www.robotpickedstocs.com

Here are a selction of stocks picked by software for short term trading on 7 July.

The detail about how it works is at : -

http://www.moneyam.com/InvestorsRoom/ShowPostList?fID=1&tID=1008

If you have any comments or suggestions (helpful ones) then please post them here.



AHT, BPG

www.robotpickedstocks.com

Haystack - cheers. That's very impressive.

I think most systems have not worked too well recently, but was quite interested as some seemed to get picked at the point they meet resistance. Of course, the system doesn't really react to that kind of thing I guess.

I'll continue to watch as it's a momentum system, unless I've picked it up wrongly, and that's something I've never been good at. Have been trying, but still find it hard to enter at new highs (or lows for that matter)! Mental thing.

Al

I never am quite sure why the software has picked a particlular stock as the process is quite long (about 6 programs are run sequentially on the whole set of UK stocks). One factor that has been important is that any movements are backed up with volume. It tends to pick stocks with rising volumes or volumes that are out of the ordinary for that stock (other things being equal).

Thanks.

It is interesting to see as well that the two good risers from yesterday are both up today another 1.5% and 2.5%. That is often the case. The picks have to be looked at over several days.

I can see that.

Like I said above, it seems a decent momentum system, but if it picks up volume spikes then it is well-founded.

I was interested in the MGCR pick the other day. I watch that stock a lot and actually went short at resistance. It proved correct, but that doesn't matter much.

Does it also look for new highs breaking out (or lows for that matter) on good volume? Not trying to do a granny and eggs thing here, but just interested.

Not specifically. It makes many measurements on all the LSE stocks and dumps them all in a file. It looks back over a time period and picks out stocks that have performed well over a few days. It then looks at the measurements of those stocks. It tries to find the factors that were significant in those stocks rising and produces a scoring model.

It then applies the scoring model across all the measurements it took at the start for all stocks and choses the ones with the highest scores. There are a few other bits and bobs that it does and there is an element of learning in the system.

It back tests the scoring model against the previous couple of years and looks at how much money the system would have made. On occasions it runs again and alters the scoring model. I have 5 different versions of the software with different biases and compare the results. The software picks more than 2 stocks per day and I usually just post the two with the smallest spread.