Market Breadth Primer: Component Changes And Their Effects

By Lawrence

iStock_000015526520XSmallStock market indices are calculated from a basket of components. These components, however, are not always the same. Components are changed for many reasons. Contrary to what many conspiracy theorists say, majority of these reasons are legit. I am going to discuss the effects of component changes and how that affects our choice in picking a custom basket to represent the index.

Legit Reasons To Change Components

There are many reasons for which components have to be changed. For example, a merger of two companies may disqualify the resulting company to be included as a component. There is also the case that two component companies merging into one, creating an empty spot for another company to fill. The breakup of an existing component company may have it disqualified. The bankruptcy of a company is another reason for a stock to be removed from an index. I am not sure if it is possible to list all the reasons here at all.

Since companies do change their business directions and the economic environment can affect what businesses do to grow (or survive), these corporation actions happen all the time.

Another source of legit component changes is the design of the index. For indices like Nasdaq 100 that is pretty much a market cap driven index, based on the largest companies listed on the Nasdaq Exchange, there is one good reason to replace its components as frequently as required. Using Nasdaq 100 as an example, whenever a component drops out of the top 100 list, it will be at risk to be removed on the next scheduled components review.

The Not So Legit Reasons

There are times some companies like to rename themselves and change their symbols all together. It is not really a change of component but most financial data services would not include the historical data of the old symbols into the new ones. That gives the new symbols a "fresh start" and mess up the statistics collection on measures like 52-week high as there is no historical data to reference from.

Indices like Dow Jones Industrial Average is pretty much an index created from components with no specific guidelines or reasons. It is a voting based method to decide if certain components have to be removed and which stock has to be added so that the index is "a good representation of the US stock market". Since the method is not scientific, it always puzzles people why certain stocks are added to or removed from the index.

Stock Market Indices Are Extreme Cases Of Survivorship Bias

Given the short introduction above, it is obvious that stock market indices are designed to take advantage of the survivorship bias although it could be just a coincident. Survivorship bias is the phenomenon of looking at historical data on the current members only, ignoring the fact that these current members are not the ones that we start with. It could be the case that the components in the past that are responsible for generating the historical index values no longer exist.

Dow Jones Industrial Average is a great example of this problem:

1. None of its first components are still in the index

2. General Electric (GE) is the only surviving component from the 1910s

3. Don’t even think that General Motors (GM) is doing fine although it is no longer part of the index. The original GM went bankrupt back in 2009.

The new GM shares the same symbol as the old company but it is not the same company. If you still do not understand what that means, think in terms of the ownership. The owners of the old GM shares are wiped out, losing their complete investment in the old company and have no part in the new one.

4. Dow is now at its all time high while GE is still stuck at 30% from its all time high price level

By referring to the historical performance of the index only, one can trash talk how good buying stocks is as a long term investment while majority of the stocks listed back in 1910s are all gone by 1930s.

The flip side of this bias gives us a tool to engage the market. The survivorship bias driven indices in general are great play for picking market bottoms because the component changes switch the surviving companies into the indices making them more likely to do better once the bottom is in place. As an individual, it is difficult to diversify the risk like an index ETF. The index ETF fund managers essentially do the necessary thing for you – cut the losers and buy the strong ones. This strategy works very well when a stock market just comes out of its bad times.

The Basket Size Matters

An index with a small basket of components will be affected by its component change more significantly than an index with a large basket. For example, the last few component changes in Dow have resulted in significant post-change selloff that are very predictable. This is mainly caused by the fact that there are many derivative products built around these indices. As the component change happens products like ETFs, etc. will have to adjust their portfolios to reflect the new changes. This in turn starts a chain reaction across all the related derivatives.

In fact, the sell pressure induced in Dow during those adjustment period also dragged S&P 500 lower due to the connections among the indices.

The change of a few components in S&P 500 or Nasdaq 100, however, would not produce such significant volatility.

This behaviour is important to remember when we design custom market breadth. It is better to use a larger basket than a small one if you do not plan to maintain the component list very often. The reason is that the market breadth generated from the larger basket will still be valid even though you have a few components not matching. For example, by updating the S&P 100 component list once every quarter will still be as good as the one you update once every month for the purpose of collecting custom breadth data.

The Individual Weighting Of The Components Matters

For Nasdaq 100, its component change is almost always at the bottom of the list based on weighting. That means, those new components and the old ones being switched out are not that important until they rise in their rank among the other components. In fact, majority of daily fluctuations in Nasdaq 100 are contributed by the top 30 components.

This weighting concept also applies to Dow but in a slightly different way. It is the absolute price value of a component in Dow that affects its significance within the index. That’s the main reason why GE does not matter in the Dow index any more. Just IBM alone has 6 times the weight of GE. One percent move in IBM is 6 times more powerful than one percent move in GE when Dow index level is concerned.

This property of heavy weighting components makes it easy to construct faster response (even leading response) custom market breadth by using only the top weighting components within the index. This is a more time consuming process as you have to maintain the component lists proactively. From my experience though, it is well worth the effort.

Downloadable Component Lists

I will have the historical component lists of Dow going back to 1991 available for download from the Historical Data Bank. The dates in the filenames are the dates the symbols going into effect. Notice that many components no longer exists. It will be difficult to look for historical data for those components.

This Dow historical component lists archive will be free for all members.

The historical component lists for other indices will be uploaded as soon as they are ready.

 

Resources

Wikipedia.org – Dow Jones Industrial Average

Wikipedia.org – Historical components of the Dow Jones Industrial Average

Wikipedia.org – NASDAQ-100

Comments
  • MidKnight January 3, 2014 at 9:08 pm

    What you say above about a few stocks being the main mover for the index….I think that is a big reason why HSI is so frenetic.

    Over 46% of the weight is in financials. 14.52% alone is assigned to HSBC holdings. 6 stocks make up 45.93% of the weight. I wonder, would you bother including the entire 50 stocks for any custom breadth when you have such heavy weighting discrepancies amongst constituents? I feel the lower weighted stocks would just create more noise. What do you think?

    With kind regards,
    MK

    • Lawrence Chan January 3, 2014 at 10:05 pm

      My usual cutoff preference is 65% weighting.

      For NDX the top 30 stocks is already 70%+ of the total weighting.

      The heavy weights lead, but the remaining ones gives you the extremes, just like what I said in STOPD how tops are formed. =)

  • MidKnight January 5, 2014 at 10:47 pm

    Thanks Lawrence. It’s been a loooooong time since my math days and I can’t figure out how to adjust the weights of a custom basket to total 100%. For example, lets say you have a stock that is an 8% weighting in the full index of 100 stocks. You put this stock into your custom index of 30 stocks – how do you adjust its new weighting? I wouldn’t even know how to explain this succinctly for a google search…..

    With curiosity,
    MK

  • MidKnight January 6, 2014 at 12:40 am

    Actually, just thinking about it some more. Probably just sum up the market cap for the index and then determine each stocks market cap percentage ?

    • Lawrence Chan January 6, 2014 at 12:00 pm

      you can do that but the custom index you get will look similar to the one you are trying to estimate but not quite the same.

      to get a better estimate you need to solve the system of linear equations based on only the custom basket of symbols for the weightings. it is much easier to do that with an automated script.

  • mel January 12, 2014 at 6:57 am

    LC any way to show comparisons? weightings kept changing right now I just use equal weightage for components stocks to gather T16 and breath…. for N225, I just basically use the highest weightage of the 20 stocks out of the 225 and use equal weightage to gather the same…. currently observing the behavior…

    • Lawrence Chan January 12, 2014 at 1:00 pm

      If you are talking about what was discussed above, it was not on breadth data collection.
      We were talking about creating your own custom index.

  • mel January 12, 2014 at 8:22 pm

    Yes LC creating the custom index

  • You must be logged in to comment. Log in