[Tradewinds]

Originally Posted by SIUYA »

. . . .
i second mms - there are not many great scanning systems out there. . . . . .
The issue is largely in stocks as potentially there are 10,000 of them around the world. . . .

Let's say you want to scan for something that needed a months worth of data. Where would the data come from to get a months worth of data for 10,000 stocks from different exchanges all over the world? I don't know. Maybe it's easier than I think, but I'm wondering how difficult or costly that would be?

A scan of historical data is basically a query in a database. If a user interface could be developed that could translate simple conditions at different levels into a multi-tiered query, that would probably be unique. For example, have two sections for user input, and allow the user to link a subquery to the main query. I've seen a scan that allows the user to define the main set of data, and then have the filters work on only that main data set, but the main data set is very generic, like "All NYSE" stocks.

Another problem with scans is that a lot of users trying to filter massive amounts of data requires a lot of computing power on the server side. Maybe I'm wrong, but that's what I'm guessing.

I would think that a scan of price data is very different than looking up something like a customers address in a database. There could be hundreds or thousands of price bars depending upon the aggregation period. If a business is looking up a customers payment history in a database, the computer finds a match for a phone number, or an account number, then retrieves the data. With a scan of price data, you could be applying a moving average that first needs to make many, many calculations. I would think it's a different level of computing power. But I don't know.

[SIUYA]

as i am not a developer, I dont know exactly how these would work, but I would imagine that first yes you have to have the data. It is available...Esignal is a data provider that comes to mind.
then you would need to be able to first filter out the instruments into sub sectors first....then ideally you would run the scan over less instruments.... Bloomberg has a sorting process like this.
Or alternatively you might have to run a long and slow scan initially to be able to work out your subsectors then build from there.
You could theoretically do this in excel - but doing it over many many instruments is not what excel is built for.

[Tradewinds]

I'm wondering if it might be better if the user had all the data right on their computer. Then the website filtered the data, but it accessed the data from the users computer. That could be problematic I guess. The user would need to be constantly downloading data, and have enough room to store it. It would be interesting to test how that worked though. That would avoid the problem of having the server on the website do all the filtering work. I don't know if it would be easier to provide downloading of data or filtering of data. I have no idea.

I'm thinking there would need to be a free and a premium service. The people who make money trading would gladly pay a fee that improved their profitability.