Boris's space

I recently got my hands into large amount of data for EURO/USD currency exchange rates for a period of 10 months in 2010. It consists of the value at 3,000,000 discrete points of time.

I had greedy intentions, but still tried to keep it scientific. I set myself to answer the ultimate question: “What is the power of the market momentum?” Most traders will tell you that the trend of the market is something very important, but none of them quantifies this “importance”. I first looked at how we “detect” the trend. We connect the dots; eliminate the small fluctuations, put more weight on the latest changes and declare that to be the trend. For instance in the chart below, most people would say that the trend at the end (Feb, 10) is flat. The expectation is that in short period of time the price will not go up or down.

EUR/USD chart from Forex.com

In attempt to mimic this trend detection in our brains, I came with relatively simple technical indicator, called “weighted moving average”. It combines the values in a given window back (say 5 days), putting more weight to the newest. The result is a smoother graph with smaller fluctuations.

Down Jones Industrial Average Index from Yahoo Finance with simple moving average displayed as a green line.

Now back to our question – how important is this trend? The question, however, is even deeper – does the trend, or momentum, really exist?

Let us go back to the promise of the trend – the market has some momentum, e.g. the DJIA is clearly going up in the second half of the graph above. Now let us translate this momentum in mathematical terms – there should be a correlation between the direction of the the moving average changes (up or down) and the actual future changes of the market (again up or down). Increase in the first shows a positive momentum that should be preserved for some short future period. I have run the model on my 3,000,000 data points with a matrix of moving average historical window (e.g. minutes, hours, days back) and the future window (again –seconds, minutes, hours into future). And yes – the momentum was preserved in whole 50% of these data points! No matter how far in the future I looked, or how much of the past I used to calculate the momentum, it never got far from 50%!

The reason is simple – it is like trying to predict the 3,000,000 tosses of a coin. If the coin is fair, you will get it right exactly half of the time. Let us look at the “momentum of the coin”. If you see heads three sequential times, the momentum will predict that the fourth time as another head. However, being independent events, this prediction will be right exactly as much as the opposite prediction. 50% of the time. The coin has no momentum. Just like the random coin tosses, the EUR/USD market for these ten months of 2010 showed absolutely no momentum.

Being humans we try to explain the unexplainable, put order into chaos. What is worse, our minds are set to ignore sudden changes and focus on predictable transitional ones. If you have played Fitness Evolved for Kinect, you may have observed this effect in totally different light. One of the tasks in the game is to break boxes with movements of your hand and leg. The boxes appear suddenly (no transition) on your screen. Most people do a good job in the beginning. Then at certain point, a wall of boxes is built on the back and it nicely breaks into pieces on its own, falling down to the ground. While this is happening on the back, another box appears in the center of your vision. Neither me, nor my friends are able to see the box in front of us, until the animation on the back stops. Yes, we all focus on the gradual and predictable. Just read “The invisible gorilla” book.

Now let us get back for a bit to the EUR/USD currency exchange market. Have I proven that this market is unpredictable? No, I have not. The market momentum is just one piece of the predictability. Non-random coin, switching consistently from heads to tails will be 100% predictable and still show no trend. If the markets are predictable, this predictability is not as simple as we wish to believe.

What I learned? Be very skeptical of analysts who use “the momentum” of the financial markets.

Hello world!

This is my first post in Word Press. I am going to say a few words about myself. I am a computer science graduate, interested in machine learning, futurism and creating new innovative business models. I work for Microsoft Advertising, building the next generation of creative ads. You can learn more about me in my home page: www.vidolov.com.

I hope you will enjoy my future posts.