I regularly find patterns amidst seas of data that regularly lose me money because they feign causation while only being coincidentally correlated.
Even more frightening…
You can find more of those goodies at http://www.tylervigen.com/
I found myself searching for examples of spurious correlations after reading this note on the infamous “Sell in May and Go Away” adage from BlackRock:
A cursory glance would suggest that there is some evidence behind this rule. Looking at S&P 500 monthly data going back to 1927, history seems to support the notion that returns are indeed weaker in the summer and early fall. Over that time period, the average monthly index return in the May-September period has been 0.35%, net of dividends, while the average return for the other months was 0.78%.
However, the real culprit is September. Since 1927 the average return for September is -1.07%, significantly below the average for all the other months. If you repeat the exercise looking at May through August, the average monthly return is 0.70%, close to the average for all the other months. In other words, while September does tend to be unusually weak, the conclusion doesn’t extend to the entire summer
Read more: http://www.blackrockblog.com/2014/05/08/sell/#ixzz31hyNib4G
Again, I turn to you. Link to pictures (or gifs) if you can.
1. What was the most painful correlation you’ve mistaken for causation?
2. What are the most interesting/helpful correlations you have seen recently in the Forex market?
