The importance of asset allocation frameworks is only understood at times when the market has fallen, when one can see the bargains in equities all around them but doesn’t have the necessary meaningful amount of liquid capital to deploy.

What is asset allocation you ask? It is simply the decision one makes at different time periods, on what % of capital to deploy in different asset classes. For a simple investor, liquid asset classes could include, cash, gold, debt, and equity. Notice that I do not include real estate into the mix as it is not easily liquifiable to raw capital.

One can use the table below, to make such asset allocation decisions across different market cycles. Countercyclical asset allocation decisions in equities and other lesser volatile asset classes can yield improved long term results for any investor.

Don’t believe me, look at how Warren Buffett invests, he always keeps cash handy for big market falls. He currently has $ 140 Bn in liquid assets to deploy in a market crash. This probably forms 30-40% of his total portfolio. In 2008, he was one of the investors approached by the biggest American banks to help save them with capital injection. He did this at terms that were extremely favourable to him.

While small investors like me and you will not have the clout to make such deals, the equity markets give us enough opportunity to reach there.

The inspiration for this post is taken from my friend Deepak Venkatesh, who shared his findings on his blog, https://peepalcapital.blogspot.com/. Be sure to check it out, I have taken only a subset of his work and expanded upon it, his original work is a must-read.

Below you will find the main table which summarizes the findings of the study. You will need some knowledge of normal distribution, percentiles to understand what is going on, I will try to simplify as much as I can.

The PE data is for NIFTY50 from Jan-2000 to 30th March 2020. We then break this down to different ranges to gather insights from the data. The middle part of the table is the ranges used by my friend Deepak, I expanded upon it to break the ranges in percentiles of 5% in the bottom table and 10% in the upper table.

Percentile is similar to how students in CAT exam are ranked, the top percentile 99.99% is the best, similarly, in our data, the higher percentile reflects higher PE range.

After percentile, we have a look at how much time the market spends in certain PE ranges, this is depicted as % of Days.

The average, median, maximum and minimum CAGR columns, denotes the respective figures for 1/3/5 year timeframes. The data for 3/5 year timeframes is denoted in CAGR terms. Essentially we have taken the rolling returns data and applied these filters to it.

If you don’t know what rolling returns are, they are essentially calculated on each rolling time period falling between a date range. For example, if we have one year of historical price data, we can have around 250 unique rolling time periods in that one year from one day to the next. There are around 250 trading days in any year.

So how to read these columns, for a given PE range (ex. 10.68-14.64), which falls below the 10% percentile, the average, median, maximum, and minimum returns across different time periods are given in the 1st row of the table.

The last 3 columns (1/3/5 Yr) are the most important to use the framework in asset allocation decisions. Under the 1 yr column, I have counted the number of rolling periods which have given more and less than 12% return. This 12% is the threshold return picked by me which is necessary to achieve a win criterion. This is the minimum return an investor should strive for taking equity risk.

The win/loss ratio is just the division of the number of winning period by the number of losing periods. A W/L of 1 means you have equal odds of winning and losing. This is similar to the case of a coin toss with a fair coin, the probability of winning or losing is 50%. Next to the W/L column, I have also provided the winning probability figures which should be easier to understand. A probability of greater than 50% means the odds are in your favour.

The PE ranges where the odds are in your favour (W/L greater than 100%) are highlighted in green over different time periods.

Caveat lector, these odds are dependant on historical data, future may rhyme with the past but there could still be back to back black swan events or ELE (Extinction Level Events) which could diverge the results from past data.

It is interesting and intuitive to see that when capital is deployed at the lowest PE ranges, the odds of one losing are almost nil, and at the highest PE ranges, the odds of losing are an absolute certainty. Therefore, one can develop an asset allocation matrix based on these probabilities. Deploy capital to equities when the odds of winning are in your favour and take out capital when they are not.

To sum up the findings from the table, historical data suggests that ~18 PE is the sweet spot below which one can be certain to deploy capital in the markets and come out a certain winner across different time frames.

Mind you I am not suggesting to deploy all the capital once the Index PE touches 18, one can start deploying incremental amounts of capital below this figure. As you can see in the minimum returns column, even at index PEs below 18, the 1 and 3-year returns have been negative in some periods.

I have deliberately not shared my asset allocation matrix with the table, as I feel everyone’s risk profile is different and one should understand the odds and allocate capital as per their preference.

I have shared some charts which are different visual ways to looks at the same data. Have provided the google drive link at the end for the images in high resolution.

PE Percentile vs Return Win/Loss Probability

Charts: Google Drive

Data Source: