It's Going to Be a Good Q3 Earnings Season

This post is strictly a display of the engineer coming out of me. As a mechanical engineer, I'm fairly familiar with how statistics are used to estimate something. It's called statistical sampling. I'm sure you've heard of it before, but you may not have realized. It comes up every few years in a democratic state...how? Well, it's all in the polls. When an election comes up, polls are taken and each candidate or party is assigned a percentage of the votes predicted. For example, you will hear that the polls predict Candidate A will win the election, with the poll being right 19 times out of 20.

A Brief Lesson on Statistics

It all sounds a little unintelligible...how is it done? Well, as you might have guessed, the polls don't actually go to every single house in the riding and ask every eligible voter who he/she is going to vote for. The poll goes around and takes a random sampling of voters, say 1000 of them. Out of the 1000 voters, their opinions will give a picture of how the election will turn out to be. Assuming a normal distribution (bell curve), the poll will use what are called confidence intervals. In our case, it would be 95% confidence intervals, hence 19 times out of 20. Basically, the poll gives say Candidate A 45% of the vote. The 95% confidence interval can be42% -47%. Because it was only a sample of 1000 voters, the actual vote could vary between 42% - 47%. And we are 95% confident that this interval captures the right percentage.

So What?

So, why am I bringing this up? How does this apply to Q3 earnings season? Remember, you heard it here first. I'm using statistical sampling to predict how the entire Q3 earnings season will turn out. Using Yahoo Finance (http://biz.yahoo.com/z/extreme.html), I'm able to see which companies have beat, met, or missed earnings.

From Oct 5 - Oct 14, 37 companies beat earnings and 10 missed. The proportion is 78.7% beat earnings. This is a small sample of the companies that have reported, but we can estimate what the 95% confidence intervals are for the entire season. The confidence intervals can be calculated with the following equation:

For those who want to learn what the heck this is, you can visit wikipedia at http://en.wikipedia.org/wiki/Binomial_proportion_confidence_interval. Anyway, the intervals turn out to be from 67.0% - 90.4%. What this means is that we can be 95% confident that by the time Q3 earnings is done and over with, 67-90% of the companies will have beat their earnings estimates. So, up to 9 out of 10 companies could beat earnings this season! This is HUGE! Even if the final outcome is on the low side, two thirds of the companies will have beat earnings. This translates to an awesome earnings season to me! With this estimate, I'm quite willing to keep my money in the market!

We can continue to monitor and improve our estimates as new data comes out. By 12:30 pm on Oct 15, 93 companies have beat and 20 have missed. The proportion is now 82.3%, which is right inside our initial prediction. With the larger sample size, our new updated intervals are 75.3% - 89.3%. So, we're definitely on a good track to a good earnings season.

But...

There are some caveats. The market is ever changing and so are earnings estimates. Market analysts are free to change their estimates as they continue to observe the earnings that are coming out. So, their estimates may drift up or down depending on how other companies are doing, which means this method is not totally accurate!

As well, our sampling may not be randomized enough...So, be careful when using this method. Having said that, I am still predicting a good earnings season, which would logically result in a good market!