Saturday, July 24, 2010

To Predict or Not to Predict?

By Dr. Israr Qureshi.

Should we stop predicting?...because of the fear that such attempts may not succeed.

Should we choose less uncertain (or to stretch it further, "certain") scenarios to predict what will happen in future? The fun (and utility) of predicting lies in the fact that the business environment is very uncertain.

Going back to the issue of prediction raised in an
earlier post at Strategy and You, let's take the example of the football (soccer) world cup. If we take a probability approach, then predicting each match would be quite difficult as there are infinitely different combinations of passing the ball towards the goal post - when we factor in the presence of so many players in the ground, and each player can take multiple attempts at the ball in each passing sequence.

If we take a capability approach, then also it is difficult to predict, as not all players are equally replaceable (substitutable). Moreover, they play in unique synergistically mutually reinforcing combinations. For example, Muller was made to sit-out in the crucial semi-final with Spain. Now, no one could predict, which player will be made unavailable (either due to injury, or red card (for the same match) or yellow card (for the next match)). Absence of Muller made Klose less effective and resulted in the loss of a crucial match for Germany.

However, does that mean we cannot and should not even try to predict... (and rely on coincidence of random picks, as shown by Paul the Octopus)... my answer is a big NO. Here is why...even though some of us may not have predicted precisely about the Winner, most of us would agree that weaker teams were out of the competition sooner than the stronger teams (e.g. exits in first round). Irrespective of Germany's loss in the semi-final, most of us would agree that they played well, so they have the capability, and they would perform well in coming years (though there will be some chance factors such as the yellow card to Muller). Spain has consistently preformed well past couple of years and they deserved the world cup victory and they will continue to do well in next couple of years.

Thus, what I am trying to point out is that we may not be good in predicting top winners (their exact relative positions) but we are good in predicting broad winners and pathetic losers based on analysis of their relative capabilities.

Does that mean we do not need refinement in our analytical tool-box? Sure, we do need to refine our analysis. We do need to take into account the context, such as disturbing effect of "vuvuzelas", cold weather in South Africa (during this time of the year), high elevation of some match locations, so on and so forth. At times we need to factor in some uncertainties, such as the yellow card to Muller and his subsequent unavailability in the next (crucial) match. However, these conditionalities should not desist us from our endeavor to use all our knowledge, analytics, and wisdom to see where are we heading and how best we should land there.
---

*Dr. Israr Qureshi is an Assistant Professor in the Department of Management and Marketing at Hong Kong Polytechnic University. His recent research has appeared in MIS Quarterly, European Journal of Information Systems and several other prominent outlets.

3 comments:

Anonymous said...

The question is not whether to predict or not, but to know the limits to the predictions we can make and be ready to admit those. I found this on Taleb's website:

"My major hobby is teasing people who take themselves & the quality of their knowledge too seriously & those who don’t have the courage to sometimes say: I don’t know...."

Israr said...

I think most of the statistical and scientific predictions come with confidence interval (which are nothing but upper and lower limits...). In other words, what these confidence interval suggest is that we "DO NOT KNOW" beyond these limits...

BI Qingqing (Claire) said...

I think analytical prediction is like that you use a ruler with fixed scales to test a variable. No matter how accurate the ruler is, you cannot always get the accurate length of this changing object. Though the ruler is not perfect designed for predicting this variable, it is still useful to reduce our uncertainty abut the target object.
Also, I think to predict or not predict depends on individual's perception of the accuracy of the ruler. Though the scales on the ruler are fixed, which enable us to compare results, different people have different perception of the power of the ruler and the different perception of the ruler's predicting ability.

Post a Comment