Have read in multiple books that
diversifying across investment models and assets is a must and wanted to check
this for myself, so here it is.
All data is weekly (since I have
weekly data readily available). For this test I have applied a simple moving
average to the Nifty Spot index and the same moving average to the S&P 10
Year Bond total return index. Let’s not get into an argument over what Is
better – my main focus of studying this is:
·
Linearity of the combined curve – i.e. was it a
smoother ride
·
Max drawdowns from peak
·
Portfolio volatility
We will be comparing 2 portfolios
– Rs. 100 allocated to a Nifty System and Rs. 100 divided equally, running the
same system on Nifty and Bonds. So at the start date Rs. 50 is allocated to the
Nifty system and Rs. 50 is allocated to the Bonds system, that’s it – and then
we buy & sell on the close as per the system rules on the day of the signal.
Again, I know buying the exact close is not possible due to slippage and commissions
but for simplicity let’s just keep the focus on the 3 things mentioned above.
First up, the linearity – who doesn’t
want a smoother ride eh? – I am using the R-squared number as a measure for a
smoother upward sloping curve. The higher the R-squared number the smoother the
ride. From the below charts we can see that the Nifty system has a R-sq. value
of 0.9141 while our combined system’s R-sq. is at 0.9622 indicating a better
linear curve.
Now on to the Max drawdowns –
again, it gets better for the combined system as majority of the time the max
drawdown of the combined system is lesser than using just the Nifty system.
Lastly, to check the volatility
of returns of the 2 systems I calculated the standard deviation (SD) of the
weekly returns of both systems for the entire data set. The SD of the Nifty system
is 2.04% while that of the combined system comes at 0.73%!! that’s less than
half of the standalone Nifty system. Although the combined system scored better
on all 3 parameters but it came at the cost of returns. Can’t get everything J
These are just baby steps and
basic building blocks; I think I will test out the combined system with annual
rebalancing in a future post, as of now we have left the starting allocation
static so will have to check if rebalancing provides a kicker. Also, if you can
point me to any books or websites that have such tests, that would be great.
Thank
you for reading! J
