Tuesday, January 10, 2012

Some "Odds" to Consider - Recent Medium Term Growth

The takeaways from this article are:
1. It is likely a good buying opportunity based on an expectation (which may be disappointed) that the future will rhyme with the past for the All Ords
2. The time (selected with the benefit of hindsight) to consider selling the All Ords in the past has been when the sum of the 1 and 2 year performance of the All Ords has been positive, generally when it has exceeded at least 40%, not when it has been lower than negative 25% as it is now, other than for a total of about 8 months out of the last 25 years.

We are often being told that the stock market is up or down x% this year, and we know that performance fluctuates, sometimes dramatically.

We have probably also heard of the concept of mean reversion. When something grows dramatically for a long time, it will probably grow slowly or fall for a long time after that so that long term growth reverts to the previous long term average growth, or overshoots to be below the long term average.

We know that market peaks always occur when growth has been strong, and bottoms always occur when the market has had a substantial fall.

So, unless we have had such a bubble that we must have an incredible bust to get back to normality, (ie we are "turning Japanese") then we could look at medium term performance of the All Ordinaries as an indicator of the odds that it will change direction. If it is at a very high level of 1 and 2 year performance, or 1, 2 and 3 year performance, then we would expect that the odds of a downturn are increasing, while if the performance over 1, 2 and 3 years has been woeful, we might, if we think that the future at least rhymes with the past in this regard, think that the chances of some good performance must be increasing.

Remember that some like Steve Keen have pointed out that Australia has had a huge increase in private debt that has funded our consumption and investment boom, that the Reserve Bank charts show credit growth way below recent averages and household saving way above recent averages and that recent house price falls will be constraining credit availability and appetite for many.

We also know that there are recession calls generally accepted for much of Europe in H1 2012 and some credible, but not generally accepted, calls for a US recession in H1 2012, China has had a continuously falling stockmarket for many months and has had unsustainable property price appreciation and infrastructure investment since the bottom in 2009 when massive stimulus was introduced.

Having said that, lets look at past percentile performance over medium terms to see whether the current performance is unlikely to deteriorate based on past performance.

To do this I have looked at adding together a number of performance measures from the 2 month, 1 year, 2 year and 3 year horizons. I didn't go longer because the average length of cyclical bear markets since 1984 has always been less than 15 months and as you go longer than 3 years you are sometimes covering multiple cycles. The median length of time between peaks has been just over 3 years.

The graph above shows the percentile outcomes of adding together 2 Month, 1 Year and 2 Year performance. About 50% of the time the result is better than 40 eg 2M = 5, 1Y =15 and 2Y =20. The result today is -31.7%. Such a bad result has happened only about 6% of the time since 1984. Similarly I looked at the totals for 1 and 2 year performance and for 1, 2 and 3 year performance. All are in the bottom 20% of outcomes. So one could suggest that if the future rhymes with the past there is say an 80 to 90% chance that the market will continue to stage a possibly somewhat volatile recovery from the levels of 26 September 2011.

We also: 
1.  know there is a small history that indicates that performance can deteriorate significantly - just look at the -130% total in percentile 1 of the chart above
2.  recognise when we think about it that if the All Ords stayed at its current level for 2 years that the total of 1 Yr + 2 Yr performance would increase to be 0 at the end of those 2 years
3.  recognise when we think about it, that the averages over fixed length periods moving forward are as much about the old scores that are dropping out of the average as they are about future performance.

Let's look at the history of this measure since 1 July 1987 for the Sum of 1Yr + 2 Yr Growth .

Virtually all of the worse than current scores occurred during either of 2 periods of extreme recession or the exit from extreme recession:
1. 4 months in 1990/91, or
2. September 08 to September 09

For a part of each of those periods  the All Ords had started to recover from its lows and the sum of the averages was in catchup mode as higher levels at the start of the period covered fell out of the averages.

One amazing  observation is that at the immediate bottom of the September 1987 crash, the All Ords still had a positive 1 and 2 year performance sum, such had been the incredible "blow off" top in 1987. It might be thought that, given we have had a long and dramatic period of underperformance in 2008/9, the excesses of 2006/7 have been exorcised.

This article is presented as information about past performance and for general financial education and is not advice or a recommendation to take any action.

No comments:

Post a Comment