Case based reasoning approach for transaction outcomes prediction on currency markets

This paper presents a case based reasoning approachfor making profit in the foreign exchange (forex) marketwith controlled risk using k nearest neighbour (kNN) and improvingon the results with neural networks (NNs) and a combinationof both. Although many professionals have proven that exchangerates can be forecast using neural networks for example, poortrading strategies and unpredictable market fluctuation caninevitably still result in substantial loss. As a result, the methodproposed in this paper will focus on predicting the outcome ofpotential trades with fixed stop loss (ST) and take profit (TP)positions1, in terms of a win or loss. With the help of the MonteCarlo method, randomly generated trades together with differenttraditional technical indicators are fed into the models, resultingin a win or lose output. This is clearly a case based reasoningapproach, in terms of searching similar past trade setups forselecting successful trades. There are several advantages overclassical forecasting associated with such an approach, and thetechnique presented in this paper brings a novel perspectiveto problem of exchange trades predictability. The strategiesimplemented have not been empirically investigated with suchwide a range of time granularities as is done in this paper, inany to the authors known academic literature. The profitabilityof this approach is back-tested at the end of this paper and highlyencouraging results are reported.