Revenue Management for Meth Dealers in Flint (Fiction) | Hacker News Reader

Revenue Management for Meth Dealers in Flint (Fiction)

analyticsmadeskeezy.com

79 pointssdoering35 comments

Who said microeconomics was not useful :-)

For those who are not used to that, it's a very simple linear regression, followed by an estimated demand equation, which you then turn into profit by multiplying each side by prices.

To find the local optimal, you look for the spot where the 1st derivate cancels (and the second is negative, unless you have a quadratic equation like that)

It's an example so well though and so simple that it should be in economics textbook, so that students see how to make more money with a simple application of sound principles.

(if the excel spredsheet was turned into an online collection, using say OLS to calculate the estimators, Victor would have had a SAAS pricing software instead of entering the stuff manually on his laptop! ain't that cool ?)

You would indeed have a rough draft of price optimization software.

Also, you could use our[0] more refined SAAS Pricing Software that takes more factors like time into consideration.

I know the HN crowd will like to know that the entire price optimization system is exposed through a REST API with code samples on github.

[0]https://ventata.com

I find the idea of putting together what is nominally dry economic analysis with a colorful story about drug dealers an interesting technique for making it more fun to read.

I'm guessing that guys who run state wide drug businesses do not show spreadsheets of their business to blogging economists :-)

Sudhir Venkatesh approximately did that. It got a bunch of attention as a chapter in Freakonomics.

He also wrote a book about it:

http://www.sudhirvenkatesh.org/books/gang-leader-for-a-day

There are a lot of questions as to the accuracy of that book and what's been described. I can't find great links (most of it might be internal academic grumblings), but this nytimes article is a start.

https://www.nytimes.com/2012/12/02/nyregion/sudhir-venkatesh...

It's an interesting idea, and I'm sure there are a few people who'll follow it now there's an interesting narrative.

Unfortunately, it still has 'math' presented in a regular way among that narrative, and many people are going to glaze over and not learn much from it.

There are some other examples of presented information that people fail to absorb. Radio broadcast weather bulletins is one good one. People listen to the whole thing, but 5 minutes later they realise they've missed the weather for their region.

Still, it's a nice blog and they have some great examples of other stuff too.

I'd think that it would be advantageous for most of them to not keep spreadsheets.

I imagine the highest level of drug dealers use not just spreadsheets, but probably things like QuickBooks.

They typically use these things called "prestigious accounting firms" and "major international banks."

http://www.theglobeandmail.com/report-on-business/internatio...

... also some (but not all) of Robert’s Rules of Order:

http://www.youtube.com/watch?v=hGo5bxWy21g

Probably more likely than paper books. At least you can encrypt files on a disk.

Reminds me of the first Freakonomics book. They discussed the economics of slinging crack on the streets (moral of the story, if you aren't at the top, you are SOL).

"But it’s been confined to enterprises like airlines, car rental companies, and hotels implementing crap like Teradata and Oracle. This field has not been democratized, so it’s kindof gone unnoticed in the big data conversation."

This is precisely what we do.

Please check out, https://ventata.com/ if you are interested in a price optimization REST API.

Our average customer is a small to medium sized business selling goods through eCommerce.

You should partner with http://spreecommerce.com and sell an integration into the spree backend. They're good people, hit them up on irc.

Site has already been knocked over..

Here is the Google cache: http://webcache.googleusercontent.com/search?q=cache:rXK-EEu...

This is ok if costs are fixed, but if costs aren't fixed you should maximize profit, not revenue. Drugs, in particular, don't have fixed costs because they can be bought in greater quantities at lower per-unit cost as demand increases; there are strong economies of scale.

Author here. You're absolutely right. The reason why I chose to do a revenue optimization rather than margin is that most industries that use this actively on their websites (hotels, cruises, airlines) treat costs as fixed and maximize revenue. For most other businesses though, this isn't the case. Doesn't change the calculus though.

And sorry the shitty hostgator site went down. Cache: http://webcache.googleusercontent.com/search?q=cache:rXK-EEu...

I just want to add that sites like hotels, cruises and airlines have demand based models that consider time which makes finding the right price a bit more complex that standard a standard revenue curve.

We[0] solved this problem for ecommerce flash sale sites and event ticket sellers by creating a limited supply strategy that allows someone to give a starting quantity, current quantity and a start and end date to maximize revenue while trying to sell out all items.

[0] https://ventata.com

With meth, you don't really need to import it. I would think the cost of meth is more fixed than other drugs.

It would depend on the cost of acquiring the chemicals necessary to produce, as well as externalities. For example, in the short run you could increase your supply of chemicals by hijacking trucks, but that increases your economic costs substantially because of the increased risk of "heat" from law enforcement. You typically can't purchase massive quantities of the necessary chemicals easily.

In the long run, you may be able to purchase in sufficient quantity that you can import entire containers from overseas, with the knowledge that some will be intercepted, and you'll incur non-negligible cost for the smuggling. Also worth noting is that the economic costs associated with law enforcement attention will tend to rise (I'd bet non-linearly) as scale increases, just look at Pablo Escobar. In any case, I'd still bet that the per unit cost of production is much lower at scale.

The only way that I see that cost could be fixed, is that the economic cost of law enforcement activity exactly offsets reductions in production costs from scaling up.

It plays out at a slight remove:

http://www.pbs.org/wgbh/pages/frontline/meth/etc/cron.html

The part I think is pertinent here is that there is little illicit manufacture of ephedrine and pseudoephedrine (so the availability of meth varies with the availability of them).

This is why drug stores are required to alert law enforcement if they notice anyone trying to buy a large amount of Sudafed, or trying to buy a small amount very frequently. So now there are people whose jobs are to drive around to every drugstore in the town, buying small amounts of Sudafed at each one, so as not to raise any suspicion.

I saw a documentary, not so long ago from one city in Napoli, where it were the biggest drug dealing in Europe, the drug dealers do it all in the walls, most of them didn't know how to sum properly. They made so much money that kind of errors were normal.

There were two parts I didn't understand: One, how was the equation for revenue derived? Is it common knowledge that Revenue, as a function of price, is equivalent to your demand times your price? This wasn't obvious to me. Second, how is R^2 derived?

If you're in excel, R^2 values is just a couple check boxes away.

The actual math is: If you have a scatter of data (x, y), then you can get a measure of the variability of your data by taking the sum of squares of (y - avg(y)) for all your data points. Call this V

Further, if your doing say a simple linear regression, then for every x, you can have an f(x) which is suppose to be an 'estimate' of y. So then for all our given data points, we can find an error, which is basically the distance of y from the fitted line (y-f(x) basically). Then we can take the sum of squares of those errors get what is sometimes called our residual sum of squares. Call this W.

The R^2 value is defined as 1 - (R/W). So we can see that for very good fits of data, R^2 will be near 1, and for poor fits of data, R^2 will be near 0.

Edit: I guess to throw another bit at you. Often in microeconomics, you assume that as long as you can meet a demand at a certain price, then you will in fact sell however much is demanded at that price. So as an lemonade stand, if there is a demand for 100 cups of lemonade at 50 cents a cup, and you have 100 cups of lemonade and are also willing to sell at 50 cents a cup, then you WILL sell 100 cups. In which case, the revenue = price * demand kind of falls out by itself.

That revenue optimization curve looks almost exactly like the Laffer curve, used to optimize tax revenue: http://en.wikipedia.org/wiki/Laffer_curve#Justifications

Has someone made a HN bot that automatically caches pages before they reach the front page, and offers the page snapshot if the page goes offline via userscript?

Personally I would've paid for the coke in cash instead of making myself accomplice to a federal felony :)

great read anyway.

I'm not so sure giving a criminal advice is a felony. I'm no lawyer though. Seems like as long as he doesn't sell coke, he's in the clear.

You do realise this story is completely made up, using colourful narratives to illustrate the problems he's talking about?

http://analyticsmadeskeezy.com/what-the-hell-is-this-blog/

I did, and I was speaking about the character. :)

In these cases they tend to charge everyone who knew about it with conspiracy. There was a recent example in the form of that article about the guy who built stash compartments for cars.

even so he could still end up as a witness

And that would be a bummer. But possibly an opportunity to sneak some math into the courtroom?

from the sheet: price per g = $100, demand 52kg, revenue $520,000 $100521000 = $5,200,000 Perhaps he's selling 5.2kg/month, not 52kg?

Reply on news.ycombinator.com