1 MARKET TIMING WITH THE ANALYTICAL NETWORK PROCESS

In this paper we develop a stock market timing model based on expert judgments and observable market valuation and sentiment indicators. We apply it to the US stock market. We use the model for monthly, weekly and daily timing decisions over the period from 1994 to 2008. Two different strategies are used: switching between holding the market index and holding cash as well as holding the index and going short. In the case of monthly timing, a buy and hold strategy would have delivered a return of 4.5%, the best ANP timing model on a monthly basis delivered a return of 8.4 % (long-cash) and 8.5 % (long-short); on a weekly basis 8.1 % (long-cash) and 6.5 % (long-short); on a daily basis 14.4 % (long-cash) and 19.2% (long-short). In the daily case we apply White's Reality Check for data snooping and Hansen's test for superior predictive ability (SPA) and for the case of long-cash we find genuine outperformance even while adjusting for data snooping