August 10, 2017
Vista Analytics has successfully demonstrated prediction in both the likelihood of settlement and the dollar value of the settlement in securities fraud class action lawsuits. The Vista solution can be applied in the initial stages of litigation before costly discovery is undertaken and therefore can aid corporates in formulating earliest stage strategy. Though securities fraud cases were used for initial research, Vista believes that this machine learning solution can be adapted and applied to most types of courtroom predictions where sufficient data is available, including cases involving IP, employment law, and class action cases regardless of underlying cause.
Co-founder and managing partner Craig Freeman notes, “The question of predicting the dollar value of a settlement will without question always have some degree of error. Settlements occur and are calculated based on a series of decisions by counsel that are not measurable. However, the goal is to calculate the order of magnitude of a settlement. 35% on either side of the median may sound large but if the system were to tell the relevant lawyers that a case had a 87% chance of settling with the settlement amount predicted at between $18,000,000 and $22,000,000 we believe that is actionable data.”
In the setting of Supreme Court decision rulings, machine learning has predicted more accurate results than seasoned lawyers and law professionals’ predictions. In patent litigation, machine learning has been used to predict the success of a patent and its financial impact. Existing settlement prediction analyses in securities fraud focuses primarily on loss causation and economic damage estimation. In contrast to the ex post diagnostic explanatory studies, Vista Analytics provides a prescriptive machine learning solution in settlement prediction.
Vista Analytics’ analysis shows that with adequate historical data, machine learning algorithms can potentially provide an 85-90% chance of predicting the likelihood of settlement. Although certain events—such as the 2008 financial crisis—can skew results in dramatic fashion, by taking into account advanced remedies regarding concept drift Vista can normalize the results much more quickly and accurately than was previously possible using a static view of the prior data.
A comprehensive analysis of the…