I’ve been working on a bot that can use Google to trade. It’s fairly simple in what it does.
I have a file for each currency pair and each time a google search query turns up a positive result for that pair it upgrades or degrades the pairs’ percentage, either favorable or unfavorable.
The bot searches everything on google from plumbers in Kazakhstan to banana growers in Peru. I have it set to only search “new” pages, etc. So it should be getting up to date results. The results are weighted on past history.
I have no idea of the results. If it is anything other then random, i would be surprised. Although had a perfect first 12 hours ! (it wont last)
I think you might be on to something. On what do you base your decision to make a trade? You mentioned you keep a pairs percentage, but how do you evaluate that and then make a trade?
The trades are just made on percentage. For every 10% or so it opens a trade. trades close at 50%. I added a basket close for all pairs.
The trading decisions are not very technical. My thoughts are it could be useful as a filter for a more advanced system? It is really just a way to add a type of fundamental trading to an automated system. Looking at just the history of price charts is very limiting in judging future price.
How accurate do i think it will be? Problem is most currency pairs have no clear signal, which is to be expected as there should be no big swings if things in the world are normal and balanced… The closer the returned values get to 50% the less accurate the signal should be.
The pair to watch is really EURGBP. Just because there is a lot of news about them right now. The bot has them at 80% sell. Which is way outside the norm for other pairs. Also has GBPCHF close to the other end of the scale.
I could be interrupting it back to front for all i know? Or it could be complete rubbish… I wouldn’t say it is random because it is based on real data. More likely i cant interrupt it correctly…
Then there is the issue … If it does work, and people know it works, then the results could be manipulated by false news reports, etc.
This is only a 30 day demo. Will see how it performs in that time period.
sorry, i had to stop this test… there was an error in closing the trades. will restart it soon. Just havent got the free serer space for the next month…
Dear Sir,
is it possible that you write some lines, how your external app filter news and the search date and interpret the scraped results ? do you use some kind of keywords ?
Thanks for your help.
Regards
Tim
