Michael,
I am currently using 57 variables (mostly fundamental) to which
I assign weights.
Each market day, I scan the QP2 database and output a daily database of about 6500 stocks with these 57 variables (plus about 35 more
which are currently not in my valuation screen).
I select the 6 highest stocks for going long and the 6 lowest
for going short.
I devised my variable weights by subjective trial and error,
combined with pursuing high long returns as well as high short returns over the 3 months starting 2-17-1998.
My goal is a hedged 50-50 long-short approach which (hopefully)
will provide returns exceeding the SP500 and with lower volatility.
Since I selected variables and assigned weights to optimize
returns for 3 months forward from 2-17-1998, I cannot be surprised that my approach has outperformed the SP500 for that time frame. However, I am pleased to see results continuing in similar fashion from 5-18-1998 forward.
In particular, I am getting about +14.5% for the July 20, 1998
market top thru Oct 5, 1998. This compares to a -15.5% return
for the SP500 for the same period.
I am currently ignoring commission, spread loss, and dividends
paid on the short position.
I would like to see QP2 add variables like employees, short term debt,
shares out over the last 4 quarters, tax rate, and operating
expenses broken out so I can add them in a weighted fashion and tinker further with this universe of variables.
Of the 57 variables, price, yield, market cap, 12 ttm price gain,
QRS, 12 ttm price gain for this SIC, dollars traded this day,
price to 50 day price, price to 200 day price, and annual sales, seem to be more significant than the other variables. |
