In Technical Analysis Part 1, I used carefully tested methods to predict where the S&P 500 would be in three weeks.
And now it’s been three weeks. Let me show you how well technical analysis works if you just trust the model (and ignore all the failed predictions). Also, we’ll answer the question: What is confirmation bias?
I Told You So
As a reminder, at the end of the regular trading day on July 23rd, 2018, the S&P 500 was valued at 2806.98.
On July 24th, I told you that come August 15th the same index would be at 2786.38, using the following graph (the thick black line is the prediction line):
Here’s the same graph with the last three weeks of S&P data superimposed:
I predicted the S&P 500 would trade at 2786.38, after a modest pullback. In fact, this is very close to the actual value of 2818.37 (-1.1% error)!
Nothing To See Here
The problem with technical analysis, as with a lot of the prediction business, is that bad predictions die a quiet death. Predict doom and destruction every day, and you’ll be ignored until the moment that destruction hits. There were four other predictions that were incongruous with each other and inaccurate in the end:
These four predictions were:
- All Time High: 2893.04 (+2.6% error)
- Fibonacci Crash: 2499.16 (-11.3% error)
- Fibonacci Rally: 2930.33 (+3.9% error)
- Elliott Wave Pullback: 2763.64 (-2.0% error)
Technical analysis provides a powerful example of confirmation bias. Confirmation bias is our brain’s tendency to value information we already believe and discard information that is contrary to our beliefs. We now know what the market has done on August 15th, therefore we go out looking for data that confirms we were right all along. Contradictory evidence is easy enough to discard and forget.
There is simply no evidence that technical analysis is anything other than the random throwing of darts onto a blank canvas. Days, weeks, and months later, we can draw a target on the wall and find which dart is closest to the bulls eye. But you have better things to do with your time and energy, right?