Friday, 22 June 2012

Timing the Market: Is it Possible?

Clearly there are times when the market is undervalued and opportunities to make big returns are possible and times when the market is overvalued and returns will be a lot lower. Is it possible to identify these times and if it is can it be used to increase our investing returns.

If you recall our model is based on dividends. The chart below shows how dividends have grown over time on the FTSE All Share index (I've used this index rather than the FTSE100 index as there is more historical data available). I've also included the FTSE All Share price level. As noted earlier dividend yields have been falling over time and this can be seen as share prices have increased at a faster rate than dividends. Another thing you'll notice is that dividends are a lot less volatile than share prices (I've added a 4 period moving average to the data as it is based on quarterly data). Share prices rise and fall whereas dividends tend to be stable or rising most the time. Also note that the scale I've used is logarithmic:


Going back to the original question in hand. Is it possible to time the market? Using our valuation model:

Total Return = Dividend Yield + (50% IMF Growth Rate) + Adjustment Rate


Adjustment Rate = ((Dividend Yield / 4.5%) ^ (1/20))-1


4.5% = "Fair Value" Dividend Yield (This varies for different indexes and for individual stocks)
20 = Number of years over which the difference is discounted (if you wanted to be more conservative you could use a shorter time frame)

The adjustment rate is only used if it is negative (this increases our margin of safety). So in the above example if the current dividend yield is greater than 4.5% no adjustment rate is needed.

At the end of 2011 the expected return was 4.12% based on a dividend yield of 3.47%. Since 1974 the expected annual return has ranged from 2.30% to over 16%. The chart below shows expected annual returns based on the model, actual annualised 20 year returns and actual annualised 10 year returns:


As you can see our model is fairly conservative in its estimations of future returns. 

There have clearly been some good buying opportunities in the past. In recent years the financial crisis in 2008 resulted in expected returns on the model of almost 9% annually. Investing at the peak of the expected returns at the end of 2008 would have resulted in total returns of 57% by the end of 2011.


Backtesting of trading systems is often abused. Curve fitting to produce generous returns on paper is often misleading. So I hope you'll forgive me here if we used the expected returns to time the market. Let's say we invest when the expected returns are over 6% and sell our investments when returns go below 4.5% (you need a decent spread otherwise you would be whipsawed trading in and out of the market too regularly and missing most of the potential profits). Also below a 4.5% expected return we are likely to be able to get better expected returns elsewhere, for instance in Real Estate, Commodities or Cash.

Between 1974 and the end of 2011 the total annualised return (including dividends) on the FTSE All Share has been 13.6%. Turning an investment of GBP1,000 in 1974 into almost GBP111,000 by the end of 2011.

The chart below shows the FTSE All Share index since 1974. The green areas represent periods where the expected return using our model has been over 6% (entry) with the white areas being times the expected returns went below 4.5% (exit).


The fit is embarrassingly good and looks too good to be true. Entry in 2008 was a little early and exit in 2010 was a little early too but still beat buy and hold. Using the model would have turned the same GBP1,000 into over GBP150,000 (if you assume you could have got 4.5%pa in the 5 and a half years that you were out of the market this figure would rise to over GBP190,000). For a total time of 31.5 years in the market the annualised return was 17.3%.

We are not so much trying to time the market as allocating our capital to the highest expected returns possible. When the returns for the market look low this is usually an indication that the market is overbought and we can get better opportunities elsewhere. Clearly a strategy that buys low and sells high is going to perform well however a lot of missed opportunities where markets remain overvalued by our model and continue to increase in value (eg Gold over the last 10 years) will exist. Finally to reiterate that going forward 17%+ returns are unlikely due to share yields being lower than historically. The model does give us a conservative valuation of the market and can help us to maximise our returns.

No comments:

Post a Comment