RSI2 Kelly Optimization
The kelly optimal allocation can be extended to strategies as was alluded to in the previous post on non-parametric statistics . In this case, we are using the conventional pearson covariance to demonstrate that the allocation model is useful for strategy allocation. The current lookback for the model inputs are again 3-years or 756 days. In this case the allocation is made between buy and hold the S&P500 index and the use of an RSI2 50/50 system on the same index. The portfolio can go long or short any of the model components. In this case the inclusion of buy and hold is important because if the RSI strategy is not effective than we simply hold the market and thus lower our chances of under-performing the same benchmark (ie buy and hold). This is a subtle point, because it is certainly possible that an effective strategy for a given market does not exist, and thus allocating amongst an inferior set of strategies will substantially increase the chance of under-performing which in turn for many managers DOES impact AUM. As you can see the performance of the allocation model is quite consistent and beats buy and hold nearly every year. A more complex allocation model will be introduced later this week.