Don’t Overthink It

Feedback in today:

Hello Michael or Michael’s assistant, I have 20 years experience in losing money investing in the stock market. I started well however having purchased far out of the money calls for the IRAQ war in 1990 and made 8x in 48 hours, all down hill since. So I read your book with interest and eager to find an alternative investment method but your explanation to me remains incomplete (or I misunderstood). I had the following comments/questions which would be helpful for readers. 1.) Tax – all the simulations about how much money after X years needs to consider tax. Siegels supposed 6-7% constants don’t address this. 2.) Close fit but not too fit – I understand the overall critic of fundamental investing, there is no algorithm. Your trend following description claims to solve this which it does except for what I see is 1 clear caveat – the not “too” perfect fit. Since the book in my opinion correctly mentions that Stats is a great method this is all undermined by then saying you can’t have too close fit because it might not work in the real world. Unfortunately now from a math perspective the algorithm or trading system isn’t a solvable equation. Unless too close is actually defined in stats terms it means that any trading system could then be justified or refuted based on this unsolved variable. 3.) Trading system for all markets – As I get it there is supposed to be one trading system (algorithm) which you then apply to X number of markets. I don’t understand why. The idea is to have an algorithm to address the non-objective or unsystematic approach of traders in each of these markets. Since these markets are selected as noncorrelated they should have different players involved, therefore logically it would make sense to have a different trading system for each market. I am not a baseball connoisseur but if you had an algorithm for baseball you wouldn’t use the same one for basketball. 4.) Monte carlo – As it is described in the appendix (in my book version), this statisically doesn’t seem possible to me. If you design an algorithm around a set of data and that data can be then changed in ANY way then it means that it could no longer have a trend, therefore I don’t see how an algorithm could still work. The point of the algorithm as I get it is just to model some aspects of human behaviour which then provides an edge. Anyway, good success with your book. The investment market is all about selling hope and you have sold me hope. Regards, Daniel

I am sure folks out there have some eloquent responses!

6 thoughts on “Don’t Overthink It

  1. Hi Michael,

    Here are my comments.

    1. You only get to keep a portion of your profits no matter what you do. Deal with it.
    2. All you are trying to do is draw a line in the sand and say “If prices go here, I’ll buy” if they fall back to here, I’ll sell.” Just about anything will work. You don’t want to get into quadratic equations. You end up accounting for variables that aren’t there. It’s all about picking a price and then acting.
    3. If I had an algorithm for baseball I wouldn’t use it on basketball but I would certainly use it on the Cubs the same way I did on the Red Sox. We are watching price. Which price isn’t all that important.
    4. This isn’t science. It is just trying to spot a trend and then act on it. The movement may reverse the next day, you don’t know. There is no way to predict. You define your entry and exit, set your money management to keep you in the game and go. That’s it!

    I think the commentator may have trend following confused with prediction. Many systems do try to do that. That’s why you need to use fancy rules to catch the nuances of each market. That’s also why they fail. Just my two cents. I hope I am close to the mark.

  2. In response to question three I think you should compare the algorithm (trading system) to a batter’s swing. I am a not a professional baseball player but I don’t think the best batters change their swing every time the pitcher changes. They see where the ball the heading and swing in that direction.

  3. The algorithm is would use for comparing basketball to baseball is: the team that scores the most points, wins. That seems to translate from sport to sport (golfers want fewer strokes, but you get the idea I hope). Sure, there may be people that NEED to know Kobe Bryant’s freethrow percentage in the 4th quarter of home playoff games, but I would prefer to know the status of the scoreboard. Sure, there may be people that NEED to know the weather patterns over the corn-growing regions of Iowa in the spring, but I would prefer to know the status of the market (price).

  4. It’s possible to apply similar algorithms to a variety of instruments. Here is a great study of the basic TF algorithms applied to stocks that illustrates the point. 13% return in the paper might not sound like a lot but that’s just the mathematical expectation (edge). If you read their conclusions, it appears that higher returns are possible by pyramiding, and it’s also a very tax efficient system.

  5. People tend to overcomplicate because in their own minds the simple strategy is too simple and therefore can not work.

    This happens in all facets of life, not just investing/trading.

Comments are closed.