Summary
- Volatility indicators are showing that we are early, but that the bulls are more powerful over the past week.
- The TSP / GGT Adaptive Timer, which uses the raw price signals from www.tsp.gov in the calculations, has all holdings LONG.
- The historical GGT timer has the VXF (proxy for the S-Fund) LONG, but the EFA (I-Fund) and SPY (C-Fund) are still in CASH.
- For Monday, March 28th, the GGT TSP portfolio will be transitioning LONG. This is an aggressive move, as not all timer methods are indicating that this is a good choice.
============
Volatility
Perceived volatility in the market, as measured by the ratio of the longer-termed anticipated volatility VXZ compared to the shorter-termed 30-day anticipated volatility VXX, has improved. Here's the chart:
As with all my charts, right-click on it to open in a new window or tab.
In the chart above, which is linked here, we have the ratio of longer-termed volatility to that of short-termed volatility. Increasing values of this ratio are considered shorter-termed bullish, and this is what the past week has delivered.
As you can see in the chart above, the ratio of VXZ:VXX has closed above the 7w moving average for the first time since the 2nd week of February, giving us reason to believe that we may be moving upward from here as far as the broad markets are concerned. Because I have no knowledge of the right side of the graph for the next couple of weeks, we must take this preliminary signal as bullish for the market in general.
I note with some caution that this 7-week line will be tested going forward. Observe the May-June 2009 sequence, where the 7-week line was penetrated or tested for nearly 8 weeks. During this time you may recall that it was difficult to make money in the TSP funds. Note though that the slope of the 7w line never went below 0, so we had confirming indications to remain long during this test/penetration of the 7w line.
Nearly 12 months later, in the June 2010 period, we closed consecutive weeks below the 7-week line, then above it for two weeks, then we straddled it for the next two weeks, shaking out the weak bulls. Note that we were negative in terms of slope, but then, it moved back into positive territory in mid-June, giving us some confidence to stay long. It took until after mid-August before we had consecutive weeks strongly closing above the 7-week line, which we all know in hindsight as bullish. The slope of the 7w line was clearly positive, giving us confidence through the tough times of June/July/August.
The situation now is a bit different. The slope of the 7w line is very negative, which means that short-term volatility has been large compared to anticipated longer-termed volatility, but it is pointing higher, meaning that short-term volatility is decreasing. We are early in terms of a call to the long side, and until this line crosses into positive territory, this present signal is weak. Hence, while the signal is long, we need to observe other signals to see if they can confirm a move to the long side.
===================
VIX Ribbon
The VIX Ribbon is an indicator that is available in TradeStation, and it basically is a set of different colored moving averages, all plotted on the VIX. The VIX is the index on the anticipated short-term volitility of the S&P500, and the VXX noted above is an ETF that mirrors the VIX performance in daily change.
Here's the chart:
The rules for the VIX ribbon are easy: when the VIX closes above the largest EMA, we have a sell signal, and when it closes below the lowest EMA, we have a buy signal.
As you can see in the graph above, I've plotted vertical lines on days that issued up or down signals.
The VIX ribbon is prone to whipsaws, so it cannot be used in isolation. What is notable about this indicator is that we issued a BUY/LONG call with the close of markets on Wednesday, March 23rd, essentially confirming the VXZ:VXX indicator above. This situation points us to a move LONG in the markets.
I note with a careful eye that the pattern established towards the 6/15/10 long call looks a lot like the sequence that we have at the present time. We must be vigilant for a reversal of volatility upwards, much like what occurred between 6/15/10 - 6/24/10.
======================
TSP / GGT Adaptive Timer
I am using a modified worksheet that I received from Dan Chiswick in Florida to individually watch the different TSP funds in terms of timing. This methodology uses adaptive moving averages, e.g. the length of the moving average "adapts" as the market conditions change. The impact is that as volatility increases the timer increases the likelihood of moving to CASH.
This timer is either a really good timer, or a really bad timer, depending on your mental state. For an example of where the timer worked well, the first sign of short-term weakness occurred with the C-Fund, which measures the S&P500, and this occurred with the close on 3/10/11. This means that you would have had to pull the trigger to move to cash on 3/11 to perfectly time this signal, and in this case, it would have worked, as 3/11 was an up day and you would have exited happy. The timer confirmed this weakness on 3/14, and it stayed in CASH until the close of Thursday, 3/24/11's action. Ideally, we would have moved long on Friday, 3/25/11, but weekly signals are good and generally do not cost us either way over the long term.
As an example of where the timer challenged you mentally, it has remained completely long through this whole down-draft for the S-Fund, which is the entire market WITHOUT the S&P500. A good proxy for this is the ETF VXF, which is the Vanguard Extended Market. Put another way, the C-Fund + S-Fund = the entire market, as a broad index.
As I stated above, the S-Fund remained completely LONG during this market downdraft, going from $22.8194 on 2/17 to a low of $21.468 on 3/16, or a drop of -5.92%. This is less than the backtested drawdown of a tad over -9%, so as we were walking in that direction I was wondering whether the indicator was going to toggle or not. Obviously, it did not, although I was completely in CASH based upon other conditions on/around 3/8/11.
The I-Fund, which tracks the EAFE index, moved to a theoretical CASH on 3/14, so again, theoretically, we would have moved in actual dollars on 3/15 close. From the 2/18 peak of $21.2762 the 3/15 close was $19.5573, or a drop of just over -8%. 3/16 was the low point for all the funds, so we perfectly timed the bottom, which is a terrible thing in hindsight this time, but in the big picture, had the markets continued south, the C-Fund and I-Fund signals would have looked kike good timing indicators overall.
To recap. According to the GGT Adaptive Timer method:
- C-Fund, which tracks the S&P500, moved to cash on 3/10 and moved back long on 3/24.
- I-Fund, which tracks the MCSI EAFE index, moved to cash on 3/14 and moved back long on 3/22.
- S-Fund, which tracks the ex-S&P500 market, never transitioned to cash and is still long.
=======================
GGT TSP Timer Method
This final method is the grand-daddy of my methods but it uses ETFs, not the actual TSP funds to determine the timer signals. Here's the most recent dashboard view:
The red/green data on the right side of the figure is the most recent data to the oldest (2-week-ago) data, from left to right.
What the figure shows us is that for the past two weeks, at least since 3/11/11, the SPY (GGT proxy for the C-Fund) has been in CASH. In fact, the SPY moved to cash with the close of markets on 3/10/11. Note that this is in contrast to the adaptive timer method discussed above, which does not use volume (this GGT Timer does use volume).
Next in line is the VXF (proxy for the S-Fund). The VXF transitioned to CASH on 3/10/11, and on 3/21/11 it moved back long. We see that on Friday it also confirmed this move long, with the "Affirmed Long" recommendation waving wildly at us.
EFA (proxy for the (I-Fund), transitioned to cash on 3/7/11 and was a small part of the rationale for moving the entire portfolio to CASH on 3/8. As you can see above, it is presently in CASH. This is in contrast to the adaptive timer method discussed above, and again, is because volume is really poor.
=======================
TSP Allocations
Here's the performance over the last month of the 4 proxies that I use for TSP funds:
As you can see, VXF (S-Fund) is up 0.14%, and EFA (F-Fund) is down -2.66% for the previous month (actually, the S-Fund is even at 0.00% for the period 2/28 - 3/25, and the S-Fund is down -3.35% for the same period). The question now is the amount of allocations.
Brute-force and simple, the following are allocations for the 4-portfolio method using the most recent performance as an allocator for future performance:
- C-Fund/SPY (26%)
- F-Fund/AGG (31%)
- I-Fund/EFA (9%)
- S-Fund/VXF (34%)
Note that these weightings are based upon the performance of the past month and assume that the near future will resemble the past (not necessarily a good assumption, but it is a good starting point). Using the past yearly history of daily volatilities, the expected return of the portfolio above with these allocations is about 7.1% with an expected volatility (fully invested) of about 10%.
Using some advanced mathematics, it's possible to improve on the balance above. If you're interested in how to do this I can point you in the right direction to do this work yourself. New allocations that optimize the reward/variance ratio (also known as the Sharpe Ratio) are as follows:
- C-Fund/SPY (98%)
- F-Fund/AGG (0%)
- I-Fund/EFA (1%)
- S-Fund/VXF (1%)
Note that this zeros the F-Fund/AGG -- this is because all we have is the past to evaluate where we have been, and over the last year of data, the F-Fund/AGG has seriously underperformed. Hence, in the optimization, it gets no weight.
Also note that almost everything moves into the C-Fund/SPY. While this may seem counter-intuitive, as heavily loading into the C-Fund/SPY would apparently increase risk, in fact, it is the combination above that minimizes risk for a given maximum return, which is our goal in the long haul. The underlying issue is one of benchmark -- we have to optimize to something. In this case I picked the S&P500, simply because it outperformed the Russell 2000, Nasdaq, and DJ30 over the past year. Hence, the optimization answers the question "How do we outperform the portfolio compared to benchmark but not take on any more risk?". The balance above achieves this goal.
The graph of this variance vs. gain phenomenon is interesting:
The graph above shows the expected performance and volatility using allocations derived from the previous 30 days of data as a yellow diamond, and the optimized portfolio's expected return/volatility is shown with a purple diamond. While there is more volatility in the new portfolio, there is more gain, which is what we expect and are after. Although portfolios do exist (on the blue line, to be exact) that can provide higher gains at higher volatilities, none are as efficient as the purple diamond above. Above the purple diamond, it is impossible to maintain the same rate of change in return/volatility, as volatility will increase faster than rate of return, or put another way, risk increases faster than return above the purple diamond.
I personally do not play the F-Fund, as I think the tying up of dollars in this fund results in underperformance. Correspondingly, here are my non-optimized allocations for the 3-portfolio method:
- C-Fund/SPY (38%)
- I-Fund/EFA (13%)
- S-Fund/VXF (49%)
Again, note that these weightings are based upon the performance of the past month and assume that the near future will resemble the past (not necessarily a good assumption, but it is a good starting point). Using the past yearly history of daily volatilities, the expected return of the portfolio above with these allocations is about 10.7% with an expected volatility (fully invested) of about 16%.
As you would expect, the 3-fund portfolio that was optimized above has the same optimal weightings here. Here's a graph of expected volatility vs. gain for the 3-fund portfolio:
In the figure above, the previous month's data/allocations are shown as the yellow diamond, and the purple diamond represents the optimized allocation. Note that the purple diamond is below the yellow diamond -- risk is higher with my proposed allocation using the last 30 days of data as a source, whereas the data of the past year suggests that the performance will be less. No matter what the performance expectations, the purple diamond represents that portfolio which has the best return/volatility value, and above this value (on the blue line), volatility is increasing faster than gain.
Recap: to best balance return/risk, the suggested allocation for Monday is:
- I-Fund/EFA (1%)
- C-Fund/SPY (98%)
- S-Fund/VXF (1%)
It's obviously up to you to choose whatever allocation meets your ability to withstand volatility.
=========================
Trading Plan for Monday, March 28th
I intend to move from the G-Fund (cash) to the optimal allocations for a 3-fund portfolio as listed above. Note that I'll make the request before 12:00, so that it is effective on the close Monday.
Regards,
pgd
