Trend Lines and Superforecasting

Friday, 12 July 2019.

Trend Lines Chart
EURUSD Daily Chart. Forecast 1.1360 @ 17:00 cet Friday 12 July.

I’ve just finished reading Superforecasting by Philip Tetlock. The title alone should be an irresistible hook for any trader, sadly experience and research shows that Day Traders are loathe to study and rather spend their time backtesting arrays of technical indicators.

Anyway, in the spirit of Superforecasting I’m going public with a forecast (see chart caption above) and at the same time posing a question about Trend Lines. First, the question.

In this chart the dashed line is a strict “higher lows” trend line, that hasn’t been fully tested. The dotted trendline skips the first low, a hideous spike that must have been caused by some rogue economic data or rumours, and forms a nice channel with the solid line above. The question is, “is the dotted line or the dashed line the correct one?” It’s a question of strictness that a hedgehog will answer without hesitation, but a fox (like me) has to write a blog post about it. Foxes and Hedgehogs are explained in Superforecasting.

The dotted line poses a problem, is that recent overshoot acceptable or does it indicate the up channel is broken? Bottom line, Long or Short here?

Getting back to the public forecast. I entered long 1.1243 on Wednesday. I could have closed yesterday and taken the rest of the week off. However, my forecast was for 1.1360 at 17:00 cet on Friday, my inner hedgehog is telling me to wait it out. Yesterday’s spike is really not helping, replies the fox. The main driver for my outlook was Powell’s testimony on Wednesday, the implications are still filtering through.

I haven’t done the full Brier score bit, I’m still new to Superforecasting. In the meantime, in the spirit of Niall Ferguson’s warning about impending hyperinflation, I’m moving my forecast to next Friday – it just needs more time, I’ll revise the target price later.

Multicollinearity

A horrible unreadable chart
Too Many Indicators, an unreadable chart

The most important thing to know about technical indicators is that what you are seeing is in the past.  So, it doesn’t matter if you use a Moving Average or MACD, it’s all in the past.  Also, MACD is a calculation based on Moving Averages so if you have both, you are cluttering your charts with two views of the same thing.  This is Multicollinearity, if your indicators are based on closing prices, you only need one of them.

Mean Regression is the justification for trading Moving Averages, the price reverts to the mean (average).  However, you just have to try it, or look at a chart to see that just as often the Average will catch up to the price.

Unlike natural phenomena, say an athletes performance, price is not bound to revert to the mean because the underlying factors that determine performance aren’t limited. If the ECB launches a new QE program then today’s Euro won’t be worth as much as yesterday’s Euro, there is no point sitting there waiting for it to revert to the mean. Price Discovery kicks in, and that’s one very important element that does not have a Technical Indicator.  In such a scenario, Technical Analysis will have to be suspended until the price settles into a new Trading Range.

Athletes performance? Yes, the Sports Illustrated effect. When an athlete in a rich vein of form makes it to the cover of Sports Illustrated, his performance drops. Aka, the Curse of Sports Illustrated. In this scenario, the athlete is actually performing above his mean for a period and gets noticed. So they put his picture on the front cover. Then, his performance reverts to the mean, and Sports Illustrated gets the blame. The reality is nothing happened, performance is variable in any sport. Each athlete has certain average or mean where he performs most of the time, occasionally raising his game, sometimes dipping below.

Disclaimer