When is a higher return not desireable in a Retirement Fund?

Question

I have been back-testing some share investment strategies for my SMSF (self managed retirement fund) and have narrowed it down to two strategies that have consistently performed the best over a 3 year period. A summary of the two strategies is summarised in the table below:

The starting capital for both strategies is $50,000 and the chosen strategy would only form a portion of the total investments in the fund which would also include some direct property investments, some cash and possibly some bonds.

Strategy A had a Compounded Annual Return (CAR) of 25.5% and almost doubled the initial investment in 3 years compared to a whopping CAR of 41.1% for Strategy B which almost tripled the initial investment over the same period. Looking simply at this information it seems like a no brainer to choose Strategy B. However, being a retirement fund, would it still be wise to choose Strategy B over Strategy A?

Notes Regarding the Strategies: In the back-testing for both strategies all trades where bought at the open on the next day once a signal was given and a 20% trailing stop loss was automatically placed on each open trade. I also plan to test both of these strategies going forward over the next 12 months in a virtual account to verify the results and avoid curve fitting the back-tested results. Once this is done and the results are confirmed I could start trading a real account in either 2016 or 2017 (depending on when I set up my SMSF), so if the results do not correspond to the back-testing I still have a further year available for additional planning.

Answer 1

If you can travel back in time, and start investing at the start of the period of the back-testing, the strategy with the higher return is better.

Past performance is not a good indicator for future performance, since markets change. But it's often the best we have.

A single 3 year period is not exactly good back-testing for a solid investment strategy. You can improve the back-testing by testing different scenarios instead of just a single 3 year period, and you will notice that in some of them Strategy A will come out on top. Make sure that you use different scenarios for fitting and for testing, otherwise you simply optimize for the testing scenarios. You will also notice that some times both strategies will lose a lot of money. Which strategy you chose then depends on which scenarios you expect to be how likely to come up in the near future, and how much money the strategies gain/lose in each of these scenarios.