Silly chart of the day, data-fitting edition

By Felix Salmon
August 29, 2009
Paul Kedrosky finds this chart in a Bloomberg story: it's the kind of thing which really reinforces one's belief in the wonders of data-fitting.

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

Paul Kedrosky finds this chart in a Bloomberg story: it’s the kind of thing which really reinforces one’s belief in the wonders of data-fitting.


The story isn’t actually particularly clear on exactly what the graph is showing, and specifically what “adjusted for currencies” means:

The Nikkei doubled between October 1998 and April 2000 in dollar terms, as the chart illustrates. The S&P 500 has risen 34 percent since March when the Dollar Index, a measure of the dollar against currencies in six major U.S. trading partners, is factored in.

So it seems that the BofA analysts who came up with this chart first converted the Nikkei to dollars, only to then convert the S&P 500, which was in dollars all along, out of dollars. Hm. And they chose pretty random start points: what makes 1980 in Japan analagous to 1990 in the US?

But the most breathtaking claim of all is right there in the Bloomberg headline: “S&P 500 May Surge 40% in Duplication of Japan”.

In other words, the people who came up with this chart are not saying that the US is going basically nowhere over the long term. Instead, they’re saying that there could be a big further rally in the S&P 500 over the short term. On the basis of the fact that the Nikkei rose sharply, in dollar terms, at the tail end of the last decade.

I feel quite safe in saying that of all the years to compare 2009-2010 to, 1999-2000 are probably not the most useful. And looking at what Japanese stocks did ten years ago will tell us absolutely nothing whatsoever about what US stocks are going to do now. No matter how many clever charts you come up with, or ridiculous justifications for the conclusions of your silly exercise in data-fitting:

“Even in economies overcoming credit booms, rallies can be powerful and last much longer than you think,” Bank of America’s Sadiq Currimbhoy, Arik Reiss and Jacky Tang wrote.

Well yes, if you’re caught up in the exuberant tail-end of a global stock-market bubble, maybe. But that spike in the yellow line that BofA thinks we might repeat on our orange line? Was basically a function of excesses like Softbank having a market capitalization of $200 billion. If you want to bet on that kind of thing happening in the US over the next year or two, feel free. But I’ll happily bet against you.

Update: Henry Blodget has the original Merrill report; it seems to admit how tortured the numbers it’s using are, and doesn’t go so far as to actually predict a 40% rally from these levels.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

Yeah, you see something like that, on a premier business news service, and you just have to wonder how the economy ever actually worked at all. I tell my wife stories about how boneheaded business writers and analysts can be and she doesn’t believe me.

Posted by Bob_in_MA | Report as abusive

with enough finagling you can make any market chart mean anything you want it to mean or predict anything you want it to predict…

“In other words, the people who came up with this chart are not saying that the US is going basically nowhere over the long term.”

Pity. All of my investing years have been in the red-hot bubbles of the last 15 years. I could really go for a 10-15 year bear market–long enough for the punters to get scared out of equities.

I dribble a little money into NIKKEI index stocks–less than I should, probably–because a market that’s had the stuffing beat out of it for years in a stable, advanced, industrialized country might just be one heck of a good buy if you let it run long enough. All those cute dancing robots Honda is working on? Just wait until they’re doing our shopping and changing our bedpans in our old age.

Posted by Craig | Report as abusive

I’ve found that Bloomberg has been doing that a lot lately — taking some statistic and then extrapolating a market direction prediction. They’ve done it with crude oil implied volatility and VIX, basically saying that the high implied volatility meant an implosion. Only problem was, that they said this back in June, just BEFORE both markets went on a bull run. Stats are stats, and cycles are cycles, but they’re not so infallible that you can make a prediction of a 40% gain.

And for what it’s worth, Bloomberg isn’t alone.

Posted by Don | Report as abusive

The real ridiculousness of this chart is how the creator expects the coincidental graph markings (10 years apart and for different economies!) to accurately predict future trends. Why not then just go around and find equally matching charts and use them? Number of Twitter users in Brazil. Holiday sales figures for Krispy Kreme Donuts. Willie Nelson’s popularity over the length of his career. Who knows? Maybe this guy is onto something!


Posted by woofer50 | Report as abusive