Albert Einstein’s famous definition for insanity — “doing the same thing over and over again and expecting different results” — is well adapted to the field of technical analysis and market timing. The continued application of good indicators to flawed data sets, while expecting different results, is as pointless in technical analysis as it is in physics.
Although there is an element of “art” in technical analysis and market timing commensurate with the skills and experience of the practitioner, the analytical ideal always should be to create objective results based on information derived from reliable data and good indicators.
With 20 Most Actives and New York Stock Exchange advance-decline (A-D) data inserted into comparative lines, we will show that the Most Actives series (MAAD) has reflected the “real” stock market more accurately, as expressed by big money, than the more widely used NYSE A-D series (NYAD) that employs inherently defective data. We will reveal how NYSE A-D data are statistically less reliable, despite modifications made in an attempt to eliminate deficiencies.
By applying this Most Actives methodology, market practitioners should find better opportunities in the stock market by identifying when and where informed money is focused, while using that information to realize greater profitability.
The A-D line has been a staple of stock market analysis since the 1930s (see “MAAD and NYAD,” below). The A-D line is created by subtracting declining issues from advancing issues and then by adding the net difference to a cumulative A-D line. When the majority of stocks is advancing, the line will rise. When the majority is declining, the line will fall. It is a popular indicator. Whether it is computed daily or weekly, A-D line data are reported by the major stock exchanges and then are published by the financial media.
One of the primary uses of the A-D line is as a divergence indicator. A negative divergence occurs when the market is approaching what turns out to be a major top and the A-D line consistently fails to confirm prices by moving higher also. If the market makes a series of lower lows and the A-D line refuses to make new lows also, A-D line action would indicate a positive divergence and a major market bottom.
Ideally, the A-D line should lead, or at least coincide with, key reversal points in the market. But theory and application have not always lived up to expectations. While the underlying premise is that analysis of A-D line trending is useful information, the problem with implementation has been that conventional A-D data are seriously flawed. Norm Fosback wrote in “Stock Market Logic” that the “A-D line is a theoretical nightmare.”
First, A-D data have been criticized because they include bond funds, preferred stocks, closed-end shares and other issues that are not pure equities. But if a scalpel is used to excise a particular issue, where should those cuts end? After all, the volume of all issues may better represent investor preferences. To expunge an issue and its volume is a lot like saying this vote or that vote in an election is not worth counting. At that point, data exclusion becomes increasingly problematic as new asset classes, such as exchange-traded funds, are added.
Second, exchange data, such as NYSE A-D data, are used to create an unweighted A-D line, like NYAD, which is then compared to a weighted price index like the S&P 500. On a particular trading day, IBM advances one point on 10 million shares, while a small-cap, low-volume stock, XYZ, declines one point on 500,000 shares. Using NYSE A-D data, the decline in XYZ cancels the advance in IBM.
This weighting problem was evident during the 2000-02 bear market when large-cap tech stocks, which constituted big portions of the major indexes such as the S&P 500 and Nasdaq, declined sharply. Because stocks other than tech issues reflected more NYSE advances than declines, the NYSE A-D line rose. MAAD remained in sync with the major indexes, reflecting big money bearishness and declined (see “MAAD and the S&P 500,” below).
Third, as the number of listed stocks has expanded steadily over the years, the comparability and consequent reliability of the NYSE A-D line has been compromised by its progressive, upward bias. Today, there are nearly 3,200 NYSE issues traded while in 1961 there were 1,200. As a consequence, the upward drift in NYAD may be the result of an increase in the number of shares listed rather than long-term market optimism.
The problem with the A-D line is not that it’s a bad theory or a poor design. In fact, the concept of an indicator that accurately reflects a majority market vote is an analytical plus. But as long as the A-D line continues to rely on flawed data, it will remain problematic. That fault was especially noticeable when NYAD predicted a premature stock market top in April 1998 nearly two years before the eventual bull market high in early 2000. Again, it failed by predicting a bottom almost two years prior to the bear market lows of early 2002 (see “Inconsistent performer,” below).
Enter the 20 Most Actives. Reported daily and weekly, the Most Actives are the top-tier issues that big money chooses to buy and sell. Large investors necessarily gravitate to high-volume, large-cap securities because in those issues they receive the best trade executions. Conversely, big money cannot effectively trade small-cap issues because of a lack of liquidity.
Unfortunately, data collection for the Most Actives can be challenging to the extent that there is no readily available historical database, and retrieval via an electronic source, as opposed to manual data collection, can create corroboration issues. For example, we may not know if the Most Actives data we have collected electronically is correct if we have no reference point, such as the weekly Barron’s Most Actives list.
In addition, to create a historically reliable Most Actives list, the series must include all issues previously traded, thus making the data free of survivor bias. If an issue is no longer exchange-listed, some electronic databases do not retain that issue and its volume, and the creation of MAAD becomes difficult if not impossible.
In “Most Actives Snapshot” (below), the first table includes the 20 issues reported by Barron’s for the week ending Dec. 31, 1964. Only three of the original 20 stocks are still exchange-listed. In the second table, while only eight of 20 issues for the week ending Jan. 7, 1977 are no longer listed, MAAD computations without those eight issues would have been inaccurate. The third table reports the NYSE 20 Most Actives for the week ending Dec. 9, 2011, but without Nasdaq.
Because of the data collection and corroboration dilemma, the most practical and accurate method for the retrieval of Most Actives data is the manual extraction from archived issues of Barron’s print and microfilm. This is the method that was used for this article.
Building the line
When Barron’s began publishing the NYSE 20 Most Actives list on a weekly basis, the magazine editors chose the top 20 issues and, for the sake of continuity, that number has remained constant over the years. In contrast, the number of daily NYSE Most Actives reported in The Wall Street Journal has fluctuated between 10 and 30 issues.
Barron’s now offers a wealth of new Most Actives information online, but with that new data comes additional challenges. The Barron’s online list now includes 100 issues from the NYSE, Nasdaq and Amex by share volume, including exchange-traded funds (ETFs), plus another 100 issues from each exchange based on dollar volume. Share volume remains the criteria of choice because that was the methodology originally chosen by Barron’s. Although dollar volume data might provide additional useful information, those data have been available only online from Barron’s for the past several years and there is currently no historical database (right).
In January 1993, the Most Actives landscape was changed significantly by the introduction of ETFs. An ETF is an investment fund backed by assets such as stocks, bonds or commodities that tracks the underlying assets so closely that the ETF can serve as an investment analog to the assets. There now are more than 1,000 ETFs listed and traded on the major stock exchanges (right).
With the proliferation of these new issues, data collection problems were exacerbated. First, should these asset-based hybrids be included in the Most Actives statistics? Some suggested they were anomalies. Others argued that a dollar spent in the market means volume is generated and that volume shows investor intent. Ultimately, ETFs were included in this study because it is likely that segment will continue to constitute a greater share of market trading volume.
However, ETFs introduce another challenge because there are inverse ETFs that move counter to the market because the fund’s portfolio is composed of short positions. For example, when the stock market declines, an inverse ETF advances in price; when the underlying instrument rallies, the ETF declines in price. To correct for this issue, a rising inverse ETF must be counted as a declining issue, while a declining inverse ETF must be recorded as an advancing issue.
Other adjustments included the exclusion of volatility-based ETF issues, such as those derived from the VIX, because they are a step removed from actual stock prices and are not asset-based.
There are other considerations that must be taken with respect to the data. Up until the introduction of ETFs in 1993, data collection for the weekly Most Actives was relatively easy. An analyst could buy a copy of Barron’s on Saturday morning, count the Most Actives issues up and down, and then plot the net difference in an A-D line. That is no longer the case.
According to Dow Jones, about three years ago the Barron’s editorial team decided to include NYSE Arca stocks in its NYSE lists instead of breaking them out into a separate category. This accounts for the difference between Barron’s weekly and The Wall Street Journal’s daily Most Actives lists in terms of the categories included.
As a consequence of the data reporting variances, we collect the Most Actives data in the following manner:
- Weekly data are extracted from Barron’s Online and are placed in a composite file of 20 Most Active issues culled from the NYSE and Nasdaq exchanges, including ETFs and inverse ETFs that must be adjusted if they appear in the Top 20. Amex’s Most Actives exchange volume tends to be low and is not a consideration.
- Daily data, although not within the scope of this article, are collected online using the NYSE Most Actives composite, including ETFs, as provided by The Wall Street Journal.
The above steps give us our MAAD line. While a cursory glance suggests that this is a more viable indicator than NYAD, deeper analysis is required to pass judgment. In the second part of this series, we will break down the historical performance of this indicator and develop a systematic trading strategy based on it.
Robert McCurtain is a technical analyst/market timer, private investor and financial markets consultant based in New York City. He is a member of the Market Technicians Association and can be reached at firstname.lastname@example.org.