Not too shabby. During 2014 the forecasts from my macroeconomic stock market models matched reality pretty well. Overall, the stock market performed a bit better during the year than the model had expected. No complaints there.
During 2014 actual market performance could be checked for 12 monthly forecasts. The first of these forecasts was made at the start of August, 2013 with a forecast for the end of January, 2014. The last forecast that can be checked was made in July, 2014 with a forecast for stock market performance through the end of December, 2014. A graph of the forecasts vs actual results appears below.
In statistical terms the comparison of actual and forecasted results showed an R-squared measure of 0.35. That corresponds closely with the long term back-tested results for the model.
But, that doesn't mean that the model was "right" only 35% of the time.
In more descriptive terms, for most of the 6 month spans of the year the model expected the market to rise moderately and that is generally what the market actually did. Four times during the year the model forecasted that the market would either be flat or fall less than 5%. One of these forecasts was spot on, but for the other three cases the market rose under 5%.
To my mind what matters is that the model did not make any really bad calls and mainly the stock market performed roughly as expected. I'll call that a win.
Happy New Year!
(Click on image to enlarge.)
No comments:
Post a Comment