I have written a suite of software programs which use OHLCV data (Open, High, Low, Close and Volume). The programs look at all the LSE stocks each night. They pick out stocks which have risen over several days in the past. They may be rising from support or rising through resistance or just rising. The programs then attempt to find what attributes these stocks had before and during their rises. A lot of factors are considered: -

Price rise
Volume rise
Volume not down
Gap up
Closed below high
Price higher now than n days ago (maybe 2, 5 etc.)
Movement above various moving averages (usually weighted)
A type of MACD
A momentum measure
% price rise over a period
number of price dips or rises over a period
etc.

etc. means quite a lot of other test which I am not keen to reveal at present.

The result is a sort of formula or 'scoring model'. The scoring model is analogous to the ones used in credit scoring. You get so many points for living in a particular demographic area found by postcode, so many points for having a bank account, so many points for owning a house and negative points if you don't etc. etc. The end result is that if you get 483 points you get a loan and any less then you don't.

The programs then apply the scoring model to all of the LSE stocks. They ignore ones with a very low NMS (Normal Market Size) and ignore stocks with a very low share price as the spread usually kills any ideas you may have of a quick trade (quick trade being a day or a few days). The ones with the highest scores are then output. These are essentially the ones which fit the factors that would have maximised any return by using the score in the past.

One of the interesting things about the method is that I have no idea why certain stocks have been chosen. I suppose I could get the software to try and explain it to me. It is also a heuristic method in that it does have an element of learning attached to it. The model gets modified every night as there are new trades to be looked at. It uses a rolling period as the range of data to look at. As each day comes a day drops off at the start. It always analyses the same time period that way and it allows for current changes in the market. It looks at its choices and their success and modifies its model accordingly.

It is a bit of fun really. I usually look at the "robots'" choices and select out the ones that I like the best. Here are their selections for Monday 18 August together with my choice of them. I would be interested in any comments or ideas of new factors that I could use : -

AGK, AGS, AU., BA., BVC, CTO, ISIS, ITA, LCI, MSLH, SCR

CTO, ITA, MSLH and SCR look to be the best bets for trading

Yes on the premium board ... he can't afford the subs despite his 150% gains on everything he touches.

:-)