Systematic Buy-the-Dip, an Update

We had looked at the difference between buying the dip vs. a daily SIP back in June-2016 (link.) The following is an update and an expansion of the same idea.

Introduction

Everybody wants to “buy the dip.” We wanted to run some numbers against what investors would typically do if they were to follow such a strategy.

  • We invest Rs. 1 every day the market is open.
  • For SIP, we buy Rs. 1 of a particular index.
  • For DIP, we either invest in cash or we liquidate the cash account and buy the index (if there is a ‘dip’), and we continue to invest Rs. 1 in the index as long as it is in DIP.

To define a DIP, we need two things.

  1. The look-back (formation) period over which percentage loss from peak is observed.
  2. The threshold percentage loss from peak.

The scenario will continue to be in DIP as long as the index is trading a threshold percentage below its peak.

The following shows the periods in which the NIFTY was purchased in red. The look-back period was 100-days and the threshold was -10%. This translates to “buy the index only when it is trading 10% below its previous 100-day peak.”

NIFTY buy the dip

Here’s the one for “buy MIDCAP only when it is trading 15% below its previous 200-day peak.”
MIDCAP buy the dip

We expand on our previous attempt by adding more indices and running the scenario through rolling 5-year periods. We then compare the terminal value of the number of units of the index purchased under DIP with SIP.

Results

buy the dip results

There are very few scenarios where DIP buying is better than SIP buying. This happens to the NIFTY when you buy it when it is trading 20% below its 100/200/500-day peak. However, this is “on average.” There were 5-year periods when DIP buying would have resulted in about 5% less assets than SIP buying under the same lookbacks.

Hence, we reassert our previous finding that waiting for the dip is not a good idea compared to a simple daily SIP.

Note that we define success as the terminal value of the number of units of the index purchased. This is different from each unit being profitable.

Code is on github.

Comments are closed, but trackbacks and pingbacks are open.