I thought I would share something a seasoned analyst taught me. It’s a simple concept, but it was not overtly obvious, to me at least, when I first encountered it. It’s the third option below. I go through the first two options for reference.

When screening equities, using metrics (or ratios), you’ve got three main options (or some combination of the three options below):

**1). Static Values** – For example, P/E < 17. You’re comparing the P/E metric to a static number. Through study and research, perhaps, you have determined that, on average over a substantial period of time, the P/E metric for all equities averages about 17x. You want to buy companies trading below this average and sell companies trading above this average. Your premise is essentially: when considering all equities as a whole, they will eventually revert to 17x earnings. The same could be said of a particular industry, although the beginnings of a problem arises here. Maybe, you’ve determined that, over a substantial period of time, retailers tend to hover around an average P/E of 20x. Here again, you could limit your screen to retailers, purchase retail companies below 20x earnings, and sell above 20x.

**2.) Dynamic Industry or Total Market Values** – As noted above, a problem arises when comparing the total market P/E average to the industry average for retailers. Mainly, the total market average (17x) differs from the retailer industry average (20x). We can safely assume other industries will carry different P/E averages as well. Not only would we need to manually enter static values for each industry and for the market as a whole, we would also need to update these static values every now and then, as the averages change with time.

What’s more, different time periods (discussed below) may carry substantially different P/E averages. At the peak of the dot-com bubble in 2000, the total market P/E average was certainly higher than at the trough of the 2008 credit crisis. Similarly, the industry P/E average for US upstream oil and gas companies was undoubtedly higher at the beginning of 2014, with WTI crude oil trading at about $100 per barrel, than today, with WTI crude oil now closer to $50 per barrel.

To alleviate the issues described above, we use dynamic industry and total market averages calculated for us by Equities Lab. In this way, you might construct a screen that purchases equities currently trading below the current total market P/E average and sells equities above the current market P/E average. Similarly, you might construct a screen that purchases retailers below the industry P/E average for retailers and sells above the industry P/E average.

Getting into the topic of this blog post (the third point below), you may ultimately want to screen for the industry P/E average, over the course of the past 5 years (rather than simply using the current P/E average). Using this 5-year industry P/E average as your baseline, you could purchase below and sell above. This method would also work for the total market P/E average for the past 5 years.

**3). Dynamic Time Period Values** – The final option is the main topic of this blog post. Using the P/E average from a certain period of time. This time period P/E average could represent the total market, an industry, or a single equity. I mention the method one might use to determine the 5 year average of an industry or the total market above.

Specifically, you would stack the “Average Within” operator around the “Average Across” operator:

Average within(average across(P/E, retailers, true) within 1250 days. I use 250 days to represent 1 year here; each year has roughly 250 trading days. Thus, 1250 day is equivalent to about 1 year.

You’ve now got a broad foundation for the various methods used to screen metrics. I’d like to end by suggesting one method in particular: comparing the current metric values of a single equity to the time period metric average for that equity. Generally, I use some variation of the following formula:

[Current P/E] / [Average(P/E) within 1250 days] < 1

Or, in other words, the current P/E metric is less than the trailing 5 year P/E average. This framework gives you a dynamic metric average, based on an equity’s own trading history. In other words, the screen will look back at where an equity has traded on average over a certain period of time, buy above that average and sell below. Structuring the formula in this way allows me to screen below or above a simple 1-to-1 comparison (i.e. inserting 0.75 in place of 1 in the formula above would allow me to screen for equities with current P/E values at 75% of the 5 year P/E average).

If you think for a moment about this method, it’s exactly what you were trying to do above, only it’s tailored to each equity that runs through the screen specifically. Rather than looking at total market or industry averages, you’re looking at the equity’s own average, over a certain period of time.

Depending on the metrics you use, this method ought to screen for undervalue, or oversold, equities. Unless the fundamentals of a particular equity have quickly and drastically changed, that equity ought to revert back to average metric values. You’re comparing the current metric to a dynamic 5 year moving average, updated each time the screen rebalances.

When screening equities, using metrics (or ratios), you’ve got three main options (or some combination of the three options below):

What’s more, different time periods (discussed below) may carry substantially different P/E averages. At the peak of the dot-com bubble in 2000, the total market P/E average was certainly higher than at the trough of the 2008 credit crisis. Similarly, the industry P/E average for US upstream oil and gas companies was undoubtedly higher at the beginning of 2014, with WTI crude oil trading at about $100 per barrel, than today, with WTI crude oil now closer to $50 per barrel.

To alleviate the issues described above, we use dynamic industry and total market averages calculated for us by Equities Lab. In this way, you might construct a screen that purchases equities currently trading below the current total market P/E average and sells equities above the current market P/E average. Similarly, you might construct a screen that purchases retailers below the industry P/E average for retailers and sells above the industry P/E average.

Getting into the topic of this blog post (the third point below), you may ultimately want to screen for the industry P/E average, over the course of the past 5 years (rather than simply using the current P/E average). Using this 5-year industry P/E average as your baseline, you could purchase below and sell above. This method would also work for the total market P/E average for the past 5 years.

Specifically, you would stack the “Average Within” operator around the “Average Across” operator:

Average within(average across(P/E, retailers, true) within 1250 days. I use 250 days to represent 1 year here; each year has roughly 250 trading days. Thus, 1250 day is equivalent to about 1 year.

You’ve now got a broad foundation for the various methods used to screen metrics. I’d like to end by suggesting one method in particular: comparing the current metric values of a single equity to the time period metric average for that equity. Generally, I use some variation of the following formula:

[Current P/E] / [Average(P/E) within 1250 days] < 1

Or, in other words, the current P/E metric is less than the trailing 5 year P/E average. This framework gives you a dynamic metric average, based on an equity’s own trading history. In other words, the screen will look back at where an equity has traded on average over a certain period of time, buy above that average and sell below. Structuring the formula in this way allows me to screen below or above a simple 1-to-1 comparison (i.e. inserting 0.75 in place of 1 in the formula above would allow me to screen for equities with current P/E values at 75% of the 5 year P/E average).

If you think for a moment about this method, it’s exactly what you were trying to do above, only it’s tailored to each equity that runs through the screen specifically. Rather than looking at total market or industry averages, you’re looking at the equity’s own average, over a certain period of time.

Depending on the metrics you use, this method ought to screen for undervalue, or oversold, equities. Unless the fundamentals of a particular equity have quickly and drastically changed, that equity ought to revert back to average metric values. You’re comparing the current metric to a dynamic 5 year moving average, updated each time the screen rebalances.