Is correlation enough to trade with?

Discussion in 'Technical Analysis' started by dnaj65000, Jan 25, 2004.

  1. I got into a philosophical debate with a Microbiology undergraduate friend of mine over beers tonite. The topic was, is it enough to base (life/trading) decisions on knowing the correlation without knowing the causation or reason why something happens?

    example: Who wins the Superbowl has a high correlation that it will either be a bull market or bear market. I say that is good enough reason to bias long or short for the year.

    example: In roulette, a table will typically show either Red or Black being landed on 52% of the time. I say the odds are in your favor to bet the color that is showing up 52% of the time that night.

    example: The only 4 women that ever showed any interest in me are all Scorpios. My ex-g/f is a Scorpio. High correlation.

    hypothetical example: scientists find the underlying DNA strand that is the cause for cancer however have not found a cure yet. A herb that was found in the Amazon, when ingested showed a 70% chance of reversing tumorous cells. Is that enough reason to use the herb for cancer?

    His argument is that I need to know the reason why things happen so that I can make a good judgment in the future. He says by knowing the reason, you can rationalize what your decision is. I don't understand why a reason is necessary because it doesn't change the outcome.

    Technical Analysis is based 100% on finding correlations. If the market does this, then it is likely to then do this. We do not know the underlying reason why it happens, other than supply and demand, so is just having correlation without reason enough to trade with? If someone proved to me that 75% of the time if the price crosses a certain line, and I take a long/short position, it will make me money...I would do it without asking why. IMO, people who ask "why" and are searching for "reasons" are the people who like to listen to CNBC and analysts explain why a certain stock went up. Knowing a reason does not change the price.

    I welcome any thoughts
  2. Quote from "Trading in the Zone" by Mark Douglas (Page 121):

    "A probabilistic mind-set pertaining to trading consists of five fundamental truths.

    1. Anything can happen.
    2. You don't need to know what is going to happen next in order to make money.
    3. There is a random distribution between wins and losses for any given set of variables that define an edge.
    4. An edge is nothing more than an indication of a higher probability of one thing happening over another.
    5. Every moment in the market is unique."

    I have read by more than one experienced trader that if they could start over, they would spend less time on technical analysis education and more time on Money Management Skills and Trading Psychology.

  3. Heavyside the famous mathematician made many discoveries that he couldn't explain. His famous quote to his critics was, "you don't have to understand digestion in order to eat".

    Moreover, we don't understand why Newton's laws of motion work but we have been using them for 300 years.

  4. lescor


    When it comes to trading, you can do anything you want as long as you use stop losses.
  5. selecto

    selecto Guest

    Note that Browning wrote "How do I love thee..." not "why."
  6. Let's take an observation ice cream consumption increases when
    swimming deaths increase
    It is causation?- no its spurious causation - it's the weather
    But is it correlation? - yes
    Would I go long ice cream futures (if I could) when swimming deaths increase? - yes (unless this has already been priced
    into the market)
    Therefore - correlation can be a predictor of market activity.
  7. Cutten


    Is there any reason why you believe that the correlation in question is not just the result of randomness/coincidence?
  8. wdscott



    It is true that technical analysis, in large part, is based on correlations and relationships in determining future price direction.

    It is alson true that:

    Quote from Mark Douglas

    1. Anything can happen.
    2. You don't need to know what is going to happen next in order to make money.
    3. There is a random distribution between wins and losses for any given set of variables that define an edge.
    4. An edge is nothing more than an indication of a higher probability of one thing happening over another.
    5. Every moment in the market is unique."

    As a technical analyst myself, I would first like to clarify the definition of a Technical Analyst. Currently, I believe Technical analysts are narrowly described as people who simply look at charts to find chart patterns.

    I know this to be false. As a member of the MTA Market Technicians Association, a true TA has to learn and understand the following subject matter in order to be be considered as a practicing technician:

    Momentum and Oscillator analysis, moving averages, trend analysis, Japanese techniques, Formation analysis, Cycle analysis, Point & Figure, Market Profile, Market psychology, Volume analysis, Price analysis, Personal psychology, mathematical analysis, Random walk theory, Fundamental analysis, Macro & Micro analysis, Portfolio analysis, Money management, Trade management, Discretionary vs mechanical trading, Regression analysis, Charting, Behavioral techniques, Risk Control, Multiple time frames, Pattern recognition, Testing, etc....

    I agree that trying to trade just looking for a Double top without consideration for other trade basics is a quick road to ruin.

    BTW, after months of reading posts of Elite Trader members, it has come to my attention that most traders do not really know what technical tools they are trading with. For example, traders routinely do not understand the difference between MACD and MACD histogram, Oscillators and Trend momentum indicators, using chart formation patterns in time frames which it was never intended . Example some chart are really not suited for a 1-minute chart, they are more Intermediate or long term but do not get used that way.

    Now I know why TA get a bad name. Sad to say, these are not what I consider practicing Technical Analysts.

    On the subject of correlations. I believe many successful traders correlate their success to a particular skill or trait they possess. Some attribute good mathematical ability, some quick with numbers, other the ability to play poker, understand foreign languages, play music, etc....

    A famous trader trader Mayer Offman who traded huge size and made probably 50-100 million in 1996-2000 attributed his success to a quick ability to calculate and retain enormous amounts of information, due to the fact that he was a CPA.

    His correlation ran out of steam when the bull market ended. And he is no longer profitable. So in fact it is true that our correlations are sometimes false or misplaced.

    In trading though, I believe a correlation or relationship has to have some basis in fact. The reason for this is we as traders, mechanical or systematic, have to develop some kind of market structure or framework ( Market model) in order to continue to execute our trades. Without this framework we will never have the confidence to place trades. For both systematic and mechanical traders, correlations and relationships must have a basis in fact.

    As far as the talking heads on CNBC, their reasons for market actions are always based on an after the fact assessment. Quite easy and truly meaningless.

    As traders, we must execute well before reasons are known. And to do this without understanding correlations and market relationships, trade management, risk, execution skills, psychology - both market and personal, etc......are not really traders, but IMO closer to Las Vegas vacationers .

    Know the reasons to enter and exit a trade, and above all know your personal risk tolerance levels.

    Good Trading

    Best Regards,
    Dave Scott
  9. jaan


    *very* good point. i've seen quite a few traders (and not only traders!) to think there's a correlation between events X and Y, whereas the "correlation" between those events can easily be explained by pure coincidence.

    for example, let's take 2 sequences of 5 numbers:

    a) 7, 1, 2, 5, 10
    b) 0, 4, 7, 10, 0

    there's a very high (anti)correlation between those sequences (-0.6). however, they are simply the first 10 numbers i got from ROUND(RAND()*10;0) function of excel! so the apparent correlation was just a random coincidence.

    back to the subject: if you see an apparent yet unexplained correlation between events, then there's *always* a possibility that you're just witnessing a coincidence. now, if you could work out an underlying explanation for the correlation, you would be much more safe in assuming the events really are correlated.

    in practice, however, the market is too complex system for clean-cut explanations to work, so even if there *is* some correlation, trying to work out the underlying reasons will just yield a meaningless pseudo-explanation (eg the "profit-taking after a run-up" type of explanations).

    in other fields of (natural) science, finding out the reasons behind observed correlations is much more feasible, and pretty much required. hence the view by the microbiologist-guy.

    - jaan
  10. No. I think it's random. But you can find correlation in random data.

    If I were to bet on the next woman, (since I'm single now and have time to contemplate and write something abstract like this thread), I would bet on Scorpio opposed to the other 11 signs based on my correlation. :D
    #10     Jan 25, 2004