Use access key #2 to skip to page content.

Evidence that past returns correlate with future returns (written by a fundamental investor)



February 03, 2011 – Comments (0) | RELATED TICKERS: CRY , STL , BLL

"In the short term the market is a voting machine; In the long term it's a weighing machine." 

Because of an interesting article by FSaucy, I decided to look into whether there is a correlation between the average price of an index over a longer term (5 or 10 year average, etc.) and future returns.  The following graph compares the "Current Price/Average Price over the Last 10 Years" (hereafter results over 1 are known as a "Price premium") to the returns over the next 10 years for the Dow Jones from 1928 through the present...

The y-axis shows the next 10 year's annualized returns (higher=greater returns), and the x-axis shows the premium paid to the 10 year average price (ex. 0% means the price is at the 10 year average, 50% means its 1.5 times the 10 year average).  Blue dots represent the results at a point in time and red dots represent the average for that "Price premium" over the entire period.  There seems to be a correlation between these 2 factors as demonstrated below...

This point is more apparent when results are divided into their respective decades...

This shows relatively clearly that although different decades have varied returns because of fundamental reasons, paying a lower "price premium" will tend to yield higher returns compared to investing at more expensive times during the period.  This is shown further by the general trendline...

Results are relatively strong at a "price premium" below 100%, with the average return over the next decade averaging 7.4% (before dividends).  The annual return drops off quickly above this point, with results averaging 3.8% (before dividends) for the next 10 years with significant volitility often experienced.  The final graph puts these results in a historic context...

This shows the Price/10 Year Average in blue (expressed as a percentage, with 100% meaning there is no "price premium"), the 10 year return in red, and the return to date in green (not annualized, with results expressed as a percentage of starting investment with 100% yielding breakeven).  The areas with the highest returns tend to have a moderate "price premium," often below 200% (a 100% premium).  If you exited the market (or reduced exposure) before it hit 200% you would have missed the internet bubble, the 87 crash, and from the looks of it (didn't have data far back enough) you would have missed the bulk of the great crash.  The recent recession didn't register as expensive, partially due to the internet bubble still impacting averages.

I am still experimenting with this and I'm curious about looking into differing time periods (such as a 5 year average, a 15 year average or a 20 year average) and weighing the averages (such as having more recent results having a greater impact).  I'd also be curious how impactful short-term movements are on short-term returns.  These results are what I've accomplished over the last few days (I got a lot done with the snowstorm); these initial results have given me enough reason to keep looking into this as a balancing element for my portfolio, where I'll adjust the amount kept in stocks vs. bonds depending on the ratio. 

Congrats for reading this far, please share any thoughts, ideas, feedback, stuff to try/check out.  Rec this if you feel this is worthwhile information to share or if you feel I might be on to something.  If anyone has tried anything similar in the past, I'd be happy to hear what you've got from it. 

0 Comments – Post Your Own

Featured Broker Partners