The Next 10, an introduction, part 2

In part 1 we looked at the list of ranked companies, how we have a set of top 10 which we think are too risky and the next 10 which we think are just right.

We presented various navigation options on the list and drilled into viewing individual companies using Netflix as an example. This builds up a fundamental understanding for how to use the AgoraOpus portfolio construction website.

In this blog post we’re going to look at the portfolios page, and see how the list from part 1 produces different portfolios and how these perform relative to our expectations.

Viewing portfolios

First of all, the default view is to see our benchmark and top 11-20 portfolios for our full instrument universe. But we also break this down into sub-portfolios based on the 50 biggest companies by market capitalization in addition to portfolios for various company sectors if you drill into the specialized portfolios. The benchmark consists of all members in the selected instrument universe.

For all of these universe subsets we have portfolios for the top 10, top 11-20, and so on, were we expect the top 11-20, or next 10, to perform better than the top 10. When we say “perform better” we’re talking about the longer term trend, and we’re measuring it based on alpha. You can naturally always find a small period somewhere when this didn’t hold. But over time we expect the next 10 to perform better than the top 10.

The portfolio performance chart is split into two. The top section show the two selected portfolios with their value shown over time, and the bottom section show the relative performance between them. If what we say about the next 10 performing better than the top 10 holds, then the relative performance should steadily go up over time.

One important thing to keep in mind about relative performance: It’s easy to, for example, use leverage or high beta stocks to make the relative performance go up compared to an unleveraged or low beta position, assuming the market goes up in general. Put differently, you’re making your active portfolio more volatile than the benchmark. This is the wrong type of relative performance, because what you’re simply doing is adding risk without adding value.

This is why we also include statistics for alpha and beta. Below the chart the current alpha and beta values are shown, calculated from the data you currently have in your view. Zoom in, change time range, or pick different portfolios, and these are recalculated.

There are many statistics used when calculating portfolio performance, and calculating alpha and beta is one of them. In simplified terms you can assume beta shows you your leverage against the benchmark, where a beta close to 1 means we have the same level of risk/volatility exposure, taking correlation into consideration. Alpha shows the value added returns on the active portfolio vs the benchmark.

Producing good and positive alpha is critical for any investment strategy.

The mechanics of building a portfolio

There are many details to consider when building a portfolio. Let’s get some of the more mechanical aspects we do listed and described below:


  • Dividends are included
  • We always use log returns
  • We always trade after the fact
  • We use continuous re-balancing
  • The portfolios are unleveraged long only
  • We use average daily prices as trading price
  • All instruments included in the portfolio have equal weighting
  • The member sets are updated on a weekly basis, when we release new rank values

Dividends are included

If any company in a portfolio pay dividends, this is reinvested.

We always use log returns

Log returns are symmetrical and mathematically easier to work with. There are other type of returns, but we don’t use them here.

Make sure you’re comparing apples to apples.

Always trading after the fact

Of course we trade after the fact!

It means that when we produce a new set of portfolio members, over the weekend, for data available at the close of Friday, we act on that information on Monday. We don’t magically trade on Friday prices or something impossible like that.

Continuous re-balancing

We keep the relationship between the portfolio constituents fixed, so when they individually increase or decrease in value we re-balance the portfolio. This is typically less of an issue during short time windows, like up to 7 days in our case, so we expect any difference between doing this or not to be small.

Long only portfolios with no leverage

Whenever we build a portfolio on our portfolios page, all portfolio constituents have positive weights. Put differently, we’re doing long only portfolios. When you’re comparing two portfolios, the active with the benchmark portfolio, you’re looking at the difference in return between these two long only portfolios.

This might be difficult to initially get your head around if you’re used to having short positions, but think of it like so: Put your short positions in a portfolio, as long positions. You can then compare your long only portfolio of your short positions against other portfolios as you would any other portfolio. If you’re any good at picking short positions, you’d expect your portfolio to show this as negative alpha.

No leverage is used when building portfolios. Again, this is to compare apples against apples, as you can easily fake “good performance” with leverage. Naturally that doesn’t mean you can’t use leverage to increase your exposure for your own personal sake, but do not use it when comparing a portfolio against another.

Average daily prices

To include for all sort of complications around trading large positions we use average daily prices when simulating a trade. That means we do not assume we would be able to fill a trade based on any one price point. This is also known as time-weighted average price.

This is important with regards to scalability: If you’re only trading tiny positions, assuming a quick fill is fine. But you can’t continue to assume this if you scale up. It’s therefore only fair to assume average pricing when measuring performance.

And if your strategy doesn’t work if not using spot prices from a single point in time, throw it in the bin.

Equal weighting of portfolio members

We pick the portfolio members, but we don’t assume we have any further allocation information available. All portfolio members are given an equal weight compared to the other constituents.

For for a portfolio with 10 members, you allocate 10% each.

Sticking to the weekly member set

As we update the ranking over the weekend, we base our portfolios on that. Whatever happens in the coming week does not change the member set.


This blog post was written by Christian, the main portfolio curator here at AgoraOpus. With a background from FinTech, he holds a MSc in Quantitative Finance and a BSc in Computer Science and Industrial Automation.

Subscribe to our mailing list



All information provided "as is" for informational purposes only, please see "Help" section for further details. Copyright 2012 - 2019.