Investing with headphones on

'Signal' and 'noise' are electronics-engineering terms that have escaped into wider usage, including that of investment research. Here's how the Wikipedia entry for signal-to-noise ratio introduces the idea:

Signal-to-noise ratio (abbreviated SNR) is a measure used in science and engineering that compares the level of a desired signal to the level of background noise ... is sometimes used informally to refer to the ratio of useful information to false or irrelevant data in a conversation or exchange. For example, in online discussion forums and other online communities, off-topic posts and spam are regarded as "noise" that interferes with the "signal" of appropriate discussion.

When using electronic devices, you could hear the noise as a hiss behind the music. In investment research, you hear the noise as news, views, data and analyses that sound like they should be relevant but are actually irrelevant or misleading. The actual signal-to-noise ratios are much, much worse in the latter. In audio equipment, the signal-to-noise ratio is generally far below 1 per cent, meaning what you hear is almost pure, unadulterated signal. In investment analysis (along with business and economic affairs, generally), you would be lucky to have less than 90 per cent of noise.

Suggested read: Panic and sudden meltdowns

The reasons are many, some of them bona fide and others less so. However, what distinguishes investment analysis is that if it is to be useful at all, we need to use the signal to make forecasts and predictions. After all, if you want the past financial data, derived ratios and stock prices for every listed company, you can have them with zero noise. Where the noise starts is when people use these to try to figure out what they mean for the future.

One of the universal characteristics of noise in real-world data and information is that the more closely you look at it, the worse its effects are. If you look at minute-to-minute stock prices, that's almost all noise. It's impossible to derive any trend out of them. Daily prices are almost as bad. Monthly ones are slightly better, quarterly ones are more so, and yearly ones are even better. It's easy to understand this in a simplified manner.

Think of the rise of a stock price or index over a long period. The Sensex, for example, rose by 1,410 per cent over the last twenty years from 5,584 on September 30, 2004, to 84,300 on September 30, 2024. This is an annualised rate of 14.5 per cent. During these twenty years, if you had looked at the Sensex change every day, you would have had no sense of what the long-term rate of change was. During this time, there were 4,962 trading days, of which 2,278 were negative and the rest were positive. In the daily change pattern, when you look at the next day, there is little quantitative sense of where the market is going.

Now, let's look at the yearly percentage changes. Here's the series: 54.6, 44.2, 38.8, -25.6, 33.2, 17.2, -18.0, 14.0, 3.3, 37.4, -1.8, 6.5, 12.3, 15.8, 6.7, -1.6, 55.3, -2.9, 14.6, 28.1. It is somewhat wandering, yet it has a better sense of direction. And here's the five-yearly series, expressed as annualised returns: 25.1, 9.2, 7.7, 16.9. It's only in the five-yearly series that you get some sense of where the Sensex is actually heading.

If you were to look at the markets only once in five years, then you would have no anxiety that looking at them for shorter periods would have. There would be almost no noise, just the signal. It's the difference between listening to music in a noisy place with speakers and through noise-cancelling headphones. This is the psychological basis of being a good equity investor, and this is why we never pay attention to what is happening to stocks over short periods.

Also read: The great moderation: Volatile headlines vs the steady reality