FT4Web by Investors FastTrack
This article discusses Momentum modeling among many sector funds. Momentum's dynamic allocation trading will often produce more return, with less risk, when compared to static allocations.
There is another interesting short article that compares dynamic to static allocation which describes an effective simpler strategy than what is discussed in this article.
There is a related article which deals with Managing Money using the FTAlpha Momentum modeling.
Fidelity and ETFs-
The easy way to start back testing sector funds
What You See is ...
Chart 1's Select AVG, red line, above is a monthly, equal assets, rebalanced average of all the Selects (including money market). Note,
The message to the investor is that Fidelity Selects are well-managed, provide good risk-adjusted return, and are built on 500 stocks of S&P.
Note: This AVG includes FNINX and FSPFX which were discontinued on 6/17/2009. This avoids survivor bias caused by removing these dogs. In fact, the annualized return between 6/17/2009-1/10/2011 did improve by 5.1% annually (not shown).
The Momentum Model
Chart 2's terrific Select AVG red line is created by selling the lowest ranking Select funds at the end of each month, and then buying the highest ranking funds on the first of the month . This differs from Chart 1 which simply held all funds in an equal amount rebalanced monthly.
Chart 2's model only sells 25% of assets each month, and reinvests in the top 10% (by count of funds in the family). This example was ranked by return (gain/loss n the last calendar month)
When you compute an AVG line, you can see what trading took place. Start your favorite text editor like Notepad. Open the file named:
The following is an abbreviated list of trades that were performed to generate Chart 1's red line.
Starting with Chart 1 displayed as discussed above.
Chart 2's Reblance.LOG
A Rebalance.log file is created each time you compute an AVG line. It shows the trading that produced Chart 2's red line.
Rebalance.Log's top line shows that starting at 9/1/88, $100,000 in assets was divided equally among the 30 Select funds which existed at that time. These is no magic in this starting strategy. We will discuss how real people start using the model further down. At the end of the month, 9/30/88, the model sells all or part of the worst funds until $26,025.14 of assets are in cash (25% of the end of month assets). The term "investable" refers to issues which existed and had assets.
These assets are then placed in equal dollar amounts into the top 10%, by count, of funds. Top funds are those which had the highest return during September 1988.
In the beginning . . .
SELECT Monthly Momentum by Return
Rebalance Log on 01-10-2011 at 17:55:00
Scrolling to the bottom of Rebalance.log, you see the final positions on the last day of the database, a 2370% gain since 1988 using a simple, not optimized strategy.
There are an unusually large number of holdings (13) at the end. This occurs in prolonged uneven trendless markets. Usually, the strategy holds 5-6 funds at a time.
As of the end of 2010, the final holdings funds were in three sectors: Energy, Financial Services, and Technology. Two or more funds in each sector were held. There was no human decision made in this allocation, the FT4Web Model did it all.
In the End
The Momentum Model by Return is Flawed
We zoom into Chart 3 to look at the technology bubble of 1999. During the 95 market-day period between the poles, the Red Monthly Momentum line returned (BP=103%) doubled. The model became totally invested in the four Select HITECH funds, then suffering the losses when the bubble popped.
There are other periods where the model became entirely invested in gold and then in energy. Any strategy that becomes nondiversified risks drawdowns. Also note, the high volatility measured in the red Standard Deviation value, SD= 7.24%, in the upper left of the chart compared to the green SD=6.29.
Using Common Sense
FastTrack has many other tools for timing and analysis. Use them. Don't be afraid to deviate from the model by putting some assets into money market when bubble conditions prevail. Although your return will likely be less than the model, your volatility will also be less, and you will sleep better.
Alternate Common Sense, use FTAlpha
Instead of having the model rank by return, rank by FTAlpha
To get the chart below starting from Chart 2, do the following
This chart below shows the years surrounding the Internet bubble.
What you see is . . .
A yellow FTAlpha Monthly Momentum AVG line with excellent return (better than the red and green lines) plus resistance to drawdowns plus a modest Standard Deviation.
FTAlpha's Theory of Operation
FTAlpha gives top rankings to good funds which are not highly correlated to your portfolio AND which have good risk-adjusted return.
Your portfolio, in this example, is defined as the yellow line. At the end of each month. FTAlpha reviews the shape and return of the yellow line, and then ranks the other issues in the family to score how well they would have contributed to better risk-adjusted return in the past month. The highest FTAlpha scores are bought and the lowest sold.
Note that from late 1998 through most of 1999, the yellow line and the red line were similar. However, in late 1999, the yellow FTAlpha line sharply diverged for the red Return line, even to the extent of going down during the technology bubble. This happened because the portfolio became so heavily weighted toward technology in 1999, that FTAlpha began to low rank technology funds. The model dutifully diversified, choosing nontechnology sectors.
When the bubble hit, the yellow FTAlpha line did quite well even in the teeth of the bear market.
To get the chart below starting from Chart 3, do the following
The red is Momentum by Return. The yellow is the Momentum by FTAlpha.
Differences between Model by Return and FTAlpha
The chart tells the story. The Yellow FTAlpha ranking of the Selects has a MUCH better risk-adjusted return over almost any period you measure. However, it has about the same return as the Momentum by Return red line.
From the market top on 10/09/2007 - the market bottom in 3/09/2009. FTAlpha lost -31% . . . far less that the near -43% lost in both the S&P-500 and the Momentum Return ranking. It would have been nice if the FTAlpha model had switched to funds that were not losing money during the period, but ALL 41 Selects lost -24 to -80% for the period . . . except for Select Money Market. However, since money market funds have infinitely high risk adjusted return (any positive return divided by zero risk), they are ignored by FTAlpha.
You are unlikely to follow a strategy (like the red Momentum by Return Line) that takes steep drawdowns. Momentum by FTAlpha is MUCH easier to follow than the Momentum by Return Strategy.
To get the chart below starting from Chart 4, do the following
What do I do today.
To find out what you should trade today, do the following exercise.
Your portfolio, in this example, is defined as VFINX, the Vanguard SP& 500 fund.
What you see is . . .
The green VFINX is the basis for the FTAlpha calculation so, by definition, it's FTAlpha is zero. In this example VFINX is serving as a proxy for your portfolio.
Keys: Select Momentum, Sector Momentum, Momentum Model, Fidelity Momentum, ETF trading Range, ETF Spread, Bid ask, Bid/Ask, underlying index, ETF underlying,