Historical data and technical analysis
- Thread starter Trex123
- Start date
tech/a
No Ordinary Duck
- Joined
- 14 October 2004
- Posts
- 20,403
- Reactions
- 6,328
For Stock or Futures?
You want to test ideas or simply look at charts and trade in a discretionary manner?
You want to test ideas or simply look at charts and trade in a discretionary manner?
There is no right answer, depends on what time frame you want to trade/invest.
I have noticed that people who look at a HUGE data set statistically, like for example, 50-100 years, are very good at recognising increased probability of crashes or rallies.
Probably rubbish with intra week though!
tech/a
No Ordinary Duck
- Joined
- 14 October 2004
- Posts
- 20,403
- Reactions
- 6,328
10 yrs minimum
20 yrs would be better.
Yahoo and other cheap suppliers will have all sorts of glitches in the data.
Premium Data if you want clean data for testing.
Why 20 yrs/
To get as many market conditions as possible in your data.
20 yrs would be better.
Yahoo and other cheap suppliers will have all sorts of glitches in the data.
Premium Data if you want clean data for testing.
Why 20 yrs/
To get as many market conditions as possible in your data.
WaveSurfer
Manic
- Joined
- 28 March 2009
- Posts
- 219
- Reactions
- 0
Use the DownLoader application that comes with metastock Trex.
Tools > Convert will convert to metastock format for you.
Tools > Convert will convert to metastock format for you.
- Joined
- 29 December 2010
- Posts
- 228
- Reactions
- 0
I have a whole HDD dedicated to historical prices. I think it totals about 300 GB in data alone. 12 years of 1 minute data. 4 years of intersecond tick data for all the major, and not so major currencies.
From there systems get coded to hard and fast rules and get run over the 1 minute data. If they look good they get tested over intersecond tick data which mimics real time trading as closely as possible.
The ability to have such data is vital if you plan to understand the risk your systems are taking, and if you have created a system which works in just the current market, or works in many different types of markets. ie does it go as well in latter 2010 as it did in latter 2007?
What effects does changing variable values make? What are the best values?
One thing which is important is to have a large sample size and a basic understanding of statistics to avoid curve fitting your system.
From there systems get coded to hard and fast rules and get run over the 1 minute data. If they look good they get tested over intersecond tick data which mimics real time trading as closely as possible.
The ability to have such data is vital if you plan to understand the risk your systems are taking, and if you have created a system which works in just the current market, or works in many different types of markets. ie does it go as well in latter 2010 as it did in latter 2007?
What effects does changing variable values make? What are the best values?
One thing which is important is to have a large sample size and a basic understanding of statistics to avoid curve fitting your system.
