Where can I find a free and simple stock screener for backtesting?

I’ve looked at plenty of systems over the years but Prorealtime is the only one that’s done it for me for backtesting.  I use ProRealTime, free version (whilst I’m fine-tuning) and registration isn’t a bore, I don’t get rafts of emails from them or their chums. Their back testing feature is superb but the free version doesn’t differentiate between buys and sells so it works best with small spreads.

ProRealTime is free unless you want intraday live data (iro €50 pcm for UK inc platform) and the scripting is easy.  Basic registration free screening, lacking features: Try Stock Screener Lite, which is free but you have to download data at the end of each month, which means you have to manually build your own data archive. Yahoo, London Stock Exchange, perhaps even Google have some free screening but you might have to use Excel or Google Docs as well. ADVFN seeing as you’re already registered, I export their lists to Excel and then run a few functions across it. There are resources on the net to show you how to write Excel code for technica analysis metrics. Then there’s FinViz for US data. There’s plenty of others if you don’t mind registering.

Apart from that, Stockopedia is also nice. The guys that run it are investors themselves and are developing the system from a retail investor perspective.  I setup a screener on there and duplicated it on Sharelock and it pretty much got the same companies coming up.

Filter, 33 stocks, how do you deal with that? …Backtesting. (bar a nose for good governance and a rapid understanding of fundamentals. After that it’s simply familiarity gained through experience, I guess!). IMO, though, backtesting is more important than the actual filter used. Good backtesting will let you sort performers by reliability and when you get one or two that have a 100% track record of giving set gains over a set period xx number of times it’s pretty easy to pick the best performers regardless of any fundamentals or governance.

I can find plenty to buy with technical analysis and I can narrow the field down by back testing the screening criteria to a 100% historical success rate but errors in the system seem to come in batches, i.e. after several weeks of success there’ll suddenly be a whole bunch of failures at the same time, possibly depending on how bullish the economy is.  What other metrics should I include, ie macro-economic or global, or market sentiment other than stock specific crossing MA’s or bouncing RSI &c?

I want to buy or sell a stock when I believe it’s going to go in the direction I want and technical analysis helps me do that.  Once you find a stock interested in buying it helps to actually watch it for a while.

One problem with backtesting is that it can get complicated with fundamentals. For example, what do you do with mergers/takeovers, suspensions and collapses? Then there’s rubbish in, rubbish out….if you scan for p/e instead of rolling price/earnings, you end up comparing companies who were a day away from having up to date data with those who have new data, making the whole thing a bit of a lottery. I have no idea whether there’s an accepted method. Sherlock attempts to do it using previous reporting points but I doubt the data has much accuracy (better than nothing).

Certainly a good point about rolling data. I don’t worry about cutting out companies that don’t qualify for a screener as there’s plenty to choose from and their day will come so if their data is too old or jumbled I’d simply disqualify them. I think there’s too much focus in trying to filter every Co when all that’s needed is one or two with a better measure of certainty than 50/50.