Ask Colin

Why is there a difference between the Rolling Year Records (52-Week New Highs or Lows) shown in the Australian Financial Review and the scan in Insight Trader?

Why is there a difference between the Rolling Year Records (52-Week New Highs or Lows) shown in the Australian Financial Review and the scan in Insight Trader?

There is a whole raft of reasons.

One thing that makes it difficult is that we do not know exactly what the Australian Financial Review (AFR) do.

The other factor is that there is a couple of inherent weaknesses and a couple of traps for young players in the Insight Trader system scan, that gives a less that ideal result and makes comparison to the undefined AFR data even more difficult.

First, we have to realise that the AFR seem (we don't know exactly) to include all securities. Now, if company BHP has ordinary shares, partly paid shares, company options, preference shares and instalment receipts, and all of them went up or down on a given day, there would be several entries for the same company on the same day. We get around that by only counting the three-letter ASX codes in the AFR list. This is all the ordinary shares. This is OK, except if a company does not have ordinary shares listed, or if they are very illiquid, but has preference shares or instalment receipts listed or are much more liquid. However, there are very few of these, so it is a good practical compromise rule to count only three-letter ASX codes.

This particular one is less of a problem in Insight Trader because the Generate Archive List only selects ordinary shares (with a very small number of exceptions), so should be comparable to counting the three-letter ASX codes in the AFR.

Next, the AFR seem to count companies that have not been in existence for 52 weeks. This means that, at an extreme, a company might be listed one day, go up the next and be listed as a 52 week new high. I also suspect they do the same with companies that change their name, reconstruct their shares or split their shares, where the AFR do not join the old data. So, if Harris Scarfe was relisted tomorrow, with a new name and ASX code, and then traded up the following day, it would appear as a new 52 week high, even though it had been falling for several years before it was suspended. Short of examining every code, we cannot correct for this. The only realistic and practical compromise is to use a consistent method. That is, use the AFR all the time or Insight Trader all the time.

Insight Trader has some problems here too. Insight trader counts back 260 days of data in each file. If a company trades every day, then this will be very close to a year, depending on whether you worry about public holidays (260 = 52 x 5). However, if a company only trades every second day on average, it will actually be looking back two calendar years. There is no way in Insight Trader at present to set a look-back period in calendar time. This reinforces that the only realistic and practical compromise is to use a consistent method. That is, use the AFR all the time or Insight Trader all the time.

Next, there are a few traps for the inexperienced or unthinking user in Insight Trader. You have to have Database Manager up to date, or you could conceivably miss some name changes or reconstructed/split stocks. I download Database Manager each Tuesday. I Generate a new Hilite List of all categories including Delisted and Indeterminate and then run the scans for the previous week.

Another trap is that you have to set No-Trades Limit as 0 (zero), so that the scan only looks at today or the date you specify for the scan. If you set No-Trades Limit at a positive value, Insight Trader will use the last day within that period, even if the stock does not trade today or on the day you specify.

So, you can see that there can be a lot of scope for the AFR data and the Insight Trader data to be slightly or even significantly different.

To illustrate these points, I took the AFR list of Rolling Year Record Highs for 11 February 2003, which had this list of three letter codes:

AIE

CCV

CHR*

CTO

ENV

EQI

GNS

IWF

JAM*

JBM

NAI*

PPB*

SCF*

SFC

ZIM

Total 15

Then I ran an Insight Trader scan and got this list:

AIE

CCV

CTO

ENV

EQI

GNS

IWF

JBM

SFC

ZIM

Total 10

There are five codes in the AFR list that are not in the Insight Trader list, which I have put asterisks against in the AFR list above. There were none in the Insight Trader list that were in the AFR list this time, although this could happen. I then looked at each of the five codes and found:

CHR - Very thin. The AFR is correct on this one. Insight Trader is picking up higher prices in 1999 that are less than 260 days of trading ago.

JAM - Same situation as CHR. So, the AFR is correct on this one

NAI - It changed its name on 5/12.02. The AFR must be ignoring the data in the old name. So, Insight Trader is correct on this one.

PPB - It has not been trading for 52 weeks, having listed 30/10/02. So, Insight Trader is correct on this one.

SCF - It has not been trading for 52 weeks, having listed 13/8/02. So, Insight Trader is correct on this one.

The tally is AFR right = 2, Insight Trader right = 3. Difference = 1

This is basically the kind of result I got when I checked a few months of data before deciding to use Insight Trader data instead of AFR data. The differences tend to cancel out and, so long as we are consistent, it does not matter very much. I have a hunch that when there are a lot of new listings, name changes, splits and reconstructions, Insight Trader tends to be more accurate. It depends on how many illiquid stocks there are trading.

Keyword: