The Contrarian Trader on... Curve Fitting, Correlations and Technical Analysis

Here at Trader Life, we live all things trading. And generally that means our discussions turn towards the technicalities of trading. And in this new series we are introducing a member of the TraderLife team who has a reputation for their ‘against the grain’ views, so much so that we feel it's time to unleash our Contrarian on you, the reader, to respond to with your own thoughts and opinions.

Regular feature writer Chris Johnston will speak to The Contrarian on a regular basis to get their latest trading views that might raise a few eyebrows, and bring them to you unfiltered. The idea is not to get to a right or wrong answer but to generate conversations around interesting and sometimes opinion-dividing topics within the trading world.

So without further ado, The Contrarian looks into these five hot trading topics...

  1. Curve Fitting
  2. Correlations
  3. Efficient Market Hypothesis
  4. Technical Analysis
  5. Retail Traders

1) Curve Fitting

Curve Fitting or ‘Overfitting’ is one of those topics that inevitably crops up when discussing any sort of back testing. Why do traders back test their strategies? To see if their strategy performs well or not. Even though past results aren’t predictive of any future results, we still want some sort of confirmation that our strategy isn’t completely useless when we click the ‘go live’ button.

Simplified example of overfitting data points on a chart. Source:

It is known that by only looking at a certain amount of data to run and optimise your strategy on can lead to overfitting the strategy, which when deployed in a live environment may completely crash and dash your dreams of becoming a successful trader, especially if it promised you 100%+ returns…

The way to avoid this curve fitting is to conduct out-of-sample testing first. You first perform a backtest on a given test period/timeframe. Then the same backtest is run on a new test period/timeframe, i.e. a different sample of data. If the parameters or settings were over-optimized in the first backtest, it’s unlikely that they will perform well in the second time period.

However….here is where our Contrarian Trader takes a different view. He says you can’t curve fit a strategy:

“…my general view is that you can't really curve fit a strategy, but it is very easy to misunderstand what a back test is really telling you. However, if you fully understand what the in-sample results tell you, it’s possible to work out what is likely to happen in the future with the strategy. The problem is not that people use in-sample data, it’s that they don't understand it or make quick assumptions with it that are frequently wrong.”

2) Correlations

Correlation within finance is a statistic that measures the relationship between the changes of two or more financial instruments over time. Generally calculated and measured using the correlation coefficient that indicates the strength of the relationship:

  • -1 is a perfect negative correlation; the variables tend of move in the opposite direction
  • 0 means no correlation
  • +1 is perfect positive correlation; the variables tend to move in the same direction

Under what is known as modern portfolio theory, you can reduce the overall risk in an investment portfolio and even boost your overall returns by investing in asset combinations that are not correlated. Meaning they don't tend to move in the same way at the same time.

However this is again where our Contrarian Trader steps in...

“Correlations…I think are generally impossible to be accurate on in finance. I understand you can break strategies into underlying parts to get factors to then work out correlations. But the fact is that correlations change so frequently over time and that things can look correlated but aren't.

Part of the reason for this is that finance data is pretty short in terms of history and is also dynamic. In these cases how do you measure correlation mathematically and consistently?

A good example of the problem is that if I toss two coins a million times, at some point their results will be positively correlated and at others they will be negatively correlated. They should, over a long enough time, be zero correlated but on different periods they won't be. But does that mean there is a correlation or not? Consider the same problem on two independent strategies, is it coincidence they are operating together or a market structure?”

3) Technical Analysis

Technical analysis is a popular form of trading where the price of a financial instrument is measured and tracked on a chart to which the trader then seeks to predict the next price move by examining past moves and patterns. It’s probably safe to say that every new trader will look into and use some form of technical analysis, especially as it’s a popular sell on many beginner trading courses.

As always our Contrarian has an opinion on this...

“Technical analysis and trading is absolute hogwash, it’s open to interpretation and personal bias, to the point where one trader thinks it’s a sell and another thinks it’s a buy….and just because price has bounced from a certain level before doesn’t mean it will react the same way when it reaches the same point again.”

4) Efficient Market Hypothesis

There has been many an argument / discussion / research into the Efficient Market Hypothesis, both for and against.

The hypothesis states that share prices already reflect all relevant information and always trade at fair value on exchanges, meaning according to this theory it is impossible to consistently generate alpha and therefore impossible to outperform the overall market through stock selection and timing.

Efficient Market Hypothesis. Source:

Our Contrarian disagrees with the hypothesis. Here’s what he has to say on the subject:

“The Efficient Market Hypothesis is nonsense. The market isn’t remotely efficient. The market is not statistical in nature as humans mess it up by playing with it every day. I like the poker analogies for this. If you put six simply programmed efficient computers playing poker, then the outcome is simply the computer with the best hand, as it will base the entire programme on probabilities. Stick six people round a table and typically it is the person who can lie the best. The market has way too many people messing it up to be efficient …”

5) Retail Traders

The retail trader is one with dreams of making millions of dollars and ending up on a private beach, probably having been dropped off by helicopter. And this is sometimes the selling point on numerous trading courses and seminars. But can these newbie traders ever beat the markets or are they just sold a pipe dream? The well known saying of ‘90% of retail traders lose all their money in 90 days’ or similar must come from somewhere right?

Unsurprisingly, our Contrarian doesn't shy away from the subject:

“With less information and worse technology than the ‘professional’ institutions who trade, retail traders simply can’t beat the markets. They can only beat other - less focused, less prepared, more naive - retail traders…”

So what’s your view? Do you agree with our Contrarian or do you think that you can over optimise a strategy? Have your say, let us know @_TraderLife_