Historical Index Constituents

[tech/a]

This is a problem which those of us without the resources of a small country face when confronted with he survivorship question.

It's a problem when using any sub index of stock in a portfolio system.
As already pointed out there are many factors including membership of the sub group which are missed on many stocks particularly over the longer term--- years.

I have pondered over the dilemma often.
My own personal conclusion is the following.
(1) The occurrences of delisting are rare and to be long in a stock at the time of delisting whilst not impossible is very rare.
(2) Portfolio trading commonly has 10 or more open trades in the portfolio at one time.delisting would result in around 10 % or less loss in any one occurrence .
(3) Splits have a similar effect and will show an exit normally as the split looks like a severe drop in price.
(4) Take overs and changes of code show one trade left open until closed by the summing of all open profit at the end of testing or by code which looks for in activity.

I currently can't do anything about survivorship.
From the many systems I have tested and traded I've not had one which trades outside of it's blue print of Montecarlo testing.
The only time I have seen this occur and it happens to them all ---- when market conditions are outside of those used in testing.

 

[holyLemon]

Hi guys,

I was wondering tech/a if you could recommend a good data provider I basically I just need some EOD data spanning back 8 years for the ASX 200. Ideally if it has some sort of indication of split points (along with their ratios, as not all splits are 2:1) that would be really great and also if the constituent list is historically accurate (as you pointed out not too crucial but for backtesting my system it would be good to always have 200 securities to test)

I'd like to download the data and import into a locally-running mysql (or sql server) database for backtesting with.

This is ONLY for past data, I don't need live data.

Quite happy and willing to pay for it.

I'm building a pair-trading type system that will rank the ASX200 daily and select the best and worst for pairs-trading candidates.

Regards
HL

This is a problem which those of us without the resources of a small country face when confronted with he survivorship question.

It's a problem when using any sub index of stock in a portfolio system.
As already pointed out there are many factors including membership of the sub group which are missed on many stocks particularly over the longer term--- years.

I have pondered over the dilemma often.
My own personal conclusion is the following.
(1) The occurrences of delisting are rare and to be long in a stock at the time of delisting whilst not impossible is very rare.
(2) Portfolio trading commonly has 10 or more open trades in the portfolio at one time.delisting would result in around 10 % or less loss in any one occurrence .
(3) Splits have a similar effect and will show an exit normally as the split looks like a severe drop in price.
(4) Take overs and changes of code show one trade left open until closed by the summing of all open profit at the end of testing or by code which looks for in activity.

I currently can't do anything about survivorship.
From the many systems I have tested and traded I've not had one which trades outside of it's blue print of Montecarlo testing.
The only time I have seen this occur and it happens to them all ---- when market conditions are outside of those used in testing.

 

[Punta]

I am new to this, but I am finding eoddata.com to be pretty good. I think they have historical indexes as well.

As with most, the history does not include stocks that no longer trade, so it may be impossible to construct your own historical ASX200 etc

 

[tech/a]

Hi guys,

I was wondering tech/a if you could recommend a good data provider I basically I just need some EOD data spanning back 8 years for the ASX 200. Ideally if it has some sort of indication of split points (along with their ratios, as not all splits are 2:1) that would be really great and also if the constituent list is historically accurate (as you pointed out not too crucial but for backtesting my system it would be good to always have 200 securities to test)

I'd like to download the data and import into a locally-running mysql (or sql server) database for backtesting with.

This is ONLY for past data, I don't need live data.

Quite happy and willing to pay for it.

I'm building a pair-trading type system that will rank the ASX200 daily and select the best and worst for pairs-trading candidates.

Regards
HL

http://www.premiumdata.net/

Call Richard Dale directly he is as up on data as anyone

 

[holyLemon]

Many thanks guys, have contacted them.

HL

http://www.premiumdata.net/

Call Richard Dale directly he is as up on data as anyone

 

[Richard Dale]

Just updating this old thread with new information as it still comes up on various search engines. Our Norgate Data service allows you to backtest on historical index constituents.

 

[beeg20]

I have recently subscribed to Norgate Data with historical index constituents. Previously i was using premium data, backtesting on all ords + delisted stocks. My backtesting results with all ords historical index constituents is vastly different. I am wondering if anyone has compared it and had a similar experience?

 

[Richard Dale]

That's expected - it's a significantly different universe.

Plus, you might also be using a different adjustment method too.

If you're using AmiBroker, you can check the Detailed Trade Log to view which trades each is taking and then work back to the individual stocks to figure out why. You might want to make use of the plottable indicator Charts -> Norgate Data -> Index Constituent too.

Cheers,
Richard.