Smoke and Mirrors. Cut through the hype to find great analytics

Non-Prime Times, the official publication of the National Auto Finance Association.  TruDecision CEO Daniel Parry’s article on analytics featured as the cover story.

James Randi, more popularly known as The Amazing Randi, is a stage magician and a scientific skeptic of the paranormal. He has exposed psychics, astrologers, televangelists and others who claimed otherworldly powers as charlatans who were using the same tricks he employed as a professional magician. He offered a $1 million prize for anyone who could prove psychic abilities, and over the years put hundreds of people to the test. No one was ever able to claim the prize.

They say it takes two people to make a lie work: the liar and the one who chooses to believe the lie. Part of the reason Randi was so successful is that he not only explained the mechanics of how to execute the trick, but also revealed how the audience played a willing part in their own deception. Some of those who were fooled held to a hope of an incredible future. Others sought to attach themselves

to some powerful force that would give them an advantage in this life. In both cases, the duped were 100 percent willing to believe in the fantastic to satisfy some other longing.

It makes me sad that James Randi is 88 years old, as there is so much great debunking left for him to do in the field of analytics. The parallels between magic and analytics are uncanny, as lenders latch on to hopes of the fantastic, yet often end up with little of value. Fortunately, like parlor tricks, the methods are easy to detect if people are willing to do a little homework up front.

Smoke, mirrors and other distractions
Make no mistake. I am a great fan of analytics and the benefit com- panies can get from the proper use of it. The problem is that many ofthose we look to for solutions are employing smoke and mirrors to sell us something that is not so. For lenders to gain an advantage, they must be able to distinguish between a ‘bill of goods’ and the genuine article.

The French word forte (pronounced “fort”) means a thing at which someone excels, whereas the Italian word forte (pronounced “for-tay”) means loud. The latter is often seen in music manuscripts to denote expression. The misunderstanding of these two words is so ingrained that many people who actu- ally know the difference still use the Italian pronunciation when referring to a skill, or specialty, so that others will not think they are an idiot.

There is a strong parallel between this and the use of analytics. The purpose of statistical modeling is to reduce the complexity of the world to something simpler, so that it may be effectively managed. Model complex- ity increases error, and vice versa. However, many falsely believe that the more complicated an analytic solution is, the better it must be.

I have personally witnessed good analysts complicate otherwise good models because they believed management would not be suf- ficiently impressed with their answer. What is sadder is that too often they are correct. When the leader is focused on the wrong objec- tive, it should not be a surprise that success is elusive. The purpose of top-tier analytics is to provide the best answer to the problem at hand, regardless of how impressive the form of the answer is.