8th Jan 2008: NIFTY closes at a lifetime high of 6,288.
27th Oct 2008: NIFTY closes at 2,524. Down 60% from the January high. Till date, this remains the largest drawdown in its history.
What would a simple and implementable system that warned you to get out of the market before impending declines look like?
Yes, I know. It can’t be done.
All the “market timing does not work” literature categorically proves it. But isn’t the truism “Buy low, Sell high”, an attempt to time the market? So let’s consider this in the safe confines of an academic study and see what we come up with.
Since we are looking for a simple system, let’s only consider the most widely discussed index valuation metric, Price-Earnings as an indicator. There are many global studies of the inverse correlation between starting index P/E and future returns. The relationship holds for India too. Higher the starting P/E, lower the returns.
So, the value investor’s starting hypothesis would look something like this: “It is possible to get better returns than the index by investing and staying invested below a Price-Earnings “threshold”, and exiting and staying in cash every time Price-Earnings exceeds that threshold.”
In other words, switch to cash when markets run red hot and turn “expensive”, and buy equities when markets cool off and become “cheap”. What could be simpler?
Since the NIFTY is a theoretical portfolio (you would have to buy the stocks in that proportion), we use the closest representation by using the NIFTYBEES in our analysis. The NIFTYBEES does a good job of tracking the NIFTY, so it this would be an implementable strategy. But is it worth implementing?
The “exit when expensive” strategy
What the “Exit when expensive” strategy would look like: Based on historical data from 2002 to Mar 2019, we know the range of Price-Earnings the NIFTY travels between. We decide what value of PE is “expensive” and anticipating a fall, we exit to cash, and re-enter only when the PE goes below that same threshold.
Based on 17 years of data, the NIFTY has been at or below 19.7x earnings 50% of the time. At or below 21.6x 70% of the time, at or below 22.9x 80% of the time, and at or below 25.1x 90% of the time. Another way to read this is only 10% of the trading days in 17 years, the NIFTY has closed at a Price-Earnings above 25. As of writing this on Mar 31, 2019: NIFTY PE is 29.
If you applied a strategy of exiting above 25x PE (90th percentile threshold), you would cash out on 1st Nov 2018 and be in cash as of Apr 2019, as PE rose to the current value of 29 (99.9th percentile – literally the most expensive in P-E terms the NIFTY has ever been). This strategy makes sense (or seems to), given you’re expecting a strong correction once Price-Earnings go to these levels. Except, consider the impact of such a strategy over the long term in the chart below.
What happened there? Two problems appear from the chart above. One more serious than the other. The smaller (seemingly easier) problem is how the more “aggressive” the threshold, the less often the strategy seems to kick in. Example, the 90th percentile (hold cash above 25x PE) rarely kicks in compared to the others. Notice the grey line largely mirrors the buy-and-hold yellow line except for a period in early 2008 and then since late 2018 to now.
The problem of using a static PE is evident when you consider how NIFTY Price-Earnings have fluctuated and have trended up over time (see chart).
Till 2007, a PE of 20 was relatively expensive, but since 2015, a 20x PE is almost cheap! Our application in the first chart is not valid because it applies an artificially high threshold based on the recent few years of high NIFTY Price-Earnings on the entire timeline. For our “exit when expensive” strategy to be applied correctly, it would need to apply different PE thresholds at different points in time depending on historical values.
The other, more serious problem is the massive underperformance of this strategy versus buy-and-hold! The counter-intuitive outcome of this strategy was to exit equities just when markets were on their way up, and to get back in on the way down. Notice how the upwards surges are lopped off in the blue and orange lines, only for them to resume copying the yellow line on their declines, only to a lesser extent because they had a smaller distance to fall.
Correcting the fixed threshold problem makes matters worse for the “exit when expensive” strategy because it ensures the strategy kicks in more often, exiting on the way up and rejoining on the way down.
So, the seemingly logical strategy of selling when markets get expensive and buying when they decline ensures underperformance versus a simple buy-and-hold strategy.
The opposite “exit when markets fall” strategy
Our value investor’s hypothesis blown up like a car in a Rohit Shetty movie, we have to look elsewhere. And the simplest thing to do is to “invert, always invert”.
What if we applied the opposite of the “exit when expensive” strategy? i.e. Stay invested when markets turn expensive and exit when markets turn cheap?
Chart below shows exits driven by a 5-year lookback of NIFTY PE.
Exiting when NIFTY PE falls below a threshold, and re-entering when it exceeds that same threshold works way better than a buy-and-hold strategy. Notice how this strategy signals an exit in early 2008 when markets start to correct, to then re-enter after the recovery was well underway. In spite of rejoining late (Sep 2009 for the 90th percentile), the higher base value ensures it matches the buy-and-hold strategy. The 80th and 70th percentile (lower threshold to exit) do much better.
How does a shorter lookback to set threshold Price-Earnings impact outcomes?
Turns out it worsens all outcomes, but the 70% threshold still outperforms buy-and-hold.
Shorter look backs to calculate thresholds mean worse outcomes, but lookbacks longer than 5 years also start declining.
Before we rejoice, we need to realise the major problem with this strategy. It would mean long periods sitting out. Note the flat stretch from 2011 to 2015, four years! Let’s face it, no investor can stand 0% allocation to equities for that long.
There are many ways these results won’t lend themselves to a longer-term strategy. But there’s enough there to warrant a closer look to be potentially applied to a value strategy. Sell stocks on the way down and buy them on the way up. The resemblance to a momentum strategy might just be coincidental.