Skip to content

Predicting the Market From the Inside Out: Adaptive Breadth

May 5, 2010

I had a lightbulb moment several weeks ago in which I theorized that certain stocks within the index were more likely to be the “drivers” of index returns than others. Different variables such as volume activity and others may be able to help separate which stocks were useful for predicting index returns versus those that were not. The only way to really tell if such a ranking has any power is to do a “walk-forward” test in which stocks are selected each day as “predictors” of the QQQQ by use of the DV2 indicator. Effectively a signal is created from this group to trade the QQQQ. As it turns out, the methodology appears quite powerful–a near linear increase in absolute and risk-adjusted returns by rank. Using the Top 10 stocks to create an adaptive breadth index (averaging the dv2 values of the top stocks) to trade long .5 significantly outperformed using the DV2 on the QQQQ alone (the benchmark). Unlike standard breadth indicators where theories and fables abound concerning how to best apply them, this method is more “assumption free.” As Jeff Pietsch of Market Rewind likes to say: “Markov would be proud!”

Using an Adaptive Rank To Determine Which Nasdaq 100 Stocks Best Predict The QQQQ with the DV2 Indicator

Backtest from 5/22/2000 to 4/30/2010 [2500 Bars]
Benchmark = Long QQQQ DV2 < .50 — Short QQQQ DV2 > .50

13 Comments leave one →
  1. May 5, 2010 10:58 am

    I like the idea of applying adaptive mechanisms to market breadth. Wish I had thought about it earlier ! Good post as always.

    • david varadi permalink*
      May 6, 2010 12:44 pm

      thanks Quantum (like the name btw) just kicking around some interesting ideas for discussion.

  2. May 5, 2010 11:05 am

    awesome. You can almost hear the market distribution of returns shifting to a different power. Can intelligence really be non-artificial? 🙂

    • david varadi permalink*
      May 6, 2010 12:43 pm

      hi alex, thanks—funny its true on that note. this area does bring up a lot of possibilities.

  3. May 5, 2010 5:17 pm


    When you say “Using the Top 10 stocks to create an adaptive breadth index (averaging the dv2 values of the top stocks) to trade long .5 ” are you saying that you get long when the average reading is .5? If so when are you exiting?

    • david varadi permalink*
      May 6, 2010 12:42 pm

      hi Al, thanks for catching that mistake—it was trading long when the dv2 aggregate was below .5 and short above .5. thanks

  4. BMB permalink
    May 5, 2010 5:53 pm

    Can you expand on what you mean by “Top 10”. Is this Top 10 by weight in the index, Top 10 by recent volume, Top 10 by some other ranking (such as your LTR ranking), Top 10 by relative performance?

    • david varadi permalink*
      May 6, 2010 12:42 pm

      hi bmb, the top 10 represent the top 10 “predictors” for the index based on using two different variables to generate signals–at the present this is a proprietary method so I can’t expand on it more than that.

  5. Don permalink
    May 5, 2010 9:09 pm

    How does it work if you just trade the top 10 stocks directly, rather than the index?

    • david varadi permalink*
      May 6, 2010 12:40 pm

      hi don, trading the top 10 stocks as indicated does very well, however the idea is to derive a method for superior prediction of a given asset–in this case the QQQQ. if our goal was to find the top 10 stocks we would have done things differently.

  6. prazor permalink
    May 6, 2010 2:35 am


    Nice idea. My guess is this triggers a lot of research out there…

    If i understand this correctly you are using some sort of
    ranking (livermore or similar) to produce the list of 10 stocks.
    And this is done daily.

    Then you trade the Qs using a weighted DV2 on the list.

    I am curious, how does this compare to trading the list itself using the same rules?

    • david varadi permalink*
      May 6, 2010 12:38 pm

      thanks prazor, your understanding is correct, and trading the list does quite well using the same rules. however even better is using the same idea to generate lists to trade each stock on the list (ok that sounds a bit dizzying) but this general concept is the basis of what a lot of algorithms are doing at investment banks.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: