Price Action Lab Blog

Premium Market Analysis

Market Statistics

Descriptive Statistics are not Predictive

Over the years, elevating technical analysis to a forecasting tool has created the false perception in the minds of a large number of people that descriptive statistics are also predictive. It should be obvious they are not.

Descriptions are easy. There are “technical analysts” that every day write long commentaries about the markets, price levels reached, price action during the day, levels of key indicators, etc. These are description of what occurred and have no predictive power unless placed in the framework of a sound model that has been thoroughly tested.

However, inclusion of technical levels and indicators in market narratives have over the time created the false perception of predictive power because many go back and point to their narrative when right but of course never do that when wrong.

Below is an example of theses misconception from a Twitter interaction that demonstrates the level of confusion and false perception.

On Friday, September 27, 2019, I posted  the following chart in Twitter:

This is the text that was posted with the chart:

$SPX data since 1960, 1948 days w/o -20 drawdown (based on closing price) second best market after 3102 days due to 90s uptrend.

The above was supposed to be a description up to this point in time of the second best market since 1960 based on 20% drawdown from highest-to-lowest close.

To my surprise, a follower replied that on December 24, 2018, the drawdown was larger than 20% and therefore my chart was not as “ominous.”

To start with I did not even think of an ominous chart. There is no sufficient sample to declare a count of 1948 days without a larger than 20% drawdown something ominous. The count can go to 3000, 4000 or even 10000. No one can tell where this will end.

The follower started nitpicking and replying with numbers. The numbers were correct but the thinking and interpretation were wrong. My chart was just a description of what has occurred based on a specific metric (close-to-close drawdown.) Based on highest high-to-lowest close, the follower was correct, as shown in the chart below.

The rise in the count of days without drawdown less that 20% from highest high to lowest close was interrupted on December 24, 2018, and the count was reset and it is now at only 189. The follower implied this is not “ominous.”

The follower of course did not realize that his thinking was flawed since I could change the threshold to 21% drawdown from highest high to lowest low. In that case, the chart would be again “ominous” in his sense (not mine.)

The count went to 1951 days without a 21% drawdown.

Why did I even write a blog post about this? Is this so important?

I think it is. The damage that technical analysis charting has inflicted on the ability of people to deal with markets and price action is irreparable and this was an example only.

Most people that use charts cannot think in terms of statistics and models. This is the overwhelming majority of market participants. This is one reason that some chartists are so popular while most quants are not.

However, the most severe problem as I mentioned above is that most people who were trained with technical analysis have suffered irreparable damage in the way they perceive markets.

For more information about the drawbacks of technical analysis see my interview with John Navin for Forbes.

If you found this article interesting, I invite you follow this blog via any of the methods below.

Subscribe via RSS or Email, or follow us on Twitter

If you have any questions or comments, happy to connect on Twitter: @mikeharrisNY

Charting and backtesting program: Amibroker


Technical and quantitative analysis of major stock indexes and 34 popular ETFs are included in our Weekly Premium Report. Market signals for position traders are offered by our premium Market Signals service

Comments are Closed

Theme by Anders Norén