Deceiving with Data

Being in the finance industry for almost 15 years has taught me a number of things. One of which is to question just about everything when it comes to data presentations. 

Using data to create a narrative is easy - and it’s also convincing. 

Recently I saw an advertisement on TV saying how gold has outperformed the stock market over the past 20 years. After initially raising my figurate eyebrows, I figured this made sense. Your starting point is the dot-com tech crash, and your ending point is the Coronavirus. But I wanted to dig in deeper. 

Sure enough, the advertisement wasn’t lying - the Bloomberg Gold Subindex Total Return outperformed the total stock market over the past 20 years (9.48% vs 6.67%) ending August 31, 2020. 

But you know what wasn’t said in that commercial? This:
A non-cherry picked, longer-time frame dataset. (1975 is as far back as data goes for the gold index. CRSP 1-10 deciles represents the entire stock market.) 

So although the advertisement wasn’t lying, it was deceiving. 

My point today isn’t to pick on goldbugs. It’s to show how easy it is to deceive with numbers given a large enough dataset and a predetermined narrative. 

While I can’t address how to counter every data manipulation out there, I do want to address the “X-to-date” one in this gold example.

Anytime you see a 3 month to date, a year to date, a 3, 5, 10 or even 20 year to date figure you should realize that recency bias is dangerously at play. What has happened the most recently in those time frame datasets will warp the total picture. 

A better way is to use something called “rolling period” returns. It takes a larger dataset and breaks it into smaller components that repeat incrementally. For example, a 10 year to date time period covers 10 years. Bear with me here, but that 10 year period also contains 10 individual 1 year periods (and even more so if you stagger the starting points, such as monthly starting points for the proceeding 12 months). 

Examining those unique 1 year time periods may give you a better idea what to expect for future 1 year time periods. 

Again, what has happened most recently really impacts any “X-to-date” calculation and can be used to deceive. And for that point, so too does the start date. 

If we want to stick with the arbitrary 20 year figure from above, let’s look at rolling 20 year time periods from 1975. If we keep it simple and consistent by just measuring August-August of each year, we have 26 separate 20 year rolling time periods. 

Care to guess how many times in those 20 year time periods gold outperformed stocks? 


See? Deceiving with numbers is easy. 

Interesting Resources

Brian Boycan
"Wisdom without humility is hubris." 

Use This Gospel Live Performance
View time 8 mins | Kanye West, Kenny G, & Clipse (YouTube)
I don't believe you have you be a closet hip hop fan like me to appreciate this. When I think of heaven, this is one of the concepts that comes to mind. Stories of redemption, a blending of musical genres, multiracial congregations, all worshipping God. I've probably watched or listened to this 20 times this week without exaggeration. (If you want to dig deeper into its surprisingly deep lyrical meaning, this is fascinating.) 
Wikipedia - whatever you want to read | Wikipedia 
I feel we need to create a better term in English to distinguish relaxed time versus wasted time. I'm still fuzzy on what the term truly means, but the idea of discerning otium negotiosum (free time to do what one wanted) and otium otiosum (idle wasteless free time) is fascinating. In our busy culture, being anything other than busy is frowned upon. Leisure has a relatively negative connotation to it. Something I'm digger deeper into - which may be its own form of otium. 
I'll be honest - I was back and forth sending out this week's content. Even showing the total return of gold vs the stock market is some element of data manipulation. Although we want to view data as the ultimate measure of purity, there's so much of it that we can twist and distort it to fit our pre-existing narrative.

My point remains simply this: be cautious. Deception is difficult enough to discern. Deception coupled with data is even more difficult. 


Copyright © 2020, Fident Financial, LLC. All rights reserved.

Phone: 717-208-2235

Not for you?  No worries - update your preferences or unsubscribe below. 
I only want to be in your inbox if you want me to be.
No hard feelings. I promise. 

This week's extra disclosures:
Current performance may be higher or lower than the performance shown.

Obviously, past performance is no guarantee of future results.

Views are my own. 

Information source data from Dimensional Fund Advisors Returns Website. Accessed October 2020.
Data Sources: CRSP 1-10 Index (market), 1/1926 - 8/2020 - Source: CRSP, total returns in USD$. Bloomberg Gold Subindex Total Return, 2/1975 - 8 /2020 : Source: Bloomberg Gold Subindex Total Return
Fident Financial, LLC (FFL) is a registered investment advisor offering services in the state of PA and in other jurisdictions where exempted.  Opinions expressed in this email are solely those of FFL, unless otherwise specifically cited.  Material presented is believed to be from reliable sources but no representations are made by FFL as to another parties' information accuracy or completeness.

This email was sent to <<Email Address>>
why did I get this?    unsubscribe from this list    update subscription preferences
Fident Financial · 1 Thunder Ln · Lancaster, PA 17602 · USA

Email Marketing Powered by Mailchimp