Quote Originally Posted by Phaedrus View Post
What do you mean by "short term", Alan? Are you talking about day trading? I would suggest that to attempt this in such a small market as NZ would be an exercise in futility that is doomed to failure.
If you are interested in short-term swing trading (holding for days to perhaps weeks) then daily candlesticks and volume are all you really need, in my opinion. These are readily available (free) and you can go back as far as you like to test your theories. You could add in other indicators from there. I would suggest On Balance Volume, Stochastic oscillator, Momentum and Relative Strength Index for a start.
To be honest, I don't really know.

It was a conversation where someone else was talking about trading systems.

My (general) view is that no 'system' is going to result in ongoing gains (above the market average), since if it did, everyone would jump on it, and the system would no longer work due to market forces.

I am talking here about the chartists etc, I'm not referring to 'value investing' (for example), but I'm not looking to start a holy war about systems!


Let's just say it got a little involved, and so I said, okay it would be easy to see if a system had performed in the past, just replay actual events using whatever you like, and see if it would have yielded a consistent gain above market.

The conversation then turned to how you would do that, and I said, well, you couldn't just look at the close each day, since the system might have told you to sell at a certain price, and perhaps that price was actually achieved (or beaten) intra-day, but the close might be outside the range both the day before and the day after, and the high or low might not have been at sufficient volume for you to fully buy or sell the volume you have or wanted (selling in full being more what I was thinking about).

For example, lets say you used SystemBob and it told you to buy 50,000 TEL at $2.00 which you did. Sometime later, it told you to sell all 50,000 at $2.10, so you set a limit order (?) at that price.

Close on Monday is $2.08. Close on Tuesday, is $2.09 with an intra-day low of $2.06 and a high of $2.11. However, the $2.11 was for 5,000 units. That would mean you could only have been sure to have exited 5,000 at that price, not necessarily all 50,000 (of course you don't know if the volume was limited by the bidder or offeror) so you can't be sure either way.

Hence my wondering if it is possible to access data to see the actual prices and volumes for every trade.

In terms of the data set, even for the largest traded securities, I don't see that it would be too large to analyse.

How many individual trades would TEL do in a day (10,000)? Excel can handle a data set of 1m records (with multiple fields in each), subject to your machine's limitations only of course.

There are plenty of DBs that can be run on the desktop that can analyse much larger data sets to test a system. How large is large enough? BP in London - maybe 1m trades in a day? I woudn't have a clue, but we are still only talking about a few billion data points - its just not that big a problem with modern systems if you can get the raw data, and the size is trivial.

Lets say, 100 bytes of data, per trade, and go with 365m trades of data in a year.

That is 365,000,000 x 100

= 36,500,000,000 Bytes

~ 36,500,000 kB

~ 36,500 MB

~ 36.5 GB


I could be out by a factor of 10 for the most traded securities in the world, but the point is that the data sets are quite feasibly handled, if you can get them.


Anyway, that is the full background, to my OP.



Alan.