After sifting through FinCen’s SAR STATS data, I couldn’t help but notice the huge uptick in Suspicious Activity Report (SAR) activity occurring in San Francisco County. I mulled over the data several times through the lens of different charts to see what could be found. It wasn’t until I compared Bitcoin’s price data to FinCen’s SAR STATS data that I made the conclusion that the rapid increase in price of Bitcoin was, in my opinion, causing the huge increase in SAR activity.

The goal of this dashboard is to compare SAR activity occurring during a period when bitcoin price is relativity stable vs a period when the price of bitcoin is increasing. For the period of stable prices, I chose September 2019 to February 2020 which is represented by the dialogue bubble on the lower left. For the period of increasing prices, I chose December2020 to May 2021 which is represented by the dialogue bubble on the upper right.

This dashboard analyzes attributes listed on cryptocurrency exchange SARs such as AML typologies, financial instrument usage and product usage versus actual cryptocurrency exchange SAR counts.  

The design of this dashboard is in the style of Neobrutalism which is characterized by bold black outlines especially in areas of shadows as well vibrant colors in the background and throughout.

FinCen’s SAR STATS data doesn’t specifically tell us what suspicious activity report data belongs to cryptocurrency exchanges. Due to this, my best attempt at isolating data from cryptocurrency exchanges was to filter for Money Service Businesses (MSBs) entities and of course to filter out data for San Francisco County.  San Francisco, CA is home to some of the largest cryptocurrency exchanges in the United States.

Because FinCen's SAR STATS doesn’t give us specific cryptocurrency data, MSBs such as local liquor stores that offer check cashing services and foreign remittance money transfers are mixed in with the data. My analysis isn’t perfect because the data isn’t perfect but it’s the best data that FinCen currently gives us.  Despite this, there are some valuable insights to be gained in the above Tableau dashboard.

Comparing San Francisco County to other counties was a challenge I came across early in my analysis. The biggest problem is that the cryptocurrency exchange industry is largely centered in San Francisco County and no other county comes close in comparison.

At first, I was comparing San Francisco County against all other counties in the United States, but I eventually didn’t think that was the best comparison because a good percentage of counties in the US have very little SAR activity.

In the end, I decided to use county population as the metric to compare San Francisco County to other counties. Once again, this is not the perfect metric but better than most in my opinion. There is a direct correlation between county population and total count of AML activity.

For my analysis, I used the five counties directly above and below (10 total counties) San Francisco County in population rank according to the US Census 2022 estimates. In 2022, the US Census estimates that San Francisco County had a population of 808,437, making it #78 most populated county in the United States. The following ten counties are in the comparative analysis:

#73 Baltimore County, Maryland (Baltimore, MD adjacent)
#74 Snohomish County, Washington (Seattle, WA adjacent)
#75 Ventura County, California (Los Angeles, CA adjacent)
#76 Hamilton County, Ohio (Cincinnati, OH)
#77 Lee County, Florida (Fort Myers, FL)
#79 Essex County, Massachusetts (Boston, MA adjacent)
#80 Oklahoma County, Oklahoma (Oklahoma City, OK)
#81 Multnomah County, Oregon (Portland, OR)
#82 San Joaquin County, California (Stockton, CA)
#83 Polk County, Florida (Tampa, FL adjacent)

In my analysis, the ten comparison counties above are listed as “Other Counties” in both Chart 1 and Chart 3. To show a greater contrast, I took the total sum of AML activity from the “Other Counties” instead of the average when comparing to San Francisco County.

Chart 1 displays a line chart comparing the increase in all AML typologies count when the price of bitcoin rises versus stable prices. An increase in AML typology count is usually associated with an increase in Suspicious Activity Reports.  The bolded numbers on the vertical axis are very revealing to the large quantity of AML activity taking place during an increase in bitcoin price.‍

Chart 2 displays a stacked bar chart showing the financial instrument usage by percentage for the top 5 AML typologies (5 vertical bars)by total count. This chart focuses solely on San Francisco County data.

Chart 3 displays a comparison bar chart showing the Top 3products used in conjunction with the financial instruments in Chart 2. Like Chart 1, I used the total sum of AML activity to show a greater contrast versus calculating the average count of AML activity for the “Other Counties”. “Prepaid Access” refers to a prepaid access card where funds are loaded and must be spent according to the terms and conditions of the card.

The large increase in the price of bitcoin that occurred during late 2019/ early 2020 was unprecedented at the time.  We may not see that level of SAR activity again but only time will tell. I wouldn’t be surprised if a lot of the SARs filed during this period of time at cryptocurrency exchanges were more “defensive” in nature than anything else. It will be interesting to see the SAR data published by FinCen concerning the current large increase in Bitcoin we are currently experiencing in early 2024.

In this dashboard, I learned a lot by experimenting with design. I really like the color scheme inspired by the Neobrutalism style of design.  As far as design in Tableau, I had to figure out how to work around the dialogue bubbles, so I broke down the bitcoin price bar chart into multiple containers. If you look closely, you can see where the bar chart ends and begins with new containers.  I often start with a design in Figma and then proceed from there.

I also learned how to swap sheets in Tableau with a parameter. This isn’t very difficult, but it was the first time I implemented something like this in one of my dashboards.  The reason why I implemented the sheet swapping feature was to organize the dashboard in a manner to save space and not make the dashboard appear cluttered with more than two dialogue bubbles.

This section of this article was written sometime after I finished building this dashboard, so I can’t help to think how I would design this dashboard differently today.  If I were to design this dashboard today, I might include the title of the chart i.e. “Total Count of all AML Typologies” inside the dialogue bubble in small sized lettering. In doing this, I would lose some of the minimalist look, but I think it might help make more clear what content is inside the dialogue bubble.