Abstract
A divergence-based procedure is applied to trading in the S&P 500 stock index. Such a procedure has previously and successfully been applied to trading in foreign currencies. Performance is tested against a benchmark, i.e. a passive portfolio replicating the S&P 500 stock index; its robustness is also checked in a few subperiods. According to our numerical evidence, higher annualised mean returns, i.e. higher final portfolio values, as well as lower annualised standard deviations can be obtained in all subperiods. However, basic Montecarlo tests are failed. The on-line extension of the present off-line implementation is taken into consideration, as it is more suited for an operational use.