Let's say you manage money for many clients. It?s the year 2000, and you?ve put together a strategy and back tested it over the last 10 years. Your back testing showed rosy results, but then it severely underperformed due to both the dot.com crash in 2001-2003 and then really did poorly in 2008. If you had back tested over a longer time span, like the early 1970?s, then you would've seen how a recession/bear market like 1974 would have affected the strategy. Going further back, you would see how a war and depression would have affected it. Clearly this doesn't give perfect forecasting ability, but it does give *greater confidence* in what you think will happen in the future. What we want to know is how far should we go back to ensure that the future results are reasonably inline with past results. Further, should we use different time periods for back testing different asset classes (e.g. real estate, bonds, or stocks)?
| Previous | Back to FAQ for Prota Challenge 2017 | Next |