m5more
1 posts
msg #358
- Ignore m5more |
1/28/2003 12:13:53 AM
I just joined and I am looking for a filter that would sort for buying and selling pressure, supply and demand, for minutes, hours days or months.
|
TheRumpledOne
6,411 posts
msg #100014
- Ignore TheRumpledOne |
3/31/2011 1:36:05 PM
You still here?
|
straken
469 posts
msg #100020
- Ignore straken |
3/31/2011 4:54:04 PM
m5more
msg #358
- Ignore m5more 1/28/2003 12:13:53 AM
I just joined and I am looking for a filter that would sort for buying and selling pressure, supply and demand, for minutes, hours days or months.
_______________________________________________________________
So is the rest of the world. I have the closest thing to it. You need streaming real tick data for buy/sell pressure and volume data. You will have to put up a data center server, get RT data from one of the providers including exchange fees, (esignal, real tick, tradestation, etc) build a queryable filter/screener to search your datasource (csv, sql, etc). You need to store the files as daily data. trying to replicate data for monthly and weekly historical data becomes a programmers nightmare. In your screener build box make sure you give yourself the option to search based upon time range and you will be able to plot monthly and weekly without the need to store other than daily files.
I took a little shortcut. I use a (modified) version of wordens stockfinder (great charting package) to search my datacenter. You have to do the (modification) yourself. Dev partners softice and a good disassembler package will help you (modify) their code. I will say no more. GLTY.
|
Eman93
4,750 posts
msg #100026
- Ignore Eman93 |
3/31/2011 8:36:23 PM
Straken,
Are you trying to scan stocks or looking at the over all market?
Large buy side or sell side volume???
|
straken
469 posts
msg #100027
- Ignore straken |
3/31/2011 9:09:51 PM
Yes,, and Yes.
I mostly trade in options, but I'm watching everything. There is no try I do, and even successfully quite often. You can get the same data on a daily basis with stockfinder or esignal but you can get only real tick data that is daily because their servers dont have the retention. If you want historical data to compare with then you have to store the data yourself. No one that I know of retains historical tick data for any length of time and offers filtering of such....only same day. My servers and apps were designed to retain tick data for 120 days with streaming RT tick data. I have no lag time at market open but I do experience up to 90 seconds lag on RT by close of market on heavy volume days. Thats still not bad for autoparsing data streams from all U.S. exchanges including CBOE ISE and PHLX option floors. This summer I'm going to code some upgrades to the option filters by trying to incorporate a realtime Black-Scholes pricing calculator. Right now I have some basic filters seeking option volume, IV/HV, vol above OI on strike +4/-4 and a gamma seeker. If successful I may be able to filter based on option pricing value as well.
|
machismo
115 posts
msg #100028
- Ignore machismo |
3/31/2011 10:58:45 PM
wow straken that is some mind boggling work you have done. your database could be a paid service by itself. what kind of hardware your system has to support all that data processing ?
|
Eman93
4,750 posts
msg #100029
- Ignore Eman93 |
3/31/2011 11:03:14 PM
If you can give me an example of what volume patterns you are looking for, I belive Wealth Lab can get all the 1 min data needed, not quite a tick but close.
I have data on a 1 min chart of the SPY for the past 5 years.
Its all C# programing, Ifs not fiesable for me to scan large numbers of stock in a 1 min time frame it just locks up my equipment in real time, but you can back test stuff.
|
straken
469 posts
msg #100031
- Ignore straken |
3/31/2011 11:48:23 PM
I currently deploy a tandem rackable Windoz Enterprise Datacenter server 2008 with (4) 500 GB hotswap drives for my dataminer. Its a 64 bit NT kernel, very fast, very stable. I use a seedbox holding 4 G-Force 8800 GT graphics cards supporting 9 monitors with spatial systems graphics engines for additional acceleration. Some of which is overkill, but I'm constantly trying to improve the system core and acceleration so having one server dedicated just to the data collection leaves the other free for quicker parsing and application deployment. I've always felt the better and faster the data the greater the edge. Sometimes it has yielded advantages such as 12000% option gains in a few hours, and sometimes it hasn't meant a damn thing. Nonetheless, it has now become a dedication.
|
machismo
115 posts
msg #100035
- Ignore machismo |
4/1/2011 12:46:51 AM
Truly awesome straken, your setup is as cool as the PS3 supercomputer.
|