# Brace for negative returns in 2018

Predicting stock markets is a futile exercise. You only need to compare historical predictions versus actual market movements to come to this conclusion. When bulge bracket investment banks and powerhouse economists with their multi-factor models of staggering complexity get them wrong more often than right, should the rest of us even bother trying to predict markets?

No, we shouldn’t. At least not to time entries and exits.

In preparing for battle, I have found that plans are useless, but planning is indispensable –

Dwight Eisenhower

Replace “plans” with forecasts, “planning” with forecasting, and “indispensable” with “mildly useful”, and that’s how I like to think about short-term forecasts.

**The exercise:**

To use three readily available NIFTY valuation metrics (Price-Earnings, Price-Book, Dividend Yield) to “predict” 2018 NIFTY returns

**Methodology:**

- Naively simplistic – Simple linear regression of each metric on NIFTY one-year returns followed by plugging current valuations to arrive at guesstimates of what 2018 NIFTY returns might look like
- Data used from Jan 1999 to Dec 2012. Output then used to predict returns from 2013 till 2018 so the extent of error is observable

**Summary Findings:**

- Of the three valuation metrics, Price-Earnings and Dividend Yield seem to have
*some*ability to predict NIFTY returns. Price-Book has negligible predictive ability - Predicted NIFTY returns are negative and range between flat (-2%) to deep correction (-36%)

### NIFTY Price-Book

Take a look at the chart below. Each point shows NIFTY Price-Book value versus one year return.

- As Price-Book increases, the range of returns tends to go lower, once PB higher than 5.5, the returns are almost always negative
- The thick cluster of points around the 3 to 3.5 mark on the x-axis means that most of the time the NIFTY tends to be priced in that range though returns show significant variation
- As of end December 2017, we are at Price-Book of 3.54 as shown by the red circle, implying a one-year return from this point of 18%
- However, R-square for this relationship (Price-Book vs Returns) is 0.2259 i.e. only 23% of the return is explained by the Price-Book value (1 is the theoretical maximum and would imply a variable that perfectly predicts another value). Thus Price-Book is not as strong a predictor of returns as the other two metrics

### NIFTY Dividend Yield

- Dividend Yield (Dividend / Price) understandably shows an inverse relationship with returns
- More importantly, the R-square for this relationship at 0.5083 is more than double that for Price-Book as a predictor. This is apparent even visually as the points are more tightly clustered around the straight line
- At 1.08, the current dividend yield is toward the lower end of the range implying a marginally negative return over the next year. However, note how there is a smattering of points in the positive return region even for this dividend yield

### NIFTY Price-Earnings

- Price-Earnings, the most popular valuation metric shows the strongest relationship (of the three) in predicting future returns (R-square of 0.5214)
- Historically, returns have
*always*been negative once PE goes beyond 24, although the sparse number of points implies when the NIFTY trades above this valuation either earnings rise or price drops quickly - Currently, NIFTY is trading at nearly 27 times earnings implying a significant correction of over 30% from this point

### Bringing it all together

The chart shows actual NIFTY returns (green line) with the returns predicted by the three metrics (Price-Book, Dividend Yield and Price-Earnings).

The gap between the prediction and actual is the error in the model, compare the error for Price-Book with Actual against the other two metrics. Note how the Price-Book prediction has stayed fairly consistent at 18% since 2009 even as actual NIFTY returns have been far more volatile.

The two more dependable metrics, Dividend Yield and Price-Earnings track actual NIFTY performance better and currently indicate negative returns for 2018. Brace yourself.

### Caveats & Conclusion

- 17 years of data only 12 of which were used to model is too little to be anything approaching confident. Also, markets are way more complex than to be explained by simple variables
- This is not advice to sell. Being an equity investor means being optimistic about the future. However, it’s worth looking at individual stocks and considering taking some gains off the table if valuations look particularly stretched
- Finally, predicting markets is a futile exercise
*The fine folks at Goldman Sachs predict NIFTY at 11,600 by Dec 2018, implying a +10% return from the 10,500 we’re at currently*

Have a fantastic year end. See you in 2018.

Wonderful article, Really like your disclaimers “predicting markets is a futile exercise”. I wonder if you have tried to use all three variables in single regression models. I think that would improve its predictability.

Thanks Ravi. The disclaimer is essential ðŸ™‚ will try the multi-factor model and post the results.

Robust common sense. Even the smartest guys have been cut to size by the markets.Why ? Simply because you can’t predict the future behaviour of millions of investors, traders or managers. That is the beauty of human dynamics. Shouldn’t we apply a little more common sense to buy and sell stocks instead of engaging in the foolish exercise of prediction. Just share what your common sense tells you about a particular stock, but of course with a sound rationale.

Pingback: Five charts that should worry Indian equity investors in 2018 | The Calm Investor

Pingback: The stocks professionals want you to buy in 2018 | The Calm Investor