Wall Street Can’t Count | Hacker News Reader

Wall Street Can’t Count

cringely.com

35 pointsastrec15 comments

This flaming of bankers is emotion fueled not logic fueled. IMO, it's not helpful. From the article:

So it’s a typo: no big deal, right? Yeah, but what a typo! It got past Bloomberg and JP Morgan and pretty much all of Wall Street before someone said, Hey, this makes no sense! ... And who was that someone? Me! A nobody

I imagine that the vast majority of bankers who saw this chart realized this error very quickly. I only have 2 data points to support that, but both the bankers that I know spotted this instantly, on Jan 21 when they first saw the chart.

I imagine that the vast majority of bankers who saw this chart realized this error very quickly.

Are you sure?

In my own entirely subjective and unscientific experience, a minority of people on wall street are very good at math.

And they are usually not in positions of authority, some times, but not in general.

I would like to suggest that showing changes in value as a square of the difference might be useful because they multiply over time.

A stock can drop of 5% that occurs 13 times is less important than a single 50% drop. 40 drops of 5% are less important than a single 90% drop. DryShips for example down ~95% from last year that's the equivalent of ~50 drips of 5%.

I suspect most people in finance understands the math, but at the instinctive level most people just don't feel it. So what's the correct way to display information like that visually? x, x log x, x^2

Behold!

http://ii.gs/fin-inst-caps-2007-2009.png

I just copied the data out of the sheet. Mistakes are mine.

A little less alarming than the circles, for sure.

Indeed. "Cringely can't chart" would have been a better title.

He visually squared the values by using quantities as the radius of the circles.

That's a good effort.

A vertical bar chart with the two measures next to each other is a better representation because you are trying to compare them.

If you were to show something cumulative, say contribution of sales of groceries, electronics and clothes towards your daily sales, then, the chart you used is appropriate.

You know, I intended to chart blue = new, red = (old - new), so the red bar would be the comparison, the total bar width would be the old value, and the blue would be the new bar (compare in a single line), but I didn't check my data, thus making me guilty of exactly the same sin that spawned the article. :) Updated below.

Bahhhhh! Thanks for the catch!

http://ii.gs/fin-inst-caps-2007-2009-fix1.png

This might be a better visualisation. http://img145.imageshack.us/img145/5507/picture39fl7.png

Note:I used horizontal bars instead of vertical as I suggested in my other comment to remove the redundant labeling.

Mine is bi^Better!

http://www.alejosanchez.com/projects/market-capitalization

It's unfinished but it already draws with HTML canvas.

Oh, and I did it when it this story was on ;)

I stopped at the glueing of info from Yahoo! Finance (got busy with other stuff) but it's all right there. The idea is to let you pick the stock symbol and get a live graph.

Errors like this a embarrassingly common in all sorts of media. Something it looks like genuine errors - charts made by someone who didn't take enough math in school, but more often it's someone who wants to highlight differences of some kind and skews the charts accordingly.

One of the favourite ways of doing this is to make a line chart of two values, maybe mortage rates from two banks, and not make the chart's Y axis based on zero (or put a zero there and make a little squiggle on the line). A minute difference is made to look huge, and it annoys the hell out of me.

I expect things like this in marketing, but what -really- annoys me is when public television does things like this on the news. They really should know better...

http://en.wikipedia.org/wiki/Chartjunk

aka USA Today Chart amongst my former colleagues.

Interestingly, I find this term is often instantly meaningful to those who have never before heard it..

Lies, Damn lies, and Statistics

This is bs. The first version was leaked and it was wrong, but it wasn't released properly.

Cringely is making this story bigger than it was. http://ftalphaville.ft.com/blog/2009/01/22/51558/that-jp-mor...

"Apparently, the version we got was a “draft”. Oops."

s/Wall St. can't count/Cringely can't research like a decent blogger/

What the hell?!

This is what counts as front-page material? Here is a question for cringely: maybe the chart is only designed to show subjective differences and the people who made the chart concluded the scaling by diameter instead of area was a better option?

And based on this flimsy evidence you denounce the entire fscking industry? Does that make any logical sense to anybody who has spent more than a nanosecond giving any serious thought to this?

How arrogant can you get.

Reply on news.ycombinator.com