Mixed Signals

Rate of change is one of the most widely used technical tools. We a look at its accuracy

Does the "rate of change" (RoC) accurately send out profitable buy and sell signals at all times? To find out the answer, we took the daily prices of the Sensex companies for the past six years and applied the technical tool on this data. We also tested the RoC in various ways to observe its behaviour pattern in the Indian market. What we found was that while the RoC could actually provide a very good indication about major trends in the near term, following it blindly and that too, as a stand-alone device can be devastating for investors.

After the extremely popular moving average (MA), RoC is, perhaps, the most widely-used tool of technical analysts. RoC is used to measure the momentum of stock prices. It indicates the percentage by which advances and declines in stock prices occur over a period of time. It is worked out on a continuous basis. For example, a 30-day ROC gives today's price as a percentage of the stock's price 30 days ago. The RoC is plotted on a chart on a rolling basis and connected as a line, around the 100 mark. A 30-day ROC above 100 indicates that the current price is above the price 30 days back and vice- versa.

So an ROC trend continuously above 100 indicates a bullish phase, while the same persistently below 100 indicates a bearish phase. A stagnant market is indicated when the ROC flits around 100.

How, then, does the RoC give "buy" and "sell" signs? Well, whenever the RoC crosses the 100 mark from below, it indicates the start of a bullish phase and can be viewed as a "buy" signal. Whenever it pierces the 100 mark from above, a bearish phase is suggested - it's an indication of a start of a bearish and can be a "sell" signal.

We, in fact, don't recommend following this rule: we recommend a better technique. And that is: you buy only when the market is oversold and sell when it is overbought.

There are no precise definitions for the terms "oversold" and "over-bought". You'll need to define it yourself. If you feel that a 10 per cent increase or decrease in the stock price over 30 days is unusual, then for a 30-day ROC, you can define the region above 110 as "overbought" and the region below 90 as "oversold".

So, buy whenever the RoC crosses above 90 and sell whenever it falls below 110.

We found that following the trading signals of RoC blindly without defining any trading band (beyond the trading-band there may be overbought/oversold regions) is detrimental. On average, 67 per cent of the signals given by various RoCs ranging from 30 days to 360 days were found to be false alarms. We recommend using a 10 per cent band - that means considering the region above 110 as overbought and below 90 as the oversold region and making investment decisions accordingly.

This technique gave around 75-80 percent fewer signals than that given by the RoC without a band, but the percentage of false signals were far lesser. Against the average 67 per cent wrong signals given by an ordinary RoC, the "band" RoC gave off only 14 per cent wrong signals.

A narrow band gives more false signals; on the other hand, a wider band gives fewer trading signals that aren't worth following. Of course, we aren't saying that a band of 10 per cent on either side is the best standard to be used for all stocks. But we found that it worked well with the Sensex stocks, on average, though it did differ from stock to stock.

For example, a highly volatile stock like ACC with a 20 per cent band on either side might appear promising.

We used the same 10 per cent band for all RoCs with different durations. Usually, for an ROC with longer duration, the overbought and oversold zones should be defined within higher levels, because while a 10 percent rise or fall in price over 30 days may be unusual, over one year such a fluctuation may be, in fact, minor. Thus, for the test, we defined 90 as the oversold and 120 as the over-bought mark for both the 180 and 360-day RoC.

Thanks to the sluggishness in the Indian market, the ROCs indicated hardly any trading opportunities. For example, over the six-year period, both the 180-day and 360-day ROC of the Sensex gave out only two pairs of trading signals each. Of course, for understanding the momentum of the market, a wide band is a must for a longer ROC.

Does that suggest that the Indian equity markets are basically short-term plays?

Some analysts suggest using a moving average instead of the original price to make the RoC. Moving averages smoothen out the price trend. We find this technique unhelpful. It only gave out a larger number of wrong signals, pushing you into making more losses.

(An RoC provides "buy" and "sell" indications on a continuous basis at various points in time. That makes calculating a meaningful return over a certain period of time extremely difficult. Thus, we abstain from making any suggestions about what kind of returns can be made; it may only be misleading.)