Fishing in the Right Pond

Having a good trading method is only part of the problem.  One must also find the right trading "universe."

The original Turtle Traders learned this, and so should we.

We have been getting some good questions about our TCA-ETF system and why we chose the IShares universe that we have reported each week for many months.

Our criteria were logical and not overly restrictive:

  • We wanted an open-ended ETF, since we did not want to be concerned with a possible discount or premium to NAV.
  • We wished to avoid ETF’s that were cap-weighted in sectors where a very few stocks would dominate.
  • We wanted plenty of liquidity, so that the program could be scaled up as our assets grew.
  • We also wanted enough trading action so that the slippage from our moves would not be too great.
  • We needed enough data so that the model could be employed effectively.

The last point has been a big restriction, since we needed about one year’s worth of data before a new fund could be added.  Unless there was a simulated history, we could not effectively employ a new ETF for nearly a year.  (At one point we were devising our own sectors and avoided this problem, but the ETF’s are a popular and inexpensive alternative.)

Some Additions

In response to a suggestion from one of our smartest and most knowledgeable investors, we reviewed several of the Van Eck Global Market Vector ETF’s to our trading universe. 

Since some of the prime candidates did not have the requisite trading history, although all other criteria were met, we asked Vince to revisit his methodology.  Without giving away Vince’s secrets, let us say that his approach does a lot of filtering to reduce the signal-to-noise ratio.  He made some adjustments to achieve greater stability.  It is non-linear filtering that involves a "pathological time series."

The happy result is that we can now include ETF’s with a four-month data history.  One of the new choices is currently in the top eight, so we will be buying it on tomorrow’s opening.  The new ETF’s will become part of our weekly report.  The new alternatives include alternative energy, gaming, coal, agribusiness, Russia, and nuclear energy.

An Invitation to Readers

We have offered our results as a general benefit to the community, reporting with a one-day lag on a weekly basis.  We now have a weekly program for investors as well as our sector partnership which trades daily.

We are open to suggestions about new ETF’s that should be part of the universe.  Our experience has been that any ETF that meets our screening tests, improves long-term performance.  Feel free to make suggestions.

You may also like


  • Bill aka NO DooDahs! May 4, 2008  


  • Saeid Nourizadeh May 5, 2008  

    Hello, Could you please send me this report?
    “Interested readers can get a report via email on participating in our weekly trading program for individual investors.”

  • Mike C May 5, 2008  

    Have you considered testing/applying your system to the various pure commodity ETFs? Maybe it provide a superior result to many of the passive commodity indices like the Dow Jones AIG index.
    It is now possible to get a pretty fine level of granularity at the individual commodity level. You’ve got USO for oil, UNG for natural gas, UGA for gasoline, DBA for agriculture, GLD for gold, SLV for silver, etc. At any given point in time, it seems like there are a few individual commodities outperforming and underperforming the passive index by a wide margin (look at oil, gasoline, and NG over the past 1-2 months versus gold and silver).
    My commodity exposure the past 4 years has always been to the passive index via the PIMCO Commodity Real Return Fund which has provided quite good returns and diversification and non-correlation to stocks, but I’ve wondered if some type of rotational system across the individual commodities would be superior.

  • Bill aka NO DooDahs! May 6, 2008  

    Hmm, a system that tracks ETFs for all asset classes, including subsets of commodities, bonds, foreign countries, REITs, currencies, and domestic industries.
    Novel idea!

  • Jeff Miller May 6, 2008  

    Mike – We have considered various commodity approaches. For the characteristics of our model and our validation process, we have not found the additions to be useful. Without going into a lot of information about our research, there were always better sector choices.
    One limitation for us is the need to do extensive testing over multiple market cycles and still preserve plenty of out-of-sample data.
    Others with a different approach to development might be successful, as Bill’s comment suggests.
    Thanks for the idea.