Explanation HAG vs. Z and an update

Today, System HAG holds Ebay and buys Valero, Broadcom, and Marvell.

To answer a few questions, System HAG is running on a purely theoretical basis. I have a set of algorithms which I put market data into, and System HAG tells me what to buy and whether or not to sell what I've bought. When calculating returns, I assume that all trades happen at the closing price and that there are no commissions.

For the most part, System HAG is a contrarian program that operates on the universe of securities that trade with the most momentum and volatility. When backtesting the algorithms on market data from 1995 onwards, the returns are quite spectacular. Many years, the returns hit triple digits, and the years 2000-2002 were no worse than the bubble years. Returns slowed down considerably from 2003-2005 as volatility in the market decreased, but never went negative.

In all, the average trade makes over a 1% return and the average holding period per trade is less than 2 days. There were many trade opportunities in the years 95-02, and there were fewer in the years 03-o5. Most significantly, monthly returns are little correlated with monthly market returns. However, the program is sometimes 100% long very volatile stocks, so overnight and daily risk is quite significant.

Though the algorithm set in System HAG is the same set that has been used in the backtesting of the strategy, the universe of stocks that the program is applied to is somewhat qualitative. I want to run the program on the most speculative stocks. Ideally, this is an anti-daytrader program that exploits emotional behavior. I deem stocks "speculative" on a qualitative basis.

System Z is a real-life trading account that tries to match, or even surpass, the returns of System HAG. Many more decisions are made for System Z: Stocks actually have to be bought and sold by me at actual trading prices. Paying brokerage costs, liquidity costs, and execution error costs significantly impacts returns. Hopefully, I can make up the discreptancy by correctly determining the best candidates that System HAG offers. By comparing returns, I can test the effectiveness of my input.