In this article, we will discuss the methodology used for back testing baskets. Since we have already covered broad methodology and bias avoidance, we will just focus on the construct.
Remember, baskets are nothing but an investable portfolio management strategy. We have 2 types of baskets – static baskets (where instruments are fixed and pre-defined and do not change over time) and dynamic baskets (where instruments are picked as per certain set of rules)
In static baskets, instruments do not change over time. And therefore, back testing is fairly simple.
Basket is rebalanced depending on the rebalancing frequency selected by the user. On the date of rebalancing, instrument weights are calculated as per the weight method, and the basket is rebalanced back to new weights by selling overweight instruments and buying underweight instruments. Buying and selling happens at next day’s price/nav.
For static baskets, back test history will be available from the start date of the newest instrument in the basket. For example, a basket of 4 stocks comprising of RELIANCE, TCS, HDFCBANK and ZOMATO, back test history will be available from 26th July 2021 (i.e., date of listing of ZOMATO).
Back testing a dynamic basket is similar to back testing a screen/signal. The only real difference is that in screen/signal back test, rebalancing frequency is assumed to be quarterly, and stocks are assigned equal weights.
In case of dynamic baskets, actual rebalancing frequency and weights (as defined during basket creation) is used.
On the date of rebalancing, screen/signal is applied to get a new set of stocks. Instrument weights are then calculated as per the weight method. The portfolio is then rebalanced by selling old stocks and buying new ones. Buying and selling happens at next day’s average OHLC price.
At sharpely, you can track the live performance of your basket after its creation without actually investing in the basket. More like the ability to paper invest in the basket.
Methodology used for live tracking is same as back testing methodology. The only difference is that the start date is assumed to be the date on which the basket was created.
User should note that even though we call it “live” tracking, the performance reported does not include the potential impact of transaction cost and taxes.