Gregory Taketa | Gregarious Consulting

Gregory’s Blog

Gregory Taketa

Gregory Taketa

Gregory Taketa is the Data Decanter, serving refined, well-breathed data analysis while keeping out the sediments.

View Full Profile →

Enter your email address to follow this blog and receive notifications of new posts by email.

Archives

Random Finance, #1

Test how you would save money:

You have a $10,000 debt to be paid in 12 monthly installments at 3.5% APR.

A credit card company has recently offered you a balance transfer.  You can write yourself up to $10,000 cash if you assume a 3% upfront fee, which will be added to your principal.  The APR is 0% for 12 months, and you will pay all the principal within a year.

Do you take the balance transfer offer?

Some of you might be doing back-of-the-envelope calculations, such as “3.5% APR on $10,000 is like paying $350 a year.  That’s bigger than the $300 fee to do the balance transfer with no interest payments.”

This estimate is misleading because interest actually accrues on progressively smaller balances.  Yes, the first month accrues on the $10,000 beginning balance. The next month’s beginning balance is smaller ($9,179.95 to be more precise), so the interest expense also gets smaller.

It turns out that refusing the balance transfer yields an $849.22 monthly payment versus the $858.33 monthly payment with the balance transfer ($109.40 annual difference).  If the form of your debt is tax-deductible (e.g. student loan, home mortgage), we fall even more in favor of the original debt.

If the after-tax APR of the debt exceeds 5.49%, then the balance transfer begins to save money.

How many of you picked the balance transfer?  Would the $109.40 difference have been enough to justify 5 minutes of financial consulting (that’s equivalent to $1,312.85 per hour)?

1st Children’s Book, “Mommy, Where Is My Tablet?”

For many years, I have wanted to be an author, and I am excited to publish my first children’s book, “Mommy, Where Is My Tablet” on Amazon for the Kindle Apps.

00 Front Cover

Click the picture to get your copy on Amazon!

When a daughter wants the tablet computer (again), what is a mother to do?

  • Short
  • Funny

To get updates on my stories, including promotions and new releases, see my new children’s website, GregGoof.com, and follow @GregGoof on Twitter.

Theranos Doesn’t Need to “Suck” Less Than Others

Theranos is currently a private healthcare company famous for its claim to conduct over 200 lab tests with about a single drop of blood (versus the 3-4 vials traditional testers use).  There has been a lot of incredulity, especially noted in this Business Insider article.  To the skeptics’ merit, there has been a dearth of peer reviews on the effectiveness of the proprietary technology.

Taking off the Scientist Hat

What if we stopped thinking like scientists and more like statisticians about Theranos?

For example, did you know that tests for terrorists and corporate drug tests are rife with false positives (accusing the innocents)?  They’re notoriously high, but contrary to your intuition, the testing equipment and methods are really, really accurate (arguably 95%+).

How could such accurate tests end up with such lousy results?  Because very few people are terrorists or drug-abusing employees in the first place (well, the latter is relative…).  This is one of the basics of Bayesian inferences.

Arbitrage Opportunity

Back to Theranos.  It seems that scientists expect Theranos, which charges less than half their rivals, to meet or beat current standards.  Did Theranos ever claim to be as accurate?  And, more importantly, does it need to be?

Any seasoned statistician knows that an alpha (Type-I error) of 1% and a alpha of 5% could be orders of magnitude in terms of sampling costs but not in terms of information benefit (and the reverse could also be true).

Traditional blood testing, which started before the 1940s, had to meet highly accurate standards because the populations for blood testing were very small.  Compare that to a much larger population today.  We have such a wide range of historical references that any 1 test could afford slightly less accuracy than the traditional counterparts.  In the worst case, Theranos could easily conduct 2 blood tests at very different times for still cheaper than the traditional method, and the law of averages and large numbers would play out over time.

In other words, there is demographic arbitrage.

This might prove to synergize with their offering to post results to a database for your quick reference.  It’s really also a way to keep an ongoing profile of a patient to manage long-term averages (my regular blood lab does NOT keep my records; they keep forgetting who I am every year!).

Ahead of the Game

It also makes sense to serves convenient masses through Walgreens and the like because it allows large sampling to occur quickly.

Most curious is how they’ve so transparently listed prices on their website.  The economist in me says they’ve been doing data analysis on testing demand and the manufacturing costs of their devices.  The statistician in me says they’ve known their accuracy and error rates for some time and have likely bought some insurance in case of error to readily know an effective price.  We could infer that incidences are low, or insurance is cheap (who am I kidding?  It must be the former).

The Taketa Takeaway

If it turns out that Theranos has been buying up insurance for the last few years, then it is likely that its business model is about cost savings through risk management (lowering accuracy rates a little for more than proportionately higher sales potential through a cheaper alternative).  The risk gets mitigated with insurance but more importantly the simple fact that a wide population and wealth of data are on their side to reduce errors.

The breakthrough of Theranos might not really be its technology but its approach to statistics, risk, and distribution.

Greg owns no stock in Theranos at the time of this writing (would’ve been nice…) and has no insider information of any sort on the company (would’ve been nice, too).