Linear regression is a statistical tool used to help predict future values from past values.

It is
commonly used as a quantitative way to determine the underlying trend and when prices are
overextended. A linear regression trendline uses the least squares method to plot a straight line
through prices so as to minimize the distances between the prices and the resulting trendline.
This linear regression indicator plots the trendline value for each data point.

Understanding linear regression

The simplest form of the regression equation with one dependent and one independent variable is
defined by the formula

y = b*x + a

where y is the estimated dependent variable, b is the regression coefficient, or what is commonly
called the slope, x is the independent variable and a is a constant. In simple words, y is the output
when m, x, and c are used as inputs.

Linear regression does try to predict trends and future values. It essentially, though not with pin-
point accuracy can answer questions like,

 What could be the price of Infosys in the next 3 months?

 What could be the price of Gold in the next 6 months?
 Or Where could the Market go if the existing trend continues into the future?

Limitations

The future is shown as a linear increase which is not true as if you observe the actual values
you see a jump up and a fall down in the prices. There is a lot of variability in the actual data but
a standard increase in the forecasted values.

The line goes only one way which means if the past has been of an average increase over a
period of time then the future will also show the average increase over the next period. In reality,
there is always a danger of trend reversal which the linear forecast can never reveal.

The values by and of itself differ from the actual by large percentages in excess of 10% which
shows it is not such an accurate predictor.

Usefulness

It can tell you on average, what has been the past trend, and like mentioned at the start, if the
future trend happens to be the same as the past trend then your prediction will be quite accurate
using the simple linear regression model which isn’t a lot of work
 Using the percent error, a range could be deduced and it can be expected the future value
could lie within this range.

Once can work out all the cases and become aware of the future values if the trend improves,
worsens, or remains the same and so have an idea of the possible ranges.

It is a quicker and easier method to apply using the most popular analysis tool which is excel
and unlike highly complex processes that try harder at forecasting accurate values is much easier
to interpret and explain to non-experts.

Conclusion

We knew from the past data the rate of the linear increase, we knew that we do not know
whether the future is going to be better than the past or worse than the past or equal to the past.