In Sherlock Holmes’ world drawing inferences, however brilliant, was relatively straightforward: a specific scene of the crime, a limited number of suspects and experience in similar crimes, allowed the data to do the work resulting in the correct conclusions. Sherlock Holmes was actually one of the first advocates of “small data”. The huge rise for big data surpasses these beneficial circumstances Holmes used to work in: a world in which we should “let the data speak for itself”, in which little or no prior knowledge seems to be required. An overkill of data did not allow any central bank or forecasting institute to predict the economic crisis of 2008 or actuaries to predict longevity effects in time. Long Term Capital was built on models and data which worked in normal times but went terribly wrong in abnormal times. Even more depressing are the latest developments on stock returns research: for almost half a century academics acted like alchemists to discover the formula for better returns. Value investing, small stocks, emerging markets and now high frequency trading were considered. In 1988 the Wall Street Journal even let monkeys throw darts to select stocks and compared their choices with professionals. The pros won which seemed to prove that “experts” can add value. The latest research however shows mainly trading-expenses matter in the long run. After half a century of intense data analysis, it appears that using common sense and not predicting the market is actually the best method for beating it. 28 BLOG // 28_06_2016 For almost half a century academics acted like alchemists to discover the formula for better return. ’’ Big data is a massive game changer and the benefits for short term analysis (consumer needs) is almost without limits. The insurance world can use its power for short term products and to predict trends in customer needs. However, for long term decisions, data can be hugely misleading and unfortunately the majority of actuaries are part of a business which has a long-term, not short-term focus. Hence, like Sherlock Holmes use the power of (small) data, but only in incidences where you are fully aware of the conditions of the crime scene. AG_ BLOGS_ 2015_2016
60 Online Touch Home