Category: Investing Insight

Investing insight to make you a better investor.

Cross-Asset Time-series Momentum

Trend-following systems typically use the past performance of a particular asset to trigger a buy or a sell on that asset. A research paper that came out in 2019 looked at whether the historical performance of multiple assets can be used to trade them.

Pitkäjärvi, Aleksi and Suominen, Matti and Vaittinen, Lauri Tapani, Cross-Asset Signals and Time Series Momentum (January 6, 2019). Available at SSRN:

From the abstract:

We document a new phenomenon in bond and equity markets that we call cross-asset time series momentum. Using data from 20 countries, we show that past bond market returns are positive predictors of future equity market returns, and past equity market returns are negative predictors of future bond market returns.

Unfortunately, the paper did not look at Indian markets to check if this worked. So, we rigged up a simple backtest to see for ourselves.


A simplified equity-bond cross-asset trading strategy at the beginning of month t can be constructed as follows: Compute the past 12-month equity return (E past) and the past 12-month bond return (B past). If:

a) E past is positive and B past is positive: Buy equity
b) E past is negative and B past is negative: Sell equity
c) E past is negative and B past is positive: Buy bonds
d) E past is positive and B past is negative: Sell bonds
e) Otherwise, invest in the risk-free rate.
Hold the portfolio for one month and then repeat the same procedure in month t+1 (source.)


We used the NIFTY 50 TR index to represent equities, NIFTY GS 10YR index for bonds and the CCIL Index 0-5 TRI for risk-free rate.

Since our risk-free index starts only from 2004, our backtest only goes back 16 years. However, the markets have been through a lot since then, so it is unlikely we are losing much by not being able to go back much earlier.

The 12-month look-back approach massively under-performs the NIFTY 50 TR buy-and-hold. We shortened the look-back to 3-months to see if we could make the strategy more responsive to trend reversals.

To our dismay, we saw only marginal improvements in overall returns but the draw-down profile of the long-only portfolio was much better.


While the approach outlined in the paper might be valid for the selected subset of markets, it fails a simple backtest on Indian market indices.

Code for the backtest can be found on github.

Moats are for never

Be wary of #neversell

When Buffett-heads are asked about when they are likely to sell a stock, they often bring out this quote from the Master:

A common-sense parsing of the actual quote would over-weight the “outstanding businesses” bit over “forever.” However, the quote has been deliberately misinterpreted by asset-managers (who are paid as a percentage of AUM) to keep investors tied into their funds in spite of extended periods of underperformance.

Corporate history is replete with examples of companies that once had the widest of moats but withered nonetheless. The most recent posterchild is Intel.

Over the last decade, Intel (INTC) trailed the broader S&P 500 and Nasdaq indices with the final death-blow being dealt by Nvidia.

Twenty years ago, could you imagine a world without Intel? Now you can. Apple, Samsung, Amazon, Google and Microsoft are all in the process of developing or have already developed processors to run their operating systems and power their data centers. For a deeper dive, read this excellent piece by James Allworth.

And it is not just hardware. Twenty years ago, could you have imagined that Infosys would beat IBM and the S&P 500? Yet, it did.

Oh! But that is tech, you might say. What about the “real” economy stocks?

Remember GE? Their six-sigma blackbelts were supposed to be cream-of-the-crop problem-solvers who could be parachuted into any situation.

While you can argue that the above companies were cherry-picked, the base-rates are more shocking

A recent study by McKinsey found that the average life-span of companies listed in Standard & Poor’s 500 was 61 years in 1958. Today, it is less than 18 years. McKinsey believes that, in 2027, 75% of the companies currently quoted on the S&P 500 will have disappeared. (IMD)

table 1 companies exiting and entering_smaller454

Often, experienced managers are experts at solving problems for an old world order. The “best” managements often miss creative destruction happening in their own backyard, like Kodak. Incentives typically are setup to reward reaching local maximas. So, it is entirely possible to hit every single quarterly number while steadily marching toward bankruptcy.

The Coca-Cola Company is an example where initial high expectations were merely met (and not exceeded) to the dismay of common-stock holders.

And what is true about individual companies it true about the broader market as well. A visit to Japan will blow your mind. But as an investment?

In fact, a simulation of historical country-index returns show that only DENMARK, USA and SWITZERLAND had an extremely small chance of posting negative buy-and-hold returns. Out of the 43 Country specific MSCI indices we analyzed, half had more than a 10% chance of giving negative returns to buy-and-hold investors. India had a 6% chance (The Buy and Hold Bet.)

No company lasts forever. No market will always remain the best one to be in. No investment strategy will always deliver market-beating returns. No investor can consistently beat the market year-in/year-out.

No investment is forever.

Volatility and Allocation

Think in terms of volatility buckets, not assets

This post is part of our series on diversification and asset allocation. Previously:

  1. Diversification and its Malcontents

  2. The Permanent Portfolio

  3. Sequence Risk and Asset Allocation

  4. Static vs. Tactical Allocation

  5. Tactical Allocation

The thrust of our previous posts on allocation was that Indian investors shouldn’t blindly copy strategies that worked well in the US. There are a lot of qualitative arguments to be made to support a India-dominant view for allocation strategies. In this post, we introduce a quantitative aspect to the discussion.

It is Volatility, Stupid!

In finance, more than any other field, it is very easy to get correlation and causation mixed up.

A man goes to the doctor and says, “Doctor, wherever I touch, it hurts.”
The doctor asks, “What do you mean?”
The man says, “When I touch my shoulder, it really hurts. When I touch my knee – OUCH! When I touch my forehead, it really, really hurts.”
The doctor says, “I know what’s wrong with you. You’ve broken your finger!”

There are no universal laws for an asset class that holds across geographies and economic systems. The reason why a 60/40 Portfolio “works” in the US has more to with the quantitative aspects of the assets being mixed than what they are called. US bonds have benefitted greatly from a 30 year slide in yields, benign inflation and a flight-to-safety bid. None of these hold true for Indian bonds. So, expecting a 60/40 Indian portfolio to behave like a 60/40 US portfolio just because you mixed the same assets together is idiotic.

The most import aspect while considering assets for diversification are their volatilities. Specifically, the correlation of their volatilities at their left tails.

To keep things simple, consider a 2 asset portfolio: Eq and X. Eq has some average return that will be held constant during this analysis. What changes is its standard deviation (aka, volatility.) X is a stable asset with zero volatility (think of it as a fixed deposit.) How does different allocations to Eq change portfolio returns and volatility?

  1. Low volatility is supportive of higher allocations

  2. Higher allocations to the higher volatility asset progressively reduces the predictability of portfolio returns

Volatility is Volatile

Asset return volatility is itself volatile.

The past performance of a diversified portfolio is based on the realized volatility of its components. However, volatility itself is unpredictable over long periods of time.


While considering assets to diversify into, look at the volatility of the asset rather than what it is called.

Don’t expect the quantitative aspect of an asset class to transcend economic systems – different markets need different treatments.

All investing is forecasting. And all allocation is forecasting volatilities.

The Permanent Portfolio

Pain is eternal

This post is part of our series on diversification and asset allocation. Previously:

  1. Diversification and its Malcontents

  2. Sequence Risk and Asset Allocation

  3. Static vs. Tactical Allocation

  4. Tactical Allocation

The Permanent Portfolio – an equal weighted allocation to stocks, bonds, gold, and cash – was devised by free-market investment analyst, Harry Browne, in the 1980s. The basic idea is that no matter what the macro environment, the portfolio will not totally crash and burn.

The American Experience

Turns out, the theory largely worked for US investors.

If you look at the rolling 3-year annualized returns of the Permanent Portfolio, never has it given negative returns. In sharp contrast to equities and gold, US bonds have been spectacularly stable. So naturally, an equal weighted allocation to all for assets delivered decent returns with low drawdowns.

Did it work for Indian Investors?

Indian investors need to be careful with their bond allocations.

The Permanent Portfolio allocates 50% towards fixed income. This is a problem for Indian investors because unlike US bonds, Indian bonds do not have a “flight to safety” bid – they tank along with stocks during market panics.

A density plot of annualized 3-year rolling returns highlights the left-tail problem with the Indian Permanent Portfolio:


Beware of people preaching simple solutions to complex problems. If the answer was easy someone more intelligent would have thought of it a long time ago – complex problems invariably require complex and difficult solutions. – Steve Herbert

This is another instance of a “copy-paste” solution disappointing Indian investors.

The common thread connecting the misfiring of the 60/40 and the Permanent portfolios is the vastly different paths taken by Indian bonds. Is there a better way to crack this nut? Stay tuned.

Euclidean Distance for Pattern Matching

Most of us have learnt how to calculate the distance between 2 points on a plane in high school. The simplest one is called the Euclidean Distance – a pretty basic application of the Pythagorean Theorem. The concept can be extended to calculate the distance between to vectors. This is where it gets interesting.

Suppose you want to match a price series with another, ranking a rolling window by its Euclidean Distance is the fastest and simplest way of pattern matching.

For example, take the most recent 20-day VIX time-series and “match” it with a rolling window of historical 20-day VIX segments and sort it by its Euclidean Distance (ED.)

Here, the ED has dug up a segment from November-2010 as one of the top 5 matches. Take a closer look:

While not a perfect match, it “sort of” comes close.

Sometimes, a simple tool is good enough to get you 80% of the way. This is one of them.