I would choose load 10,000 .This will most likely load everything you have. I'm not sure if this is what your asking. If you mean how much is ideal I think 7+ years is good,at a minium I personally would not use less than five years of historical data for testing.

I've tried quite a few,but might favorites are e-signal,and Quocenter. Right now I'm playing around with futures data ,so i'm trying this vendor that supposedly has good historical data,going back many 10's of years.-don't really need that much though. How long would it take? I't depends, anywhere from 3-5 minutes to infinity.With this your internet conection,how your downloading,how many symbols ,are you cutting and pasting,etc.,etc., etc., comes into play. With the above mentioned data vendors and a cable modem you are probably looking at maybe 10 min. I've never used databull so I really can't offer you much input on them.Maybe someone else in the forum might be able to help you with that and compare experiences.I believe Quocenter,and esignal offer a free trial. P.S. I have two computers on one I keep my metastock really clean and on the other I really don't care too much I have lots of historical data and its where I play... creating,testing ideas..formulas and whatever comes to mind.

II As data is what you are going to be analysing, it is ESSENTIAL that you have ALL of the CORRECT data. There are many data providers; some are good, some not so. Some are free and some are quite expensive. If you don't have good, reliable data, then how can you make good decisions? Bad data is worse than no data. I recently swapped over my data providers (in Australia). I thought I was happy before, but am even happier now! I didn't really know there could be a difference, but there is. Do some shopping around, get the service you want for the price you want. Services might include: daily downloads, historical backups, intraday snapshots, split/merger/consolidations, watchlists/custom lists/custom folders, composite data (TRIN/Adv/Dec etc), download at close, download after midnight, separation of stocks, options and warrants, international indicies and currencies, delete expired stocks etc etc etc Remember two rules: "If you don't ask for it, you don't get it" and "You get what you pay for." My data provider gave me a historical data download, then each day I take my incremental daily data (I also get snapshots of the market during the day) This means I dont have to download the entire history every day. I believe just about all good data providers have this service. It takes me about 15 seconds to download the EOD data and about another 45 seconds or so for it to be written to my HD, and I dont have anywhere near 1.5Mbs!!! (Dreaming.... of the day!) -- How much data do you have to load? For explorations and indicators like the EMA that always keep some part of history in their computations it is essential to have "enough" data loaded. Some people have done tests to quantify how much is enough, but the general rule of thumb is about, five times the length of the function. e.g. if you are going to use Mov(C,100,E) load 5x100 bars on your chart and in the exploration. You can load up to 65500 bars of data for each stock. For my charts, I load all the data (my machine can handle it - well at least let me say I haven't had the troubles that some others have had) and show one year on the page. In my explorations, because I do not use EMAs, Wilders etc I can get by with loading minimum data, but I always load 500 bars (approx two years) just to be on the safe side. I don't think there is a hard and fast answer, except to say, experiment a little. Do a couple of explorations using EMAs etc and compare the exploration results with the chart. Load different amounts of data for the chart and the exploration and compare the results again. If you notice any differences, try to figure what is causing the aberration. If you cannot figure it out, just ask here. Hope this helps. wabbit :D