Data - Where from and how much?

[tech/a]

Re: Data

Nizar.

Montecarlo testing returned 20-28% from the 20000 portfolio's tested.
which was un leveraged.
Actual trading has returned to the higher end of the scale.
Traded margin the resulst have of course ben amplified by the leverage.
The system was always designed to be traded on around 2:1 leverage. CFD's were not around at the time. However you wouldnt trade the method above 3:1 leverage as drawdown becomes un acceptable---well to me it does.

So return was similar to that expected from testing.

 

[nizar]

Re: Data

Nizar.

Montecarlo testing returned 20-28% from the 20000 portfolio's tested.
which was un leveraged.
Actual trading has returned to the higher end of the scale.
Traded margin the resulst have of course ben amplified by the leverage.
The system was always designed to be traded on around 2:1 leverage. CFD's were not around at the time. However you wouldnt trade the method above 3:1 leverage as drawdown becomes un acceptable---well to me it does.

So return was similar to that expected from testing.

Really?

Im surpised considering the conditions of testing wouldve probably been (far) less bullish than the 2003-2007 period.

 

[tech/a]

Re: Data

Was from 2001 back to 1993 ish.
Wasnt that bad. Bit of tech boom and bust.

Hey I dont argue with the results.

 

[Richard Dale]

Hi everybody,

I'm from Norgate (Premium Data) so I thought I'd add a few points to this conversation.

Yahoo historical data is accurate to 1c. Unfortunately (for Yahoo) the ASX trades in 0.1 and 0.5c increments up to $2.00. So for any security in this range (out of 28077 securities taded today, 63% of them were below $2), they will be inaccurate.

1992 is a pivotal year in my opinion for one reason - July 1992 is when compulsory employer superannuation was introduced. In one swift stroke, the Australian Government caused securities on the ASX to enjoy forced buying demand changing the character of the market (at the large and mid-cap levels) forever. Backtesting before this time period starts to lose significance unless your candidates are mid cap-low cap-spec stocks.

Let me know if you have any further questions about our data or just data in general. I live, eat and breathe this stuff!

Cheers,
Richard.

 

[tech/a]

Thanks Richard.
Great points.
I had a look at Premium Data site today.
Very impressed.

http://www.premiumdata.net/

Do you do realtime?

 

[nizar]

Hi everybody,

I'm from Norgate (Premium Data) so I thought I'd add a few points to this conversation.

Yahoo historical data is accurate to 1c. Unfortunately (for Yahoo) the ASX trades in 0.1 and 0.5c increments up to $2.00. So for any security in this range (out of 28077 securities taded today, 63% of them were below $2), they will be inaccurate.

1992 is a pivotal year in my opinion for one reason - July 1992 is when compulsory employer superannuation was introduced. In one swift stroke, the Australian Government caused securities on the ASX to enjoy forced buying demand changing the character of the market (at the large and mid-cap levels) forever. Backtesting before this time period starts to lose significance unless your candidates are mid cap-low cap-spec stocks.

Let me know if you have any further questions about our data or just data in general. I live, eat and breathe this stuff!

Cheers,
Richard.

Thanks for sharing that Richard.
You bring up a good point about pre-1992 backtesting.

 

[nizar]

Thanks Richard.
Great points.
I had a look at Premium Data site today.
Very impressed.

http://www.premiumdata.net/

Do you do realtime?

Tech,
Did you try developing techtrader for some equivalent for US markets or other international markets and if not then why not?

 

[tech/a]

Its used by a few in lots of markets in its present form.

Of course there were changes to the universe of stocks traded.
I havent personally traded anything other than AUST.

Interestingly the best result was from the Hong Kong Bourse.
USA was also very good.
Choice of universe becomes an important factor when trading outside the Margin lists (As the lender does some fundamental analysis to allow it in the list ).

 

[Richard Dale]

Hi tech/a,

No we don't do realtime nor are we likely to do so. The only way to make money in that space is to have a lot of clients to support the massive infrastructure requirements (servers with oodles of capacity, redundant servers with same, high speed dedicated links to the ASX etc.). We've seen other players enter and exit that market pretty switfly so are happy to keep our niche as end-of-day (and intraday hourly snapshots for ASX).

Cheers,
Richard.

 

[nizar]

Hi everybody,

I'm from Norgate (Premium Data) so I thought I'd add a few points to this conversation.

Yahoo historical data is accurate to 1c. Unfortunately (for Yahoo) the ASX trades in 0.1 and 0.5c increments up to $2.00. So for any security in this range (out of 28077 securities taded today, 63% of them were below $2), they will be inaccurate.

1992 is a pivotal year in my opinion for one reason - July 1992 is when compulsory employer superannuation was introduced. In one swift stroke, the Australian Government caused securities on the ASX to enjoy forced buying demand changing the character of the market (at the large and mid-cap levels) forever. Backtesting before this time period starts to lose significance unless your candidates are mid cap-low cap-spec stocks.

Let me know if you have any further questions about our data or just data in general. I live, eat and breathe this stuff!

Cheers,
Richard.

Richard.
You say 1992 changed the dynamics of the ASX forever and so backtesting beyond that is not of much significance.

Is there anything like that in the US markets? One year that changed everything? Or, when designing a system to trade the US markets, it is practical to do longer term backtesting?

Id be keen to hear your (and others') thoughts.
THanks

 

[Richard Dale]

Richard.
You say 1992 changed the dynamics of the ASX forever and so backtesting beyond that is not of much significance.

Is there anything like that in the US markets? One year that changed everything? Or, when designing a system to trade the US markets, it is practical to do longer term backtesting?

Id be keen to hear your (and others') thoughts.
THanks

There's several events that have changed certain parts of the US markets but none with an individual impact like compulsory superannuation.

Here's some of the things that have driven up volumes/changed spreads/allowed smaller traders into the sytem over the years:

1966 - Automation of floor data systems
1976 - NYSE Automation of small orders and odd lot trades (<100 shares)
1978 - Intermarket Trading System (ITS) established alowing brokers to find best prices across multiple exchanges nationwide.
1992/1993 - SEC approves Exchange Traded Funds and they begin to trade in 1993
1997 - Trading in Sixteenths (was Eighths)
2001 - Introduction of decimal places instead of trading in sixteenths

One thing about backtesting - you should always sample delisted securities otherwise you introduce a significant survivorship bias.

I'm in the process of compiling history for US markets back to 1950 for listed and delisted securities - a big job!

Cheers,
Richard.

 

[caribean]

Has any of you tried MLD downloader?
i've been using it with metastock and i'm reasonably happy with it, one off fee of approximately $70 for all stocks & indicies & forex around the world.
For historical and EOD
The data is from yahoo & other sources, & even though it's not very accurate
it's good enough for me.

 

[Pager]

For Futures, CSI is the best, heres a link.

http://www.csidata.com/

They also cover US and UK stocks.

Not the cheapest though, i subscribe to there budget market mix, 15 Futures markets and 50 stocks.

The budget mix is the cheapest option and if you pre pay for 12 months its US$335 plus a one off payment of US$50 for the licence fee, it includes a 10 year data base.

Cheers

Pager

 

[nizar]

There's several events that have changed certain parts of the US markets but none with an individual impact like compulsory superannuation.

Here's some of the things that have driven up volumes/changed spreads/allowed smaller traders into the sytem over the years:

1966 - Automation of floor data systems
1976 - NYSE Automation of small orders and odd lot trades (<100 shares)
1978 - Intermarket Trading System (ITS) established alowing brokers to find best prices across multiple exchanges nationwide.
1992/1993 - SEC approves Exchange Traded Funds and they begin to trade in 1993
1997 - Trading in Sixteenths (was Eighths)
2001 - Introduction of decimal places instead of trading in sixteenths

One thing about backtesting - you should always sample delisted securities otherwise you introduce a significant survivorship bias.

I'm in the process of compiling history for US markets back to 1950 for listed and delisted securities - a big job!

Cheers,
Richard.

Thanks for your response.
How far back does Norgate currently offer for historical US markets data?
I agree very important to test and include delisted stocks - i assume your data already includes this (as per what the website says).

 

[tech/a]

Nizar.

Richards data is damned good.

If I didnt have the deal I have with Just Data I'd be with Richard.
Its serious data for serious traders.

 

[kaveman]

I have Justdata, got it for about $460 5 years ago, lifetime supply. However I find it very slow to update the databases.
Premiumdata is very fast for the daily updates. Add to this the update facility for the sectors etc and premiumdata come out in front. I can update a week of data in same time bodhi takes to do one day of data.

 

[nizar]

I have Justdata, got it for about $460 5 years ago, lifetime supply.

What a great, great deal

Thanks for your feedback kaveman.

 

[chops_a_must]

Just wanted to know where every1 gets their data and how much do you pay?

Obviously im after quality data that would include spilts, adjustments and delisted stocks (for historical data for backtesting purposes).

Any1 use just data or premium data?
Any feedback about those guys?

Is their data compatible with an amibroker/tradesim set-up ?

Thanks

I'm looking at a similar course of action over the next few months Nizar.

EOD data from premium data looks perfect for what I would be using it for.

Now I need to resurrect an old thread for metastock/ amibroker q's. Lol!

 

[bvbfan]

Richard.
You say 1992 changed the dynamics of the ASX forever and so backtesting beyond that is not of much significance.

Is there anything like that in the US markets? One year that changed everything? Or, when designing a system to trade the US markets, it is practical to do longer term backtesting?

Id be keen to hear your (and others') thoughts.
THanks

March 1988, Reagan signed off on the Working Group on Financial Markets (aka the Plunge Protection Team)

 

[Strw23]

I use DataHQ. Reasonably priced, all EOD, and Adjusted. Cover ASX, US, Singapore, FX and World Indicies. Cost $77.00 for history data and that covers every market, and then a monthly subscription fee. About $30.00 for ASX and $44.00 for US. Unsure on others.

Scott