Minimizing the Negative about Prediction

Minimizing the Negative about Prediction

By Karl Kern, Founder/President, Kern Analytics LLC

FP&A Analytics, FP&A Trends, Predictive Analytics

Prediction is an important part of the work that FP&A practitioners do.  This work has many challenges.  One way to address these challenges is by minimizing the negative.

The first step in minimizing the negative about prediction is to establish a framework for action.  A way to establish this framework is the application of a quote from Bill James in his praise of the book The Signal and the Noise by Nate Silver: “We tend not to take prediction seriously because, on some level, we know that we don’t know.”  The keywords in this quote are “we don’t know.”  We may know about situations generally, however, we don’t know about situations specifically.  We don’t specifically know about the situations with customers in regard to their ability to maintain or increase their level of purchases.  We don’t specifically know about the situations with vendors in regard to their ability to maintain or increase their level of sales.  Framing around our ability to not know specifically can guide our actions when minimizing the negative about prediction.

The second step in minimizing the negative about prediction is to identify errors in judgement, i.e. biases.  Perhaps the best source to help us identify biases is the book Thinking, Fast and Slow by Daniel Kahneman.  Daniel Kahneman has spent decades identifying biases in decision making so we don’t have to “recreate the wheel.”  This book, therefore, should serve as a valuable resource in minimizing the negative about prediction.  Its value comes in two forms.  The first form is its description of biases that occur in real-life situations.  The second form is its description of addressing biases that occur in real-life situations that serves as the basis for third step in minimizing the negative about prediction.

The third step in minimizing the negative about prediction is to manage biases.  Biases are errors and we as humans are not perfect so no matter how hard we try the elimination of biases will not happen so we must manage rather than eliminate.  Managing is about organizing activities into a process and this serves as a way to address biases when making predictions.  Like identifying biases Daniel Kahneman in his book Thinking, Fast and Slow provides a number of ways for us to manage biases.  One way is to establish baselines for predictions based on data from similar projects that were previously done.  Another way is to use pre-mortems during the prediction process; pre-mortems allow decision makers to look into the future and determine reasons for the failure to achieve predicted outcomes.  These ways allow us to manage bias due to optimism.

Prediction can be a difficult task.  Its difficulty is due partly to our inability in having complete knowledge about situations.  Prediction within this environment can cause adverse effects however there are methodologies in place that can help us minimize the negative effects.