How To Calculate A Screen's Performance
The question has come up recently from some readers: how are the Spell performances calculated? In particular, how are the "All-time" performance numbers calculated, and what do they mean?
In this article, we'll go over how the performance numbers are derived and why they may be different than what you would see by looking at a mutual fund or ETF tracking the market.
The Problem With Finding a Screen's Performance
A stock screen, like our Spells or a custom screen you can create from the Spell Caster, is not like a regular stock portfolio or a mutual fund.
In a regular stock portfolio or mutual fund, there is a finite number of total value you can use to easily measure performance from point A to point B. For example, if you wanted to compare your portfolio's perfomance against the market, you could take its value on January 1st and compare it to its current value, while doing the same thing for the S&P 500 and comparing their relative performance.
A stock screen is different. There is no finite total value to use, because there is no monetary weighting to each position. In order to create one, we have to assign a weight to each position in the screen.
Weighing Positions And Single-Day Performance
Obviously, the most fair thing to do is assign an equal weight to each stock in the screen. What that means is, if we were allocating money, each position would get the same amount of investment.
In that scenario, calculating a screen's performance on any given day is pretty simple: we simply average the daily gain/loss of each position in the screen.
Let's take a simple example. Our screen consists of just 3 popular Magic Recipe stocks: Apple (AAPL), Cisco (CSCO), and Gilead Sciences (GILD). Let's look at a single day's performance: Monday October 10, 2016.
Apple closing price 10/7 = $114.06. Closing price 10/10 = $116.05. Gain/loss = +1.74%.
Cisco closing price 10/7 = $31.47. Closing price 10/10 = $31.47. Gain/loss = 0.00%.
Gilead closing price 10/7 = $75.24. Closing price 10/10 = $75.49. Gain/loss = +0.33%.
Screen performance 10/10 = Average(1.74, 0, 0.33) = +0.69%.
To compare that to "the market", we look at the S&P 500's close on 10/7, which was 2,153.74, and on 10/10 which was 2,163.66, a gain of +0.46%.
Given all of that, we can say that our simple screen outperformed the market on 10/10 by (0.69 - 0.46) = +0.23%.
Calculating Performance Over A Span of Time
As we saw, calculating a screen's performance in a single day is relatively straightforward. This is what you see in the "Today" performance for the Spells at the top of the page.
But what about calculating a screen's performance over a time span of a week, a month, a year... all time?
There are a few difficulties here. The first thing we have to acknowledge is that a screen's constitution can easily change from day to day, as certain stocks fall out or enter into whatever criteria we have set for the screen. For example, let's say that on day 2, Gilead falls out of our screen but Biogen (BIIB) enters it. For day 2, we need to calculate the screen's performance with AAPL, CSCO, and BIIB.
The second difficulty is one of balancing. In our above example, at the end of day 1, our positions in AAPL, CSCO, and GILD would be out of balance, as both AAPL and GILD outperformed CSCO. We would no longer have the equal weighting we need to calculate the screen's performance. To achieve it, we would have to either sell all of our positions and re-buy at equal weight, or "trim and add" to get back to equal balance every day.
The third difficulty is one of cumulative returns. Let's illustrate this by building out the single day screen performance for 10/11:
Apple closing price 10/10 = $116.05. Closing price 10/11 = $116.30. Gain/loss = +0.22%.
Cisco closing price 10/10 = $31.47. Closing price 10/11 = $31.04. Gain/loss = -1.37%.
Biogen closing price 10/10 = $304.59. Closing price 10/11 = $298.30. Gain/loss = -2.07%.
Screen performance 10/11 = Average(0.22, -1.37, -2.07) = -1.07%
S&P closing 10/10 = 2,163.66. Closing price 10/11 = 2,136.73. Gain/loss = -1.24%
Screen relative performance = +0.17%
Now, you might think that calculating the 2-day performance is a simple as averaging both days together, but that's not the case because of cumulative returns.
Imagine we invested $10,000 in the screen's results. If we just averaged the two daily returns we would get Average(0.69, -1.07) = -0.19%. That would indicate that at the end of day 2, our value would be $9,981.
But remember, the return is actually cumulative and compound. We have to calculate the day 1 result and roll that into day 2. So it looks like this:
$10,000 * 1.0069 = $10,069 (end of day 1). Day 2 = $10,069 * (1 - 0.0107) = $9,961.26.
The actual value is about $20 less, because that daily loss on day 2 applies to a larger sum of money.
How Our Spell Performance Is Calculated
The balancing and cumulative problems we can account for. To calculate the "All-time" (or any period) performance, I assume a daily equal weighting of positions, while also calculating cumulative returns.
This is actually pretty simple in practice. To start with, I simply calculate the daily performance of each screen as above and store it. Then, to get performance over a period of time, I start with $10,000 and feed it in turn through each of the daily performances to get a final number.
To illustrate, let's see the performance of the Magic Recipe screen over the past week:
Starting capital: $10,000
10/10: +0.47% ($10,047)
10/11: -1.66% ($9,880.22)
10/12: +0.37% ($9,916.78)
10/13: -0.47% ($9,870.17)
10/14: -0.10% ($9,860.30)
Over the past week, the screen's performance would be calculated as ($9,860.30 - $10,000) / $10,000 = -1.40%. You can repeat this pattern over any period of time.
Drawbacks and Conclusion
I found this was the simplest and best way of doing long-term screen performance. In essence, it is very similar to how a mutual fund calculates its performance.
There are a few drawbacks. Daily rebalancing has associated costs involved, both tax-related and in commissions to brokers. The time commitment in doing a daily rebalance is also quite large and unrealistic.
That said, we are not really trying to measure a real-life portfolio strategy. Instead, we are measuring the efficacy of our screens and Spells' basic philosophies of buying stocks with particular fundamental characteristics. So far, we are seeing some substantial market-beating philosophies across all of our Spells! Try out our very low cost membership today to see a whole universe of attractive investing opportunities.
Joel Greenblatt and MagicFormulaInvesting.com are not associated in any way with this website. Neither Mr. Greenblatt or MagicFormulaInvesting.com endorse this website's investment opinions, strategy, or products. Investment recommendations on this website are not chosen by Mr. Greenblatt, nor are they based on Mr. Greenblatt's proprietary investment model, and are not chosen by MagicFormulaInvesting.com. Magic Formula® is a registered trademark of MagicFormulaInvesting.com, which has no connection to this website. The information on this website is for informational purposes only and solely represents the views and opinions of the author. No warranty is provided or implied as to the accuracy, completeness, or timeliness of this information. This information may not be construed as investment advice of any kind, nor can it be relied upon as the basis for stock trades. DON'T RELY SOLELY ON THIS WEBSITE'S INFORMATION OR STATISTICS! Please do your own research before buying. Alexander Online Properties LLC, the proprietor of this website, is not responsible in any way for losses or damages resulting from the use of this information. Alexander Online Properties LLC is not a registered investment advisor. All logos are trademarked properties of their respective companies.
© 2008-2017 Alexander Online Properties LLC