I should amend this. The key is that what you buy and what you sell, the spread between ranks should be greater than or equal to 4. So say you have a bullish outlook. Then you could chose things that are less than or equal to -4 to sell (long puts), and choose things that are more than or equal to 0 to buy (obviously choosing things that have a tendency to move in similar % moves). Or chose things that are >=4 and sell things that are <= -1. Etc. That way you are short some things and long others. Clearly there is an infinite way to do this, and not only that, there is an infinite weighting scheme you can use. You could do the diversified theory, choosing five to ten stocks from different sectors from your buy/sell list. Or you might be able to do pair trades. Etc. This is all research to be done.
T has been held back in sympathy with VZs problems. That said, it is now at a rank of 1, which while in this market that is a "good" ranking, there is no reason for me to hold it over say SPY. In other words, I am (predicting) not getting above the market rewards. So exited at 28.62.
Later today, I will be ranking Rentec's "top holdings" portfolio: http://www.stockpickr.com/pro/portfolio/renaissance-technologies/ They have over 2800 positions. If I could get them all, I would rank them here.
T, 1 GE, -1 SHY, -1 BIIB, -2 NEM, 0 ABV, -3 VZ, -3 ALK, 0 LULU, -6 NVS, -2 VOD, 0 AMGN, 0 KO, 0 MJN, -3 CMG, -6 JPM, 0 PCLN, -6 BMY, -3 MCD, -2 ALTR, -3 NVO, -3 XOM, -1 CL, -3 PM, -2 PG, -2 GOOG, -4 INTC, -1 LO, -1 LLY, -1 AAPL, -4
ENTR, 1 As I see relatively strong stocks, I will post them. Getting a measly 1 in this environment is like a 5 in normal times.
Because the entire world seems to be talking about the XLF and its components, I thought I would bring perspective to it by posting cold hard numbers and no opinion: XLF, -1 BAC, -1 C, 0 GS, 0 JPM, 1 MS, -4 TRV, -1 WFC, -1 As you can see, other than MS, the XLF and components contain no more model-risk than anything else in the market currently. Short MS Long GS suggests itself.
Hang on... so you have what appears to be a quantitative model spitting out some sort of ranks/scores. You have no way of backtesting it. How do you know it works? How do you know those ranks correspond to expected returns in any way?