Update: 3x2 System / Crash System

The variant of the StockPickr 'Crash System' that I follow has just moved up to a new equity high.


The variant of the StockPickr '3x2 System' that I follow has noticeably broken out to new equity highs since the last update in July.

Trend: Near Perfect Straightness

The trend continues to be near perfectly straight up. The time the 15 day trendiness has remained at an extreme level above 0.9 (gray line) is now longer than during the market collapse from February to March 2009. See the plot above. We cannot be too far away from some sort of counter trend move?

Trendiness Extreme

The 15 day trendiness has reached an extreme level. (See the plot above.) It is now over 0.9 (horizontal gray line). There will likely be some interruption of the straight line move up in the near future. In late February/early March 2009, the trendiness stayed elevated as the market collapsed. Are we in an equivalent panic period or will we quickly see a counter trend move?

I remember reading something that made a reference to the three week time period as being significant in terms of the human mind accepting a change. I believe that it was related to plastic surgery. Anyone know of the paper referenced, or the three week period as being significant in human psychology?

QQQQs Up 9 Days in a Row

The QQQQs have been up nine days in a row. What happens after the QQQQs have been up eight or more days in a row? The chart above summarizes the results for purchasing at the close and holding for one, two, five, and ten days after purchase. The results are good for a two, five, and ten day holding period. The results for a single day hold are not good. We may be looking at short term weakness followed by a further continuation of the rally.

Update: 3x2 System / Crash System

The '3x2 System' is still holding near its high water mark of the last two years.


The 'Crash System' equity curve has also moved up nicely and is closing in on its highs.

A diversified portfolio of active trading methods certainly looks to be the way to go. However, the massive drawdown of these two methods during the crash of late 2008 suggests that some sort of disaster filter is needed.

The New Normal?

The statistics for the broad market indices (daily data) have become stationary for the past year. The plot of SPY above shows the stationarity for a rolling one year, two year, and three year period. When the colored lines are below the critical value (horizontal gray line), the statistics for the corresponding period are stationary. The statistics for the past two years and three years are not stationary. The past year has become the new normal. Will the coming year exhibit the same tendency for monster moves as the last year?


The more volatile emerging markets (EEM) also have stationary statistics for the past year. The statistics for the last two years and three years are not stationary, but are closer to the threshold than SPY. Will emerging markets continue to lead the way?


The long term treasury bonds (TLT) statistics are not stationary for the past one, two, or three year periods. TLT was even more destabilized by the October 2008 market crash than the stock indices. Can there be an economic recovery with instability in the bond markets?

Shannon Rebalancing Revisited

What if we were to apply the Shannon method to the Permanent Portfolio discussed last week? The Permanent Portfolio used for testing was equal parts of VTI, TLT, SHY, and GLD. The plot above shows the results for the last three years for buy and hold and for Shannon style rebalancing. The rebalancing has helped a little bit since March 2009.


If we add a short SP-500 position (SDS) to the asset mix, the results are of course better as the market has had a rough couple of years. But, notice how much better the Shannon style rebalancing has done compared to buy and hold. This plot above shows the results for the last three years for a portfolio consisting of equal parts TLT, SHY, GLD, and SDS and a double allocation to VTI. Perhaps active rebalancing between an index and a short position in another index is what is need in uncertain times?


The correlations of the portfolio components are listed in the table above. Are there any asset classes that have a negative correlation to the stock market other than bond funds and short funds?

Update: Asset Class Rotation

In a post back in August 2008, I examined the results for a simple relative strength asset class rotation scheme. The approach buys the top N of M asset classes each month.


The plot above shows the results for the last 4 1/2 years for buying the top 5 of 8 asset classes each month. The return was better than SPY. The asset classes in this case were actual ETFs (AGG, EEM, EFA, IWM, IYR, SPY, and TIP) and one commodity index (DJAIG).


The results were significantly better by just buying the top 3 of 8 asset classes. The plot above shows the results over the past 4 1/2 years using the same set of asset classes.

While there was a big drawdown during the market crash, an active approach to asset class allocation would seem to be a good approach to helping to smooth out returns during turbulent market periods.

Update: 3x2 System and Crash System

The '3x2 System' has continued making a nice recovery. The equity curve is at the highs of mid-2008.


The 'Crash System' has been moving up and is closing on the highs.

The two systems have complemented each other relatively well over the past few months. A diversified portfolio of active trading methods may be the best approach for the uncertain future.

Dividend Aristocrats -- Not Good

The low beta/low R-squared stocks like the 'Dividend Aristocrats' have not showed much advantage over the market in the last two years. See the plot above. They have shown a small advantage over the market (SPY) since the crash during the Fall of 2008. The devastation of the financial sector has caused many of the Aristocrats to lose their place on the list.

Time Trading

Trading just based on seasonal tendencies is an yet another trading method to examine. The plot above shows the equity curve for a system that is always in the market either long or short. The trades are entered and exited based only on the date. It works fairly well. Over the last 10 years it has done much better than buy and hold. However, the method did poorly in 2006 and 2007 as the market surged higher. But, it caught the big market crash of late 2008.

Update: Market Neutral

QQQQ continues to outperform SPY. Market neutral positions can have advantages during volatile uncertain market periods. The QQQQ/SPY spread has improved since the last update.


SLV has pulled back to the -2 standard deviations point. Can it recover and again outperform GLD?

Update: 3x2 System and Crash System

The '3x2 System' has been making a nice recovery in the last two months. The equity curve is within striking distance of the mid-2008 highs.


The 'Crash System' moved down a bit in the last two months after a spectacular recovery since late 2008.

The two systems have complemented each other relatively well over the past few months. A diversified portfolio of active trading methods may be the best approach for the uncertain future.

SPY/VIX Correlation

The rolling correlation of SPY and the VIX has been become much less negative over the past two months. The plots above show the 15 day rolling correlation of SPY and the VIX index. The spikes to less negative levels tend to occur prior to the market going sideways to down. The longer term average rolling correlation is -80%. The present correlation of -70% is at an extreme considering the last two years, but is less so when looking at the last eight years.

Update: Shannon Method / Rebalancing

The Shannon Method of rapid rebalancing has some potential to capture some of the random fluctuations in stock prices. I have pursued combining a small number of relatively uncorrelated stocks in a portfolio and rebalancing them as they move 3% from the prior rebalance point. See my August 2008 post for a more detailed explanation of the Shannon method.

The plots above show the results for a portfolio starting with four stocks (GG, MSFT, O, VLO) weighted 20% each and a 20% cash position. (See prior post.) The rebalancing has outperformed since December 2008. The rebalancing also exhibited some advantage during the 2008 crash.

T2108 Over 80%

The plot above shows when the TeleChart T2108 indicator has been above 80% (red line). Such high levels have tended to mark times when the market was going to move sideways to down. There have been exceptions, such as 2006 where the market was in a runaway move up. During the last couple of years, very high values of the indicator have marked the beginning of a down moves.

Asset Class Rotation With Filter

The simple asset rotation scheme previously described, has taken a beating during the recent market crash. The asset classes used were US Stocks, foreign stocks, US REITS, US Bonds, and commodities. The plot above shows that the monthly rotation scheme outperformed SPY, but was not any better during the high speed crash late in 2008.


The performance of the basic top 3 of 5 asset classes scheme is substantially improved with the addition of a filter to avoid taking positions if the asset class is below its 200 day moving average. Take a look at the plot immediately above. Although the filtered rotation approach will lag if the market continues straight up to its 200 day moving average, the avoidance of large losses is also an important consideration.

What Happens After SPY is Up Big

The image above shows what has happened one, two, and five days after SPY has closed up 5% or more in a single day. The results are not statistically significant. The results vary wildly with the most recent instance (3/10/2009) working out very well.

Update: Market Neutral

After a recent pullback, QQQQ may still be able to outperform SPY. A market neutral position may still be worth while.


SLV spiked down and recovered this past week. It may continue to outperform GLD.

Market Chaos

The market remains in a relatively unpredictable state. Various measures show the market has the potential for chaotic action.


The statistics for SPY for the past one, two, and three years are not stationary. The plot above shows the rolling stationarity for the three time periods. All periods are well above the threshold of 19 (gray horizontal line). Any of the patterns you see in the SPY over the periods indicated may merely be coincidental.


The SPY average true range (ATR) has been increasing over the past few weeks. See the plot above of the three week ATR for SPY. The ATR is far above the ranges seen in the years prior to the recent crash.