The Sector Update — Current Recommendations

Today we begin a new type of report:  The Sector Update.  We will post on a weekly basis the holdings and results of Vince Castelli’s TCA-ETF model.  This is an inside look at how a trading model has been developed and implemented, information that is usually not publicly shared.

This report is aimed at two different groups:

  1. Individual investors.  But not for the reason that one might think.  We are NOT making a trading recommendation for individual investors.  Each investor is different — in goals, risk tolerance, life situation, tax issues,  and other special circumstances.  The individual investor must consider all of these factors and should consult a professional advisor.  On their own, individual investors significantly lag the overall market averages.  Investors should read this series not for recommendations, but to see how difficult  it is to design and implement a trading system.  It is not as easy as the TV ads, where the online brokerages make money by persuading you to trade frequently.
  2. Traders.  We invite comparisons to other systems and rankings.  We invite comment.  While we have a strong method, we always seek to improve what we do.  Our readership includes many astute market observers and professionals, and we welcome reactions.


Our investment company was founded with trading  plans based upon sector investing.  We call ourselves "The Sector Experts."  In nearly ten years we have seen many changes in sector rotation.  There have been several major developments during this time.  Here is some information that we do not think readers will find anywhere else.  These are the changes in sector rotation:

  • A shift away from mutual funds by many investors.  Mutual funds, especially the biggest, are quasi-index funds that gain a performance edge by overweighting certain sectors.  Nine  years ago we could easily track their actions, because it took months for these funds to shift positions.  This led to multi-month trend persistence in sector investing.  This has changed.
  • The rise of ETF’s.  Many individual investors choose Exchange Traded Funds instead of mutual funds, using their own methods for choosing the right sectors.  Right or wrong, these individual traders move markets.
  • Financial consultants and hedge fund managers.  We read many reports from other managers.  Many do not do the intensive homework needed to trade individual stocks, but believe they have a "feel" for market sectors.
  • Faster rotations.  Many managers make swift changes in overall allocations.  This has reduced the period of trend persistence.

The Model

Most developers of systems — both individuals and professionals — make a crucial mistake.  They take a big chunk of recent data and find a system that would have worked during that time period.  This is back-fitting of the model.  Even those who hold out a subset of data may find that their models are not really predictive.  When a model developer repeatedly uses the same out-of-sample data, it gradually becomes "contaminated" — almost as if the test data were part of the development data.  It is the most common mistake.  Any developer who is not constantly on guard for this has probably become a victim.

A first-echelon modeler (like Vince, known well to those on WealthLab) has strong methods for avoiding this problem.  These are described in various books, but beyond our scope here.  Our method introduces even one more check.  We took Vince’s work and applied it to a universe of sectors that he had not even thought about when developing the model — and we used a different time period for the test.  We did not do this dozens of times, we did it once.  When the model showed good results, we had the nearest thing to real-time testing that one could achieve.  This sort of rigorous back-testing is much better than simply trading the model, since one gets plenty of variation in the simulation.  If the methods result in excessive drawdown or risk, that becomes clear from the simulation.

The biggest problem for the system trader is to maintain confidence when things seem to be going wrong.  Unless you  have complete confidence in the development process, you will bail out on your system the first time you see losses.

The Chosen Target

We selected iShares as the target ETF.  We did this because the components were known and the funds are open-ended.  This assures that there will be no "style drift" from managers changing the stock allocations.  The iShares portfolios also have more diversified exposure than some alternatives that heavily weight a few stocks.  Finally, there is little overlap in the underlying stocks, so we can achieve diversification.  (We have no promotional relationship with iShares).

The Rules

The TCA-ETF model (Trend, Cycle, Anticipation) first identifies whether the sector is trending or part of a cycle.  The method builds in a factor for anticipating both entries and exits.  The average holding period is about thirty trading days.

The model does not rank sectors.  A sector either qualifies or it does not.  This creates a problem.  If one allocates a very small proportion of the portfolio to each sector, it maximizes diversification when  there are many sectors.  It also leads to significant underinvestment when there are fewer strong choices, but overall prospects are good.  Sector trading has both a great reward and a concomitant risk.

Reward. To get the reward requires overweighting certain sectors.  That is how sector investors beat the market.

Risk.  Reduced diversification means greater risk exposure, particularly if the trade gets "crowded."  Others may have similar systems.  Even if these are inferior overall, they may be sharing your positions at a given time.  It is important to have a careful exit strategy.

Result.  We base the portfolio on a total of eight positions.  We watch for and limit similarities in the underlying stocks.  If new sectors are proposed by the model, we cannot generally buy when already fully invested.  We might choose to substitute a new sector to increase diversification.

Current Portfolio Sectors

Here are the current iShares positions, showing our date of entry and results to date.


There have been ten sector recommendations since our last addition.  In a future post, we will also list these additional choices.


We present this information to join the debate in the ETF community.  It is not our only method.  Regular readers of "A Dash" know that we have specific stock holdings and option positions based upon fundamental analysis and our view of overall market prospects.  Some of our fund offerings hedge positions.

Each method has its own discipline.  All have been successful, but frequently at different times.

The list reflects the strength of foreign investments and sectors not linked to the economy.  For those with a different time frame, a group of economically sensitive stocks might be stronger.

It is a fascinating feature of markets.  Those with different time frames might be buyer and seller on the same trade.  Both could be correct!

You may also like


  • David Merkel September 14, 2007  

    Jeff, you know I rotate sectors as well. I blend it with a value discipline. The two are a tough balance, but when in doubt I have let value dominate.
    Your methods are interesting. I would assume you have more momentum in your models, because you seem to have a shorter holding period than I do. My sector rotation tends to be more fundamental, driven off of pricing power, and more anti-momentum in nature.
    Anyway, nice article. Thanks for sharing it with us.

  • Bill aka NO DooDahs! September 14, 2007  

    I hope Vince enjoyed the paper I passed along. Hmm, FXI, IEZ, INP, EWH … some of the things we’re looking at are commonalities.

  • Tim September 14, 2007  

    Jeff, if I read you correctly, you have a universe of sectors and you invest in the 8 highest rated? A new top 8 would replace the lowest current rated sector?
    Are you investing funds based 100% on what your model is forecasting, staying 100% invested in the aligned ETFs?
    Curious mind(s) want to know. Thanks.

  • Jeff Miller September 16, 2007  

    Thanks to all for the comments on this new theme.
    David astutely points out that time frames differ in sector rotation. We are talking about actively managed accounts and trying to find a sweet spot for rotation. I need to write more about this.
    Bill sent along a nice paper, and I am still trying to get some comments from Vince, who has been off in some “seclusion” for something. He could tell me about it, but then it would be the end of the blog!
    Tim has his finger on how we do the rotation, and we are considering adding some numeric ratings.
    Thanks to everyone!

  • mxt52 June 26, 2008  

    I wish there were more details on how the sectors are evaluated (what is behind (Trend, Cycle, Anticipation)?). I understand it may be proprietory, but some insight that does not reveal an exact mechanics could be useful. Thanks.