Developing an investing/trading system

Here I am taking a fresh approach to how to invest.

1. "Cut your losses, let your winnders run".  This bit of conventional Wal Street thinking seems to hold up well.  The reason is that markets display persistence.  Price movements that appear random in the short term actually tend to have some momentum.  Stocks rising in the past are slightly more likely to rise in the future than stocks that fell in the past, which are slightly more likely to fall in the future.  If one randomly enters many positions and exits all of them that show a small loss, the portfolio should beat the market because a few of the "winners" will produce outsized returns.  I am assuming that the portfolio would have equal exposure to the long and short sides.  I have read that this works even if the entries are random. Yes, the "shorts" are limited to 100% gains, but since prices tend to fall faster than they rise, returns on successful shorts can be faster than returns on successful longs.  So this insight dictates the main strategy for exiting positions.

2. Next, I want to do something more sophisticated and targetted than just randomly entering all names in some universe from either the long or the short side.  In designing this next step, the key assumption is that fundamentals matter in the long term while technical do not, while in the short term the technicals matter and the fundamentals do not.  Furthermore, the technicals have no real predictive power in isolation.  At best, they can only give you slightly better odds in predicitng some aspects of the future, such as how far it might go IF it goes up or vice versa.  The direction of the next tick is wholly random.  Also, the fundamentals are harder and harder to predict the further out in time one goes. 

The process should involve dividing names into three watchlists (buy, sell, and neutral) designed to provide an edge.  These three watchlists would be based on the fundamental, or the long-term, perspective.  Technical indicators would then be used to determine the time to enter a position. E.g. TOS has a list of about a dozen "crossover" signals.  The "buy" list gets entered only from the long side, and vice versa for the "sell" list.  Neutrals could be entered on either side.

3.  The process is subject to continuous improvement.

To begin, I am going to use 52 week highs and lows to create the initial lists.  In TOS, one can create a watchlist via Scan> Load scan Query > Public >New Yearly Highs or New Yearly Lows.  Criteria for this preset scan also include a stock price > 1.00 and volume greater than 1.  You could adjust the lookback period from its defaulted value of 252 (trading days in a year).

4. Items in my daily routine before the open. Review all my positions to be sure that I know what they are and that abort stops are in place.  Enter onto Excel sheet.