0
$\begingroup$

Say I'm running a few backtests, say 1980-now, 2000-now and 2010-2015. When I'm calculating the Sharpe ratio of these backtests, do I use the risk-free rates associated with those time periods? or should I use the current risk-free rate for all of them?

My concern is that is that it might misleading to compare the Sharpe ratios of these different backtests to assess the viability of the different trading strategies I'm backtesting, because the Sharpe will be dependent on, say, what the T-bills at the time were looking like at the time of the backtest and not necessarily how profitable the trades are. However, if I use the current risk-free rate for all of them it will (I think) suggest what their risk-adjusted returns would be in the current market.

Is this a valid concern? Or should I just be using the historical risk-free rate for the times I'm backtesting. Thanks in advance!

$\endgroup$
  • $\begingroup$ I don't fully understand your concern, but markets work based on information available at the time. It never seems right in a backtest to use current (today's) information and apply it to a past period. Use historically correct interest rates. $\endgroup$ – noob2 Nov 17 '16 at 18:13
1
$\begingroup$

You should use the historical risk free rate so that it becomes viable for you to analyse your investment. But make sure your risk free rate should match the duration of the investment which was before while performing a back test. Because if you take the current one, it is not viable to compare your investment in history with the current risk free rate because the risk free rate is the average return over the time period under evaluation. And if you are backtesting a period (say 3 years) then you can take the average of the risk free rate (over 3 year period) over the same time period.

Hope this helps!!!

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.