AusDM '12 Proceedings of the Tenth Australasian Data Mining Conference - Volume 134
| Hi-index | 0.00 |
Two data mining techniques were compared for their ability toimprove the prediction of abnormal returns using insider stocktrading data. The two were neural networks (NN) and MultivariateAdaptive Regressive Splines (MARS). In the comparison, bothanalyzed abnormal stock market returns from the same 343 companiesover the identical 4\frac{1}{2} year period (1/93-6/97). The majorfindings were: 1) both NN and MARS generally identified the sameindustries that had the most predictive abnormal stock returns 2)both found that predictions further in the future (12 and 9 monthsahead) were more accurate than predictions closer to the tradingdate (6 and 3 months ahead) 3) both obtained better predictiveaccuracy using four - rather than two - months of back aggregatedstock data 4) NN identified a substantially greater percentage ofstocks in the group with the highest explained variance than didMARS 5) data from small and midsize companies led to higherpredictive accuracy than data from large size (S&P 500)companies using NN, but not MARS. The findings illustrate that thevery complex interaction between insider trading data and abnormalstock returns can be systematically analyzed using non-lineartechniques. Of the two assessed, NN led to comparatively moreaccurate predictions than did MARS.