Monday, October 10, 2011

Mark Hulbert and the Use of the VIX as a Timing Model

Fitch downgraded Spain's debt by two notches on Friday, taking the rating to AA- from AA+. The outlook remained negative. Fitch downgraded Italy's one more notch to A+.  Portugal was kept by Fitch at BBB-, just above junk. We are nowhere near the end of the sovereign debt crisis. The austerity measures by Western governments will likely have further negative impacts on GDP growth and will consequently place additional pressure on the debt ratings of several developed countries.

Some people want to know whether a junk bond is safe. What can you say in response to a question like that one. The mere asking of the question reveals a lack of common sense. If you ask it, and are serious in wanting an answer, then you need to immediately cease from managing your own money. Obviously, no security that fluctuates in value is safe. Is this Bond Safe? To ask whether a junk rated bond is safe indicates a total lack of the most basic and rudimentary knowledge necessary to manage money. It is unfortunate that so many individuals make investment decisions based on so little.

Andrew Bary argues in his Barrons' column that the plunge in junk bond prices over the past several weeks have made them alluring to patient income investors. This assumes, of course, no Lehman like financial meltdown or another deep recession anytime soon. Of the bonds owned in my Junk Bond Ladder Strategy, a few thousand in the red now, one is mentioned specifically in that column, the Texas Industries 9.25% senior bond maturing in 2020. That bond is now trading mostly in the 75 to 79 range recently. FINRA I purchased just one bond in July 2011 at 97.5 which seemed reasonable at the time, but the bond has cratered since then with the rest of the junk bond market. Bought 1 Senior Texas Industries 9.25% Bond Maturing 8/15/2020 at 97.5 As long as Texas Industries survives to pay par value, I have only lost the opportunity to buy that one bond at a lower price.

The Bank of England announced a new quantitative easing program last week. The Bank will create £75 billion in money to buy bonds. The Bank of England's Governor said the world is facing "the most serious financial crisis we're seen, at least since the 1930s, if not ever". This remark started a debate among staff members here at HQ on whether hedges need to be purchased for a potential upward acceleration in volatility and a waterfall slide in stocks.

The Washington Post reports that Iraq is providing assistance to the Syrian dictator in cooperation with Iran.

This is going to be a long post. I view the topic as important.

1. Mark Hulbert and the Use of the VIX as a Timing Model for Stock Allocation: I exchanged a few emails with Mark Hulbert about the use of the VIX to time movements into and out of stocks. I sent him several links to my discussions, published over the past three years in this blog, including the one explaining my model in relatively simple terms. Vix Asset Allocation Model Explained Simply

Long time readers are already familiar with that model. In Mark's columns, he mentioned that a simple allocation model would be to buy stocks when the VIX was below 20 and move into cash when the VIX went over 20. MarketWatch 10/3/11 MarketWatch 9/6/2011 MarketWatch 8/1/2011 I did not notice his first two columns until I spotted the last one. Near the end of this section, I will discuss some of my comments about those two columns found at Marketwatch after I go into more detail about my own Model, first developed in a few minutes in 2007, after looking for the first time at the long term VIX chart and scrolling through the VIX price information. The pattern was both clear and obvious.  

Based on historical evidence, an investor using my VIX Asset Allocation Model could have successfully timed movements into and out of stocks over the past twenty one years. A hypothetical investor, desiring to make only one decision, stocks or bonds, would have been able to navigate the last 21 years successfully. That simple minded investor would have been in stocks during the Stable Vix Periods and in bonds during the Unstable Vix Periods. It is just that simple. 

Generally, the VIX will be negatively correlated with stocks as to directional moves. Most of the time, if stocks move up, the VIX will move down and vice versa. A paper found at the CBOE website says the inverse relationship occurs three out of four days historically. I am not concerned with daily movement, most of the time, but in the longer term trend movements. For the most part, I am not trying to use the VIX to predict short term moves in the market.

Humans do not handle spikes in volatility well. The natural reaction is to flee or to require a lower price to buy stocks in order to compensate for highly volatile movement. Humans need tranquility, calmness, and comfort. Those conditions are met when the VIX moves continuously below 20. Those conditions are not met when the VIX moves out of that stable pattern of relatively low volatility and starts whipsawing up and down at higher levels. 

A Stable VIX Pattern is a continuous movement in the VIX below 20 continuing for at least three months. That kind of movement breeds tranquility and comfort with stocks. Stocks have performed admirably during Stable Vix Patterns.  

When the VIX spikes toward 30 in a way incompatible with that Stable Vix Pattern, then that is a signal to reduce or to eliminate the stock allocation depending on the unique circumstances of each investor. I call that event a Trigger Event. Something in the real world has disrupted the comfort and injected fear into investors, particularly knowledgeable ones. It will most likely be a rational fear causing that spike out of the Stable VIX Pattern of continuous movement below 20.

The CBOE VIX data starts in 1990. Since that time, there have been only two Trigger Events that terminated Stable Vix Patterns.  

The Stable VIX Pattern starting in May 1991 was terminated by the October 1997 Trigger Event. That Trigger Event occurred in late October and November 1997, with the VIX spiking out of the stable range into the 30s and remaining at elevated levels for several days. VIX Historical Prices 

The 1991 to October 1997 Stable Vix Pattern was a long term one, where exposure to the S & P 500 worked extremely well.  In March 1991, when the VIX started to move continuously below 20, the S & P closed at 373.5 on 3/14/1991 and had risen to 984 by mid-October 1997VIX and S & P Compared 1990 to 1997   

Historically, the stock market will calm down after that first Trigger Event and will start to climb again. Under the model, that rally must be sold. The one decision hypothetical investor would then sell stocks and buy bonds. Given the duration of this cycles, a five year treasury note might be the safest option. The market did rally after October 1997 and the VIX fell to below 20 in February 1998.

For purposes of illustration, I will take the relevant data for February 27, 1998. On that day the VIX closed at 18.55. ^VIX Historical Prices Volatility had calmed down after the Trigger Event. The S & P 500 closed that day at 1049.34, ^GSPC, higher than its level during the October 1997 Trigger Event. The worst day during that event was 10/27/1997 when the S & P fell to 876.99, ^GSPC, and the VIX spiked to 31.12 from a 23.17 close the prior business day. The elevated readings in the VIX in late October and early November 1997 were the Trigger Event. The market gave the investor an opportunity to change their asset allocation by reducing or eliminating stocks a few weeks later. 

I also start the current long term bear market from that October 1997 Trigger Event. Please note that the market is close to the October 1997 S & P 500 level now. Dating the Start of the Current Long Term Secular Bear Market That is another topic but very important to keep in mind when using the Vix Asset Allocation Model. The Roller Coaster Ride of the Long Term Secular Bear Market The Big Picture Questions Underlying Cause of the Current Long Term Bear Market is Too Much Debt The Importance of Identifying the Underlying Causes of Long Term Bull and Bear Markets  Generally, I will be selling securities bought for smaller profits, in rapid fire fashion, when the stock market is in an Unstable VIX Pattern, within the context of a long term structural bear market, and has arguably entered a new cyclical bear market within those patterns.

The Unstable VIX Pattern is what happens after the Trigger Event. This pattern will be a dangerous and difficult market for the individual to navigate. There will generally be a lot of volatile up and down movement in stocks, with the end result being a long period of going nowhere. Going nowhere on a roller coaster is also the hallmark characteristic of a long term bear market, where the buy and hold investor would likely lose principal after adjusting for inflation The long term bear market can last for 15 or more years. The Vix Model is  a shorter term signal and its signals can be given in long term structural bull and bear markets.   

Phase 1 of the Unstable VIX Pattern will be market by repetitive movement in the VIX between 20 to 30, with temporary spurts to over 30 and below 20. A potentially more ominous pattern, Phase 2, would be a burst into the 40s that could lead to even further elevation into the 50s and beyond, which would be associated with catastrophic stock market losses similar to what happened after the Lehman failure in September 2008. There was a Phase 2 formation in late September 2008 that gave investors two days warning to get out. SEPTEMBER 2008: FORMATION OF THE DEADLY PHASE 2 OF THE UNSTABLE VIX PATTERN The operative word after a Phase 2 signal is caution. 

The Unstable VIX Pattern, Phases 1 and 2, can last for a very long time. The one decision investor could just sit it out in five year treasury notes, rolling them over if necessary, and save themselves a lot of heartburn. Historically, it would be difficult for a skilled trader to navigate the Unstable VIX Pattern to beat the return of a five year treasury note.   

The more adventuresome investor, such as myself, will play the volatility with the substantial cash raised after the First Trigger Event. Generally, this will be a mechanical trading strategy. During Phase 1, stocks are sold when the VIX moves below 20 and hedges are bought for the stocks that are kept. When the VIX spikes toward 30 or over, stocks are bought again and the hedges are sold. This will work until Phase 2 comes around which may never happen. It did happen in September 2008. Then the investor following this path, which included me, would likely have no hedges, having sold them when the VIX spiked to the low 30s or high 20s, and would have added some stock positions back. Then, an adjustment has to be made when it becomes apparent that a Phase 2 of the Unstable VIX Pattern has formed, with the recognition that danger lurks at ever corner.

If I bought good companies when their price had fallen to reasonable levels for a long term holder, I can become a long term holder, start reinvesting the dividends, and potentially buy more shares during this catastrophic phase of the Unstable Vix Pattern with the cash stash. I could then wait for the movement to below 20 in the VIX and then decide what to do with the position added just prior to the Phase 2, Unstable VIX Pattern formation coming out of a long standing Phase 1 pattern. Trading and Asset Allocation in Stable and Unstable VIX PatternMore on VIX AND ASSET ALLOCATION;  Stable and Unstable VIX Patterns Impacting Changes in Allocation to Stocks, Bonds and Cash (November 2008 Post). In the case of my PEP shares, I was able to sell those shares before a significant price drop and then buy them back after the price fell over $20. Volatility, Catastrophic Event Formation, Asset Allocation Decision for Pepsi September 2008 Similar, when the VIX recently spiked in a similar fashion, I sold my position in Heinz which had not fallen much and realized a large percentage long term capital gain. Sold HNZ at $50.10/Probable Formation Phase 2-Unstable Vix Pattern/Fear and Enhanced Volatility in Certain Classes of Income Securities      

When the catastrophic event formation occurred in September 2008, the Phase 1 Unstable Pattern had been in force since August 2007, and I could trade that pattern based on the movement in the VIX until Phase 2 emerged, and then there is no guidance from the model. The crap has hit the fan.

As shown in great detail in this blog, I carefully deployed cash flow to buy mostly income producing stocks that appeared to present good values during the September 2008 to March 2009 period, and this worked out just fine. It may not the next time. The Great Depression Part Two was avoided by the efforts of governments around the world. It remains to be seen whether it was merely postponed rather than avoided altogether.

During the Phase 1 of the Unstable VIX Pattern, there will be temporary movement in the VIX below 20. That movement will be associated with a stock market rally. As long as that pattern continues, the rally will not hold and the VIX will spike again. That spike will be associated with a market downturn. So, in that pattern, it would not be advisable to buy stocks when the VIX falls below 20, unless there was a very good reason for a particular selection. Instead, that rally needs to be sold since it will not last.

This is where I differ with Mark Hulbert's two posts. It would not be productive to buy when the VIX falls below 20 when the market is in an Unstable VIX Pattern. Nothing will be gained by a buy other than a quick sell, most likely at a loss. There are frequently weeks where the VIX moves below 20 during such patterns before popping back over 20. The rise above 20 would be associated with a market decline, likely causing the investor to sell at a loss.

I will give two examples of the VIX temporarily moving below 20 during an Unstable Vix Pattern to illustrate my point:

VIX August-September 2000 Buy or Sell Stocks?
The VIX did not continue moving below 20 for three months. It popped out of that bullish movement on 9/29/2000 and then started to accelerate up again. ^VIX Historical Prices  In my Model, the market was in an Unstable VIX Pattern since October 1997 and this movement below 20 would be a sell signal rather than a buy signal. In August 2000, the DJIA had moved from 10,606.95 on August 1, 2000 to 11,310.64 on 9/6. ^DJI Historical Prices By 4/3/2001, the DJIA closed at 9485.71. ^DJI Historical Prices  By 4/3/2001, the VIX closed at 34.72 and was trading in the high 20s and low 30s.

A more recent example was the movement below 20 in May 2008, a pattern characteristic of an ongoing Unstable VIX Pattern, Phase 1:



This was a shorter duration move under 20 than the previous example. Again, the VIX quickly spiked above 20 and into higher readings. Investors remember vividly was about to happen. Yet, the DJIA had rallied in May 2008 to close over 13,058 on May 2, 2008. ^DJI Historical Prices Clearly, in retrospect, my model gave the correct signal. The VIX had been in an Unstable VIX Pattern since August 2007 and this movement below 20 had to be sold, not bought.

I would add that the Model would have signaled the purchase of a hedge during the May 2008 time period.  Hedges can take many forms. One temporary hedge would be a double short stock ETF, at least for those without margin accounts or option accounts. Experienced option traders would probably want to protect their portfolio with options at that time. A margin account owner might want to short some stock ETFs.

The double short stock ETFs do give more bang for the buck but have serious tracking problems after one day. For me, that is my only option since I only have cash accounts and have never learned to trade options. A purchase of the double short ETF for the S & P 500 could have been made at $55.22 on May 2nd, soon after the VIX fell below 20. SDS Historical On 9/15/2008, the VIX spiked from 25.66 to 31.70, and someone strictly following the model would have sold that security at that time. SDS closed that day at $74.01SDS Historical Okay, that worked as a hedge. The reason for taking the hedge off is that a characteristic of this Phase 1 pattern is a whipsaw action that returns the VIX to below 20. If the hedge is held, it would likely lose value, rather than cushioning the impact of the market decline.

The problem is that the market soon went into a Phase 2 Unstable VIX Pattern. SDS paid a $4.01 dividend later in September and then hit $114.94 on 10/27/2008. I do not need to be reminded of what happened in October 2008.

Our HT did not follow the Model too well, having an incomprehensible desire to change the rules based on "modifications" made in contravention of the system. Still SDS did cushion the downside some, but nowhere near an optimal level even with the last sell before the Phase 2 pattern unfolded in late September 2008:

2008 SDS +$604.49

LB wants everyone to know that it was the RB at the helm during the pertinent period at issue.  RB noted that only the Nerd Machine "modifies" its stinking rules.

A possible revision to the hedging strategy would be to keep some hedges when the VIX starts to spike higher into the 30s, just to see what happens, and that decision can be made after an appraisal of what is happening in the real world. The VIX Model normally requires attention only to the VIX chart and the Model for allocation and hedging decisions. An analysis of what is causing a change in the VIX pattern is necessary to decide what to sell.

The VIX hit 15.95 on 7/7/2011, with the S & P 500 closed at 1,353.22 and the DJIA closed at 12,719.49. SDS was then at 19.58. During the VIX spike in August 2011, SDS closed at $27.44 on August 8th with a S & P 500 closing that day at 1,149.46.

{The last period of movement below 20 in the VIX started on 3/23/11, after a vix spike to the high 20s. While this is a judgment call, based upon what was happening then in the real world, I decided to start the count over based on The VIX's movement into the 20s (6/15; 6/16; 6/17; 6/24; 6/27; 7/14; 7/18; and then above 20 consistently starting on 7/26/11). This was a close call however, and was resolved toward restarting the count based on an opinion that real world events would soon take the VIX higher. Before that movement in the VIX below 20, a similar period lasted from 12/2/10 to 2/18/11, and was followed by several readings above 20s before a spike to 29.4 on March 16, 2011. That movement was not consistent with the continuation of an Unstable Vix Pattern, Phase 1 and influenced me also to restart the count based on the 20s movement in June 2011. The Model is not entirely mechanical in this respect and may require a modest decree of judgment when moving from an Unstable to a Stable Pattern.  All of the movement since August 2007 has been consistent with the Unstable Vix Pattern, either Phase 1 or Phase 2}

Again, under my Model, the rally that takes the VIX under 20 during the Unstable VIX Pattern period would need to be sold and hedges bought to protect at least part of what is kept. Eventually, the Unstable VIX Pattern will turn into a Stable VIX Pattern. The Model does not permit hedging of a stock portfolio during that Pattern, and all hedges would then have to be sold probably at a loss. The move then is back into stocks. The strategy becomes buy and hold with infrequent trading. No sales of stock ETFs for broad market indexes are allowed during the Stable Vix Pattern, and this would include such ETFs as IYY and SPY.

Mark Hulbert's idea of buying below 20 and moving into cash when the VIX is over 20 will work in my opinion better than a pure buy and hold strategy, at least when the investor moves into a high quality bond fund fund or a five year treasury note as the alternative.  Why? The investor would be kept in stocks for the entire Stable Vix Pattern, and that is extremely important. The Stable Vix Pattern period is where the investor is likely to make most of their money in stocks. The Unstable Vix Pattern is unproductive for the buy and hold investor but can be extremely profitable for the trader. The other reason that his simple model will work is that it gets the investor out of the market for the really nasty periods.  I have not ran the numbers, however. While this simple model would drastically reduce volatility, it would create far too many moves into and out of stocks.

I pointed out to Mark, and I believe this to be true, that I could have successfully navigated the stock market over the past twenty-one years with nothing but a Vix Chart and the Model. This is a summary of the one decision results, stocks or bonds:

Four Decisions May 1991 to Present
Stable Vix Pattern: May 1991 to October 1997-Stocks
Unstable Vix Pattern: October 1997 to January 2004-Bonds and/or Cash
Stable Vix Pattern: January 2004 to August 2007- Stocks
Unstable Vix Pattern August 2007 to the Present-Bonds and/or Cash

Long Term Bull Market 1950 to 1966
Long Term Bear Market 1966 to August 1982
Long Term Bull Market August 1982 to October 1997
Long Term Bear Market October 1997 to Present

In each case where a Trigger Event terminated a Stable Vix Pattern, the investor could have waited for the VIX to move below 20 before reducing or eliminating stocks. This would have allowed for an exit at better prices than prevailing during the Trigger Event, when prices would be falling. There will be a large group of investors who will view that fall as a buying opportunity, unwilling to recognize that conditions have changed in a fundamental way.

I did recently show how the Model would have worked for that one decision investor, using only the stock ETF IYY and the bond ETF IEF, starting from March 2004. Item # 3 More Discussion on Asset Allocation in Unstable Vix Patterns (October 6, 2011 Post)

This model is based on historical evidence. I use it as a guide. I did reduce my stock allocation and raised cash in 2007. Unfortunately, I did not sell enough stocks. While I was still hit hard after September 2008, the cash allocation and the movement into short term investment grade bonds in 2007 did allow me to recover quickly from that latest stock market disaster.

The future may be different, possibly radically different from the past. All that I can say is that a successful allocation strategy could have been made since 1990 based solely on the movement in the VIX, with no other information, compared to a buy and hold investor. The Stable VIX Pattern will capture most of the positive stock returns. The Unstable Vix Pattern can be rewarding to the trader but not to the buy and hold investor.

There are still tweaks to be made. An example is what does the Stock Jock do during Phase 2 of the Unstable Vix Pattern. Those periods are treacherous, to say the least, but also present some of the best opportunities to create wealth. They will be extremely difficult periods to navigate successfully. Some investors might just want to take a pass on the entire up and down roller coaster, and the Model allows for that option also. The reaction to the signals given by the Model needs to be tailored to each individual's unique circumstances, including risk tolerance and situational risks.

The foregoing took up all of my allocated time for writing. I will write in a stream of consciousness and will generally only do minor editing when I am finished. Due to time constraints, I will discuss my remaining trades from last week in the next post.

No comments: