In Short

- Contrary to how it might seem in financial media, momentum as an investment strategy has been around for the better part of a century.
- Empirical studies going as far back as 1801 have shown momentum works, not just in equities, but across asset classes. Momentum outperforms broader markets.
- Momentum outperforms benchmarks in India. We extensively backtest a momentum strategy against portfolios of randomly selected stocks, and against the benchmark indices.
- But the outperformance comes at the cost of higher volatility and higher point-in-time drawdowns in the short-term. Stick with it though, and it does measurably better than the market.
- Finally, we look at the basic momentum investing strategies (Simple vs GainLoss vs Volatility-Adjusted Momentum) to see if some consistently do better than others

### What is Momentum Investing?

At its core, momentum investing is built on the simple premise that stocks that have gone up (down), will continue to go up (down), at least in the short term. Therefore, buying a portfolio of such (long up, short down) stocks *should* offer better returns than the broader market.

This idea is not new. It has been around for almost as long as organised investing has been around.

Momentum investing probably became mainstream in the early 90’s when a paper by two professors, Narasimham Jegadeesh and Sheridan Titman, at the time at University of California, was published in the Journal of Finance. The paper is called “Returns to buying winners and selling losers”, and tested NYSE stock returns from 1965 to 1989. At the time, they called it a “relative strength” strategy.

Not only that, evidence of momentum’s outperformance goes all the way back 212 years. Believe-it-or-not, to the year 1801.

If you consider the evidence, momentum investing has more hard data backing its persistence as a strategy for outperformance than any other school of investing thought, including value.

### But surely India is different?

It’s all very well to have confirmed that momentum as a strategy works in developed markets like the US and Europe.

**But does a strategy that buys winners purely based on price returns work in the Indian markets?**

Since there aren’t any published studies, we did our own set of backtests. We compared the returns of such a strategy versus randomly picked portfolios, and also against the benchmark index.

Why compare against randomly picked portfolios? Stock prices tend to move up more often than they move down. So it’s important to be able to tell whether any active strategy is better than just throwing darts to pick stocks.

Backtesting challenges. In general, and in Indian markets (and how we work around them):

- Data. Getting reliable price data for Indian stocks (accurate, and adjusted for splits and bonuses) going way back is a challenge. We were able to get split and bonus-adjusted data going back to 2007.
- Liquidity. At any time, there are over 2,000 securities listed on the NSE (more than double that on the BSE). As of writing this, only ~750 (34%) of those securities saw trading more than ₹ 1 Cr (10 Million).
- Survivorship Bias. A lot of backtests get this wrong. Using the current list of traded stocks, going back in time, and picking from them makes any strategy look better than it is. To do it right, our data also included all companies that went bankrupt or stopped trading.

### Competing with dart-throwing monkeys

Here’s how a randomly picked portfolio of 30 stocks, rebalanced monthly, would have done between 2007 and 2019.

Bear with me a moment to make sense of this chart. It is the outcome of a Monte Carlo simulation that picked random portfolios 1000s of times to tell us where we are *likely* to have ended up with such a strategy.

Here’s how you read it. The top row of percentages from 1% to 99% implies the percentage of times you pick random portfolios. The red bars are the maximum returns from those portfolios. i.e. Out of all the times we pick purely random portfolios, only 1% of the time, we see a 100% drawdown (the extreme left bar). At the other end of the chart, 99% of the time that we do this experiment, our return is 7% or less. Or to say it the other way, only 1% of the time that we picked random portfolios, did our annual return exceeds 7%.

In plain English, if we picked random portfolios of 30 stocks and rebalanced every month from 2007 to 2019, our annual return would almost definitely be under 7%, and most likely be between -7.43% and 5.78%.

If we run our momentum backtest and get returns in the above range, we’d have to conclude that momentum investing is no different from picking stocks randomly, and so has no basis to be an investment strategy.

### A basic momentum strategy

There are a few different ways to measure “momentum”, but for this part of the discussion, we go with the most basic implementation.

- On the first trading day of the month, we rank all stocks based on their absolute return over the last 52 weeks (1 year).
- Buy the top 30 stocks in equal weight from this list. This is a long-only strategy. The long-short version of this strategy would also short the lowest-ranked 30 stocks.
- Rebalance every month.

Why 30 stocks? 30 offers a reasonable amount of diversification and ensures the results are not dominated by one or two outliers, and so is a good test of this strategy. Note that our actual implementation in the Capitalmind Wealth PMS varies in some ways which we will get into in subsequent posts.

Here’s how a simple momentum investing strategy did between 2007 and 2019.

₹100 in Jan 2007 in a simple 30-stock monthly rebalance momentum strategy would be ₹464 today. The same amount in the NIFTY and the NIFTY 500 would be close to ₹300. This translates to an annual return of 12.6% from momentum, while the NIFTY and NIFTY 500 did 8.9% and 8.7% respectively.

I’m not a fan of such cumulative return charts. They are susceptible to starting point bias and do not offer insight into the consistency of a strategy’s performance.

Breaking down the comparison by year provides a little more insight.

Momentum outperforms both indices in 6 of the 13 years. But also underperforms spectacularly in some years, most notably in 2008 and 2016.

This table offer better context around the relative performance of momentum.

Note that Maximum DrawDown in this table refers to the worst drop from its peak value. It looks scary because it is. Here’s the chart of just drawdowns compared.

Does a 60% drawdown feel significantly better than an 80% drawdown when it’s happening?

Let’s revisit the first chart showing the performance of a random stock-picking strategy, and compare against the simple momentum strategy.

The random strategy would have given us negative returns roughly half the times we implemented it. The most important takeaway from this chart: Simple momentum not only delivers positive returns over the long-term almost every time, 90% of the time this strategy beats the NIFTY.

Here’s what we can conclude so far:

- Higher returns: Momentum investing works in India. A consistent approach to picking stocks showing the highest relative momentum delivers a non-trivial alpha compared to the index. This premium exists in spite of a particularly poor year at the start of our backtest period.
- More “up” months: Momentum delivers a higher percentage of up-months compared to the index (63% vs 54%)
- Deeper drawdowns: This is not a strategy for the faint-hearted. Momentum underperforms markets by a distance when markets correct sharply.

### Can we do better than using absolute returns as a measure of momentum?

Remember the basic premise of momentum investing is to buy winners. So far we used a simple measure: absolute return over the last 252 days to pick our 30-stock portfolio.

What if we kept everything else the same but modified the way we identify the strongest momentum stocks?

#### Gain-Loss as a measure of momentum

What if we tried to identify stocks that are seeing more definite uptrends? One such measure is called Gain-Loss. It is the percentage of up-days in the last 252 days of trading multiplied by the absolute return.

The hypothesis is, higher the Gain-Loss Ratio, stronger the upward momentum of the stock.

Here’s how such a strategy would have done:

₹100 in Jan 2007 in a Gain-Loss Momentum strategy would be ₹594 today compared to ₹464 with Simple momentum. That’s 14.7% from GainLoss momentum versus 12.6% from Simple momentum.

#### Volatility-Adjusted Returns (Sharpe) as a measure of momentum

This is another way to quantify momentum. The absolute *excess *return over the last 52 weeks divided by the annualized standard deviation of daily price movements.

Essentially, we are penalizing stocks that see big up and down moves and favoring stocks that see smooth upward trends. Hence the name, Volatility-Adjusted Momentum.

₹100 in Jan 2007 in a Gain-Loss Momentum strategy would be ₹749 today compared to ₹594 with GainLoss Momentum, and ₹464 with Simple momentum . That’s 16.8% from Volatility-Adjusted Excess Returns compared to 14.7% from GainLoss momentum and 12.6% from Simple momentum.

Here’s how they compare on drawdowns:

Applying criteria that include quality of momentum in addition to just quantity, we are able to reduce the pain such a strategy has to endure.

Comparing the three strategies more comprehensively, and against the benchmarks:

Conclusion:Momentum strategies beat the benchmarks consistently on overall return. There is a price to be paid in terms of being to handle sharper drawdowns. Introducing measures of “momentum quality” help improve both return and downside performance.

Our analysis shows, with the right set of starting hypotheses, it is, in fact, possible to reduce *some more* of the downside of a momentum strategy while retaining most of the upside.

In further posts on momentum investing, we examine decisions like the optimal size of a momentum portfolio, whether looking back one year is the right period, if stock-level stop-losses help, and if portfolio-weights matter. More to come.

Please note that backtested results are hypothetical and do not represent actual results. An investor applying this strategy would incur transaction costs and taxes on short-term gains which are not reflected in the backtest results. On the plus side, the backtests do not include dividends or their reinvestment. Our own momentum strategy at Capitalmind Wealth while built on the same principle varies in some ways from the above implementation and undergoes periodic reviews to use the most relevant factors.

*In the Capitalmind PMS, CM Momentum is one of the portfolios we invest manage. We also offer a Momentum Portfolio to our Premium subscribers. The CM Premium Momentum Portfolio is also accessible on smallcase.*

*If you’d like to know more, tweet / DM us on twitter @Capitalmind_in or **@CalmInvestor*