« Frank Deford on Greatness: Feder, Brady, and Tiger | Main | First Full MacBook Air Review »
Latest Stories
- Excel Wankers and Recession Averages
- Sorry, New York is Closed. Check Back Later.
- Catching Falling 2009 Earnings Estimate Knife
- Survivorship Bias in Global Markets
- Talking Positions on a Lazy-ish Retirement Portfolio
January 23, 2008
Dow Bouncebacks: A Trip Through the Data
Lots of people are rightly remarking on the savage Dow bounce-back today. After all, at one point in the day the Dow was down 200-plus points, only to turn the corner and close the day up 300-plus points. That's enough to require a decent whiplash lawyer.
Anyway, inquiring minds want to know: Where does today's snapback rally rank in the great scheme of such things in Dow history back to 1931? To answer the question, I took the last 77 years of trading history, filtered for days that swung at least 1% negative, but then ended the day positive. That produced a list of 962 trading days out of a possible 19,914 days in the entire period.
Preamble complete, let's get to the data. First, measured in points terms, today's 625-point Dow swing -- from the day's low to the close -- was #2 all time. Only a 701-point swing in the summer of 2002 beat today's market change of heart.
Here is the entire top ten in points terms:
Interesting, of course, but the trouble with measuring such things in points terms is it doesn't normalize things against changing Dow levels. There are many ways to do that, but the quick and dirty is to take the one-day swing as a percentage of the prior day's close. That gives you a sense of whether the point swing mattered in the context of the Dow level at the time.
So, where did today rank in those adjusted percentage terms? Measured that way yesterday came in 30th all time based on a 5.2% swing. The top one-day bounce-back of all time was October 20, 1987 (ring a bell anyone?) which delivered a 10.1% carom off the cushions.
Here is a table of the top ten swings in percentage terms. (Again, this is only for days that swung negative by at least 1% during the day, before closing on the positive side.)
Some fairly neck-snapping stuff in there, with two double-digit intra-day swings on Dow days that fell by more than 1% at some point during the day. Also notice that seven of the top ten percentage-based swing days were between 1929 and 1937.
Finally, and this is really riding the data way out there, I had a quick look at what the market did over the next day, week, month, or quarter after one of these big swings. The answer: In all four cases the market was, on average, up marginally, but nothing to write home about.
Sphere It
| 
Digg it
| 
Bookmark it
| 
Stumble it
| 
Facebook it
Of course. And you could reasonably add that the likelihood of getting a refereed academic journal article out of the above is nil. But so what? Some historical context is better than none at all.
Good stuff Paul!
On a percentage basis, yesterday did not even make the top 10.
BTW, what did you use as the data source and filter?
Now do the same thing for the Nasdaq and the SPX and you have something . . .
"I took the last 77 years of trading history, filtered for days that swung at least 1% negative, but then ended the day positive."
Can you tell me what tools you use to crunch the data?
Thanks
Paul,
Interesting stuff for historic context.
You are missing an "r" in the title. Big DOW sings are indeed rough stuff :)
Gitawa, this is crunching that can easily be done in Excel. The format of his output even looks like it.
Boy it bugs me when dramatic swings in the stock market are compared to the past by point bases. But news reporting is about drama and attracting eyeballs, so "625-point swing!" is going to be the headline, not "5.2% bounce." It's bad enough that a so much stock market action is emotion and mood based rather than real hard facts; do we really need irresponsible sensationalism in reporting it as well?
A few people have asked where I did the crunching. It was all Excel, as someone else surmised. This sort of thing is straightforward and speedy to do in Excel; it's just surprising how rarely people bother trying.
big swings appear to be bear mkt rallies?
You know Paul, the Dow is a pretty much statistical quicksand for historical comparisons.
a) it's price-weighted
b) it has a small number of components which change frequently
Historical extreme calculations on top of it are data porn for data porns sake. Not that theres anything wrong with that.