monitoring for a possible change. A trigger does occur in April 1987 with the first spike above 30. If I only had VXO now I could use it with some adjustments. But this close above 30 would start the time running for a forced reduction in stocks.  The forced reduction would occur at any time between July 7, 1987 when the VXO fell to 19.85 until August 10, 1987. 

We all know what happened in October 1987.  I am looking at the daily closing prices in October for the first time for VXO and it shows the stress, using rounded closes at  150, 140, 73, 102, 99, 113, 97, 81 etc.  It was not until November 1988 that the VXO fell to below 20 again. The S & P hit 224 on October 19th, a fall from 307.4 on the first day in the forced reduction period.  This looks like a maximum 1 warning before the storm hit. Unlike subsequent periods where there could be 2 warnings before the crap hit, you would have been smacked after 1 using VXO as the model in 1987. I would not use this index to make allocation decisions now due to the small number of stocks in it and probably would not even consult it. It does give me some data four years before the CBOE VIX chart starts to test my thesis and theories which is the only reason for bringing it up. 

So, that is where I am now. I do not own SSO or SDS. I am not likely to do the SDS part of the swing trade more than one year into the start of the bear market, given the usual length of these events. In the future, I will start SSO after three months of a stable pattern or possibly a few days earlier provided my comfort level is extremely high about the formation of an investable bull market. I will add SDS and keep it throughout the next unstable period as insurance and as a hedge rather than as an investment. I will continue to add selectively to my stock positions during the current unstable VIX period with cash flow or the proceeds from sales until the S & P 500 closes below 815, restarting that program if stopped by a close below 815 as described above. I will allow some short bond positions to mature in May and September 2009 without re-allocating back into short term bonds, reserving the funds for stocks including at some point the SSO buy. I will continue to modify the model as circumstances warrant. Lastly, I do not make large allocation decisions based on any model or an opinion that may pops into my head like every five seconds. At most, I would use my models to reduce stocks by 30% because I never know going into the bear how bad it will be. Reinvesting dividends in the bear market ultimately will prove to be fruitful assuming no 10 year type of great depression.   If the Dow returns to 14000 in a few years, reinvesting dividends at the current prices will look good later on, if no so hot now. 

Judging from the experience with VXO in 1987, a major spike in volatility as experienced in October 1987 may take a year or more to resolve itself into into an investable bull market.  We entered a similar extreme volatility phase in October 2008 so I am not anticipating now a quick return to a stable VIX.  As some are fond of saying, not me of course, this time may be different. 

No comments:

Post a Comment