Clear Skies Ahead for an S&P 500 Tech stock

“History is a mirror of the past and a lesson for the present” – Persian Proverb

We have an interesting market call this week looking at a large cap Tech stock.  Before we get to the stats, we wanted to bring some additional oomph to our commentary.

At EidoSearch, we apply pattern matching technology to generate predictive analytics from financial data like prices. We understand that there’s skepticism to the idea that we can predict future return distributions based on price patterns alone, so we wanted to take a step back and provide a brief introduction on the use of patterns (or conditional returns) from history and why they contain valuable information.

To do prediction at EidoSearch we use conditional returns, meaning that we take the price pattern of a current security and find mathematically similar patterns historically within a peer group, and then capture the market reaction to these previous similar environments to provide information about the future. This would be quite different then looking at all returns historically, or unconditional returns, to predict future returns.

This approach to doing prediction, using specific conditions, is very similar to how people forecast the weather. Weather prediction is something we all utilize and accept, and it impacts how we decide to dress and what we decide to do on the weekends.

Here’s an example. You can say that there’s a 10% chance it’s going to rain in Los Angeles tomorrow. Based on ignoring current conditions (unconditional weather patterns), if you look back 10 years and count the number of days it rains in each year, you’ll find that it rains about every 10 days. There’s valuable information in knowing that fact alone, but it ignores the current conditions and weather patterns that provide the more valuable information on what the weather is going to be tomorrow or next week in L.A.

Similarly, in the stock market, you can look at the S&P 500 (or an individual stock if you want), go back 10 or 20 years, and capture an average monthly return (about 75 basis points for the S&P). The base case, or average, is once again interesting from a relative perspective and it is how many investors rank their performance. But, similar to predicting the weather in L.A., you’re ignoring the current market conditions and trading patterns that contain the most valuable information about the next month.

To provide 1 day, 5 day and 10 day weather forecasts to their audience, meteorologists look at recent weather patterns as a primary factor. They track weather systems coming off the Pacific, tracking the speed, scale and trajectory. They’re models have seen these weather patterns thousands of times in all kinds of different environments, e.g. extreme drought, during winter months and in summer months. Based in part on those conditions they are able to provide accurate predictions on the likely weather scenario for us going forward. Sometimes the weather moves in faster than forecast, or with greater extremity (heat), but most of the time meteorologists are within a tight range using “conditional” or similar weather patterns to today from the past.

Price patterns contain the same kind of information. We’ve seen the current trading pattern in the S&P 500 many times before throughout history, with similar volatility, trajectory, etc. These similar patterns have occurred during all kinds of different environments, with Republican or Democratic Presidents, with high or low priced Oil and in good and bad economies. These conditional patterns provide valuable, predictive information about the likely return scenarios going forward. It’s not a crystal ball and accurate within a basis point or something, as there are way too many factors and other that drive the market. There is however valuable information in the range of expected returns and in the probabilities.

It’s no easy task to find the conditional patterns historically that have meaningful forward information in the securities you care about, but that’s where EidoSearch comes in. We do half a billion pattern comparisons every day, and we’ve validated the predictive nature of this approach and the accuracy of our return forecasts through millions of predictions.

So, let’s take a look at the current trajectory and pattern for Ciena Corp (CIEN) and hone in on the probable return distributions for the coming one month. We’ve also provided a couple of different studies, adjusting the conditions, to show the impact on the projection as well as the persistence of valuable information.

In projection #1, we use a 3 month price pattern, and look for similar patterns in Technology stocks historically across all market caps. Based on the 100 most similar instances, the average return in the next one month is 7.0% with almost 6x the upside to the downside.

In projection #2, we were a bit more narrow with our pattern search, looking only in market caps of $300 million to $10 billion and only looking at Communication Equipment stocks as the peer group for matches. The average return of those similar instances is 5.9% and there’s 2.5x the upside to downside.

Varying the parameters adjusts the projection a bit, but we’ve seen these patterns historically many times and they contain information on how the market reacts to these similar conditions historically.

If you live in the U.S., there’s a 30.9% chance it’s going to rain or snow tomorrow based on the average precipitation in major cities over the past 40 years. If you live in L.A., based on our understanding of current weather patterns and conditions, it’s likely going to be 70 degrees and partly sunny tomorrow.

Have a great week.