Most of us enjoy seeing magic tricks. We all know the woman in the box was not really cut in half. Illusions are created to trick our brains into believing we see something we’re not (like Donald Trump actually being the Republican nominee). Our on-call resident Dr. Data sat down with Tom and Tony last week to discuss how statistics can be manipulated to create an illusion.
Markets often have knee-jerk reactions to economic announcements such as unemployment. Currently, that number stands a little above 4%, down from the high teens when the current administration entered office. While the drop is good and more people are working, it sounds less encouraging considering unemployment is higher than it was 81% of the months from 2000 to 2009.
Manipulating statistics can also happen right in front of our eyes. Take for example the slide below comparing an average daily P/L amongst several strategies. The example on the left gives the appearance “strategy three” is most effective. But look closer (go ahead, I’ll wait). The graph on the left shows P/L increases in $0.10 increments. Change those increments to $0.50, as the chart on the left does, and suddenly “strategy three” does not stand out nearly as much:
Another common way we see statistics skewed in a certain direction is through sample size. Tom might argue he’s the best free throw shooter in the room and his argument might be correct, if he’s talking about being in the studio with only Bat. But bring Tom out near the research guys and have Kristi join them and suddenly Tom is not even the second or third or fourth best.
tastytraders use statistics to trade. It is at the heart of understanding our methodology and strategies. Therefore, we should also be the first ones to recognize how statistics, if not given in their proper context, can mislead. Accepting data at face value is no better than passively investing. We are better than simply accepting without questioning.
Josh Fabian has been trading futures and derivatives for more than 25 years.
For more on this topic see: