Tuesday, January 09, 2007

Trade Ideas: Time factor

Instead of a list of stocks for today I am instead going to show you the previous years worth of data I have been collecting on the Trade Ideas scan real-time stock picker- using my "Base Scan" similar to the one available in the Trade Ideas widget in my right-hand margin. When I am looking at the market I am not just interested in the stocks which the scan suggests, but also the rate at which these stocks appear (an indication of the level of buying in the market). I use a sample size of 8 different stocks - for no other reason than that is this was the size I could fit 8 charts on my monitor without have to scroll :), but any number (preferably over 5 stocks) should suffice.

For the record, to get 8 different picks for Tuesday would have required 10 minutes.

I also took the closing value for the Nasdaq on those dates readings were taken and plotted that alongside the time required for the scan took to select 8 different stocks. When I talk of a top or bottom in the Nasdaq I am referring to this data and not the Nasdaq as a whole.

The first chart is the raw data and it is clear it is a little ragged. You will notice there are three spike days when it took the software the entire day to come up with 8 different stocks. When the market was moving upwards from July 14th to November 22nd, the scan bumped along to the extent that 72% of the 8 picks were selected in 10 minutes or under, with only 5% of those 8 picks requiring an hour or more (n=57 days). The average time to pick 8 stocks for this period was 11 minutes. Looking at the rate of buying from the market high on April 19th to the low of July 14th, the percentage of stocks selected in 10 minutes or under fell to 55% and 8 picks requiring an hour or more rose to 14%. The average time for this period was 36 minutes (n=29).

The second chart shows a 10-day MA for both the market and the time taken by the Trade Ideas scan to pick 8 stocks. Whether there is anything to this or not, but it was interesting to see a significant slow down in the rate of buying during February and March 2006 before the market peaked in April. This was followed by a another slow down as the market declined during the late spring and early summer (no surprise there). The next slow down occurred in December 2006 - which may be nothing more than a reflection of holiday trading, but there is another more worrying spike taking shape in January 2007 which could spell trouble.

One other interesting aspect to the data was the period of June 14th to July 14th 2006. The Nasdaq continued to decline as traders fished for a bottom, but the Trade Ideas scan had started to show signs of life; for this period, 70% of the 8 stocks were selected in 10 minutes or under, and at no time did the 8 stocks require more than an hour (the slowest day was 40 minutes). The average time for this period was 12.2 minutes (n=10). So the Trade Ideas scan may give an inkling of a bottom based on the rate of individual stock buying in real-time, before the market actually completes its bottom - although the sample size in this case was small.

But hindsight is easy. How can we use this data in real time? I then looked at the performance of the scan between each of the all-day spikes of non-buying on the chart. The advantage of this is analysis is one can start or finish the analysis based on the day the an all-day spike emerges, which in itself gives some inkling as to a shift from accumulation to distribution. Unfortunately, spikes don't emerge during bottoms to the same degree as (given the nature of the bottom) the market shifts to buying over selling.

So - when did these spikes emerge?

March 6th 2006
May 17th 2006
November 27th 2006

For the period of the March to the May spike (excluding the spike day) the Nasdaq traded flat: moving from 2,294 to 2,310. The percentage of stocks picked in 10 minutes or under was 58% and those requiring an hour or more came in at 15%. The average time to select 8 stocks for this period was 19.6 minutes (n=26).

For the period of the May to November spike the Nasdaq traded from 2,180 to 2,466 - but not before it took a dive to 2,037 in the process. The percentage of 8 stocks picked in 10 minutes or under was 67% and the percentage requiring an hour or more was 8%. The average time to pick 8 stocks was 15.9 minutes (n=76).

As we have yet to see a spike to contain the data from November I shall present the data to date: The Nasdaq has traded flat for the current period: moving from 2,413 in November to 2,443 today. The percentage of 8 stocks picked in 10 minutes or under was 57% and those needing an hour or more came in at 19%. The average time for this period is 23.3 minutes (n=21).

Based on the most recent data it would appear the next spike (which occurred in the previous 3 instances on a 2.4%, 1.7% and a 0.7% loss in the Nasdaq) could foreshadow the timing of next major correction (after the May spike the Nasdaq shed another 8%). When the market starts looking receptive to some buying will show up with some steady buying in the Trade Ideas scan as the Nasdaq works itself through its bottom. This buying will maintain itself up to the next spike which will mark the start of the next phase of distribution.

Based on the data from November 22nd, we are in a distribution phase. Although we have yet to see a binding spike to 'seal' the data. The other (big) caveat is this is all derived from two data blocks - whether this reflects a true pattern will be dependent on future data.

Stay tuned to my blog - I will publish the day I see one of these spikes occurs and take it from there. The best way to do this is to subscribe to my feed.