My Books

I have five books currently in production, with over 90,000 copies in circulation.

The Books page will take you to detailed information about each book, including sample chapters and downloadable code.  (Use the navigation bar above)

This Post Has 6 Comments

  1. Hi Howard,

    Thank you for your great detailed explanation on why the examples focus on long/flat.

    I finally got the MTSP book, and am enjoying it a lot so far. I do have a question on the Monte Carlo simulation of roulette.

    There is a difference in expectancy between the -4.26% (page88) and -5.26%(page91). I can calculate the numbers and see that they result different, but why are they both called the expectancy? Roulette should only have one true expectancy (-5.26%) right?

    An other minor thing I caught my eye on was that the references to the figures on page 77,83,85,86 and 101 do not correspond with the shown figures (in all cases the text sais figure 5.x while we are in chapter 6) Maybe you already revised this.

    With kind regards,


  2. Hello Howard,

    I have a question about your latest book where in chapter 5 you show the difference in risk between short and long positions (page 133).

    As you state that there may be other reasons for not taking short positions, I was wondering what your reasons are that most of your presentations, and the theory in QTA are focussed on long/flat trading rather than short/flat?

    With kind regards,

    1. Hi Rudy —

      I focus on long/flat for several reasons.
      1. For a given tradable issue (particularly individual stocks and equity futures and ETFs), long/flat and short/flat models must be different to capture the differences between periods of rising price and falling price. Rises are longer in length, drops are shorter in length and often steeper. Consequently, the model that signals a good entry for a long position is often not the best exit from the short position, and vice versa. Discussing both long/flat and short/flat adds complexity and possible confusion to the text without adding new material.
      2. Many issues, again primarily equity-related issues, have a higher proportion of well defined bottoms than well defined tops. It is easier to detect a good entry for a long position than for a short position.
      3. Long positions get an advantage from underlying economics and sentiment — gains in productivity, increase in population, increase is standard of living, patriotism for longs, inflation — making longs both better trades and more socially acceptable trades.

      All of that being said, for issues where the price is more symmetric than equities — such as currencies — the same techniques that are used to develop and manage long/flat systems work for short/flat systems. Even for those issues that are being traded long/flat/short, I recommend developing long/flat separate from short/flat. The models are simpler, with fewer components, fewer rules, and fewer adjustable parameters. If the long/flat system goes out of synchronization, it can be managed separately without concern for the short/flat portion.

      Best regards,

  3. Hello Howard,

    In light of your recent comments on the Yahoo Amibroker forum, could you please give us an update on the progress of the new “Machine Learning” book you are currently writing?

    Kind Regards

    1. Hi Richard —

      The book is coming along nicely. It is one project of several I am working on. I do not have an estimate of the date it will be available, but several months at the very least.


Leave a Reply