The Gateway to Algorithmic and Automated Trading

αlphability

Published in Automated Trader Magazine Issue 07 October 2007

Automated Trader’s new tradability metrics were launched in the Q3 edition of the magazine and have triggered an exceptional response from readers. Andy Webb addresses some of the requests, observations and suggestions that have been made.

alphability

Firstly, a heartfelt thank you to our readers, who have been hugely responsive to the launch of Alphability in Q3's Automated Trader. One wish expressed by numerous readers was that the next Alphability column included some illustrative data on US stocks (particularly those at the top and bottom of each metric's spectrum) together with an illustration of how the metrics related to the stocks' prices. That request and other reader input therefore form the basis of this issue's Alphability.

US stocks and Trend Alphability

Figures 1 and 2 show the top and bottom three Dow stocks ranked on the basis of their daily Trend Alphability1. In Figure 1, General Motors (bars in orange) was the highest-ranked Dow stock over the period in terms of smoothness and strength of intraday trend in relation to total intraday return. Its highest Trend Alphability score of 5.65 (highlighted by the red arrow) was achieved on October 12 and the intraday price chart for that day shown in Figure 2 hopefully makes obvious the reasons for that high score.

figure 1
Copyright Algorithmic Media Ltd 2007

Figure 1

*For an explanation of how Trend Alphability is calculated, see page 56 of Q3's Automated Trader.

figure 2

Copyright CQG Inc 2007

Figure 2

At the other end of the spectrum, one could be forgiven for thinking "Crisis? What crisis?" when looking at Johnson & Johnson in Figure 3, as the late summer's events seem barely to have disturbed its slumber. The lowest-scoring Trend Alphability day for the stock was September 27, which it largely spent trading in a range representing just 0.38 per cent of its closing price. (As a comparison, GM's trading range on October 12 represented 6.73 per cent of its closing price.)

figure 3
Copyright Algorithmic Media Ltd 2007

Figure 3

Stationarity

Another suggestion from readers was that we include statistical analyses of Alphability metrics, such as tests for possible stationarity. A quick glance at Figure 4, which shows the Trend Alphability of GM plotted against its average for the period, gives an inkling of the potential of this approach. While Figure 4 indicates that GM has trending days and non-trending days (hardly a blinding revelation), there does appear to be a tendency for GM's Trend Alphability to revert to/across its mean. There is of course the caveat that Figure 4 is an extremely small sample and unlikely to satisfy a classical test for stationarity. However, it does raise the possibility of using Trend Alphability as a filter for position sizing of intraday trend following models in order to boost returns.

figure 4
Copyright Algorithmic Media Ltd 2007

Figure 4

Reversal versus Trend

One question raised by a number of readers was whether Trend and Reversal Alphability were diametrically opposed? If a stock had a high Trend Alphability score did that automatically imply that it had a low Reversal Alphability score? The short answer, as illustrated by Figure 5 below, is no. Here one can see that GM, which had the highest Trend Alphability score for the period, also has a Reversal Alphability score far in excess of the comparatively trend-less Johnson & Johnson.

While a Reversal Alphability score can benefit from a range-bound day with frequent reversals, similar reversals can also occur on trending days. For example, Figure 6 shows a GM chart for October 2, a day with an above average Trend Alphability score but also with a high Reversal Alphability score of 1.34. This can be explained by the fact that while GM showed two significant intraday trends on that day both contained equally significant (and numerous) reversals2.

figure 5
Copyright Algorithmic Media Ltd 2007

Figure 5

figure 6
Copyright CQG Inc 2007

Figure 6

2 For an explanation of how Reversal Alphability is calculated, see page 58 in Q3's Automated Trader.

Alphability Online

One of the most consistent messages we have received from readers in the wake of the launch of Alphability in the Q3 issue of Automated Trader has been the desire for greater breadth of data. Readers have not only been keen to see coverage of additional markets (ranging from major market equities to the most exotic currency pairs), but they have also bombarded us with suggestions for additional metrics.

The sheer scale of the response made us realise that Alphability was going to be a far larger project than we originally anticipated. This has prompted us to temporarily postpone the launch of Alphability data downloads from www.automatedtrader.net in order to put sufficient bandwidth and hardware in place to support the service properly. This process is now underway and the first phase of Alphability Online should be fully live by mid-January 2008 at the latest.