Hindenburg Omen: Doom?
July 7, 2008 1:41 AM Subscribe
Stock Market on Crash Watch due to a Hindenburg Omen.
More info on a Hindenburg Omen.
Even More info
And more and more information in a PDF.
(Editorial - No idea if this is legit or not, but I thought it was interesting. It sort of reminds me of Baseball statistic mining.)
More info on a Hindenburg Omen.
Even More info
And more and more information in a PDF.
(Editorial - No idea if this is legit or not, but I thought it was interesting. It sort of reminds me of Baseball statistic mining.)
No worries, mon. They don't let the stock market crash anymore.
posted by Avenger at 1:51 AM on July 7, 2008
posted by Avenger at 1:51 AM on July 7, 2008
A rare signal or event in the Shannon’s information theories (the backbone of the modern day digital communications) is considered to contain a higher amount of information.
What's the false discovery rate?
posted by Blazecock Pileon at 2:00 AM on July 7, 2008
What's the false discovery rate?
posted by Blazecock Pileon at 2:00 AM on July 7, 2008
Feels sort of spammy? Do tell?
Pump and dump spam? Pepsi-financial services?
It might be dry, incorrect, crap, or just boring, but spammy?
posted by Lord_Pall at 2:11 AM on July 7, 2008
Pump and dump spam? Pepsi-financial services?
It might be dry, incorrect, crap, or just boring, but spammy?
posted by Lord_Pall at 2:11 AM on July 7, 2008
I should clarify my previous comment.
"The omen has appeared before all of the stock market crashes, or panic events, of the past 21 years", speaking about 1985 to 2006. Having a signal that can generate sharp market declines is appealing to all active traders, but this signal is not as common as most traders would hope. According to McHugh, the omen only created a signal on 160 separate days, or 3.2% of the approximate 5,000 days that he studied.
So I would ask, how many false positives are there from thisalgorithm Omen: from how many of the 160 separate days on which signal is observed between 1985 and 2006 are there "true" stock market crash or panic events?
Five? Ten? 80? 160?
Assuming, at best, 80 stock market crashes between 1985 and 2006, that's a 50% false positive rate: the reliability of thatsignal Omen is no better than flipping a coin and hoping baby gets a new pair of shoes.
Other than 1987's Black Friday and the LTCM crash in the late 90s, how many crashes have there been in that time period? More than 78?
posted by Blazecock Pileon at 2:25 AM on July 7, 2008 [2 favorites]
"The omen has appeared before all of the stock market crashes, or panic events, of the past 21 years", speaking about 1985 to 2006. Having a signal that can generate sharp market declines is appealing to all active traders, but this signal is not as common as most traders would hope. According to McHugh, the omen only created a signal on 160 separate days, or 3.2% of the approximate 5,000 days that he studied.
So I would ask, how many false positives are there from this
Five? Ten? 80? 160?
Assuming, at best, 80 stock market crashes between 1985 and 2006, that's a 50% false positive rate: the reliability of that
Other than 1987's Black Friday and the LTCM crash in the late 90s, how many crashes have there been in that time period? More than 78?
posted by Blazecock Pileon at 2:25 AM on July 7, 2008 [2 favorites]
There may be some math behind this. If we get ups and downs at the same time, we got Hindenburg -- but maybe that's an expression of a chaotic period where things can go either up or down, no odds either way. If this is an expression of chaos then other factors (a very strong economic force of some kind) may influence the up/down aspect of the Stock Market.
posted by CCBC at 2:36 AM on July 7, 2008
posted by CCBC at 2:36 AM on July 7, 2008
A couple of the links seem only to be concerned with profiting from the crash, or avoiding loss, not with any broader implications of what this kind of shock would mean for our future economic viability as a nation. "Take comfort! With this information, you can drown last!"
posted by SaintCynr at 2:39 AM on July 7, 2008 [2 favorites]
posted by SaintCynr at 2:39 AM on July 7, 2008 [2 favorites]
Your Source for Free Numerology Online
posted by Blazecock Pileon at 2:44 AM on July 7, 2008 [1 favorite]
posted by Blazecock Pileon at 2:44 AM on July 7, 2008 [1 favorite]
Blazecock: the wikipedia article seems to answer your question:
Looking back at historical data, the probability of a move greater than 5% to the downside [a crash] after a confirmed Hindenburg Omen was 77%, and usually takes place within the next forty-days. The probability of a panic sellout was 41% and the probability of a major stock market crash was 24%.
posted by Tlogmer at 2:49 AM on July 7, 2008 [1 favorite]
Looking back at historical data, the probability of a move greater than 5% to the downside [a crash] after a confirmed Hindenburg Omen was 77%, and usually takes place within the next forty-days. The probability of a panic sellout was 41% and the probability of a major stock market crash was 24%.
posted by Tlogmer at 2:49 AM on July 7, 2008 [1 favorite]
If you're looking for a more famous source, here's Barron's take.
posted by Tlogmer at 2:52 AM on July 7, 2008
posted by Tlogmer at 2:52 AM on July 7, 2008
The probability of a panic sellout was 41% and the probability of a major stock market crash was 24%.
That seems to imply roughly 40 major stock market crashes — P(major crash event given an Omen) times #(Omens) — between 1985 and 2006. Unless the Omens are not directly used to predict events.
posted by Blazecock Pileon at 2:57 AM on July 7, 2008
That seems to imply roughly 40 major stock market crashes — P(major crash event given an Omen) times #(Omens) — between 1985 and 2006. Unless the Omens are not directly used to predict events.
posted by Blazecock Pileon at 2:57 AM on July 7, 2008
Oh, the Humanity!
posted by chillmost at 3:03 AM on July 7, 2008 [2 favorites]
posted by chillmost at 3:03 AM on July 7, 2008 [2 favorites]
I'd think there are more solid reasons to worry about a stockmarket crash:
Bursting asset bubble? Check.
Soaring oil? Check.
Soaring inflation? Check.
Sinking dollar? Check.
Likely rise in interest rates? Check.
Wobbly financial markets? Check.
Incompetent economic leadership? Check.
posted by Skeptic at 3:06 AM on July 7, 2008 [4 favorites]
Bursting asset bubble? Check.
Soaring oil? Check.
Soaring inflation? Check.
Sinking dollar? Check.
Likely rise in interest rates? Check.
Wobbly financial markets? Check.
Incompetent economic leadership? Check.
posted by Skeptic at 3:06 AM on July 7, 2008 [4 favorites]
Surely this... will not wendell.
posted by Poolio at 3:10 AM on July 7, 2008 [3 favorites]
posted by Poolio at 3:10 AM on July 7, 2008 [3 favorites]
BP, I would assume some clustering of signals prior to a market move, rather than a 1:1 ratio.
posted by bashos_frog at 3:10 AM on July 7, 2008 [1 favorite]
posted by bashos_frog at 3:10 AM on July 7, 2008 [1 favorite]
This kind of stock-market stat geekery always reminds me of biorhythms, or horoscopes. Or those guys you see in the lounge at the racetrack with huge binders of statistics and "proven systems" of picking winning horses.
posted by jbickers at 3:12 AM on July 7, 2008
posted by jbickers at 3:12 AM on July 7, 2008
Skeptic, half the things you mention are related to energy companies experiencing all-time highs, while financial companies are experiencing major lows. I think this is an interesting indicator for an unhealthy market, which is likely to correct downward.
posted by bashos_frog at 3:15 AM on July 7, 2008 [1 favorite]
posted by bashos_frog at 3:15 AM on July 7, 2008 [1 favorite]
Wow Blazecock Pileon is spot on with his criticisms!
Well, at first glance one might first ask where is the market gonna crash from? After all, its not exactly overvalued at current levels - not really dirt cheap either - and in fact has been beat to hell as anyone with any type of market exposure already knows.
Ok here is the backdrop as I see it. I'm not gonna put out a lot of numbers and analysis as that would be counterproductive and just not polite.
We've got lots of inflationary spikes working their way through the system right now. Oil is just the tip, pretty much any other commodity has been hit and those price increases - PPI shocks - previously absorbed largely by the intermediaries are now being passed through to the consumer, so we're seeing CPI shocks.
So at the same time the consumer is getting squeezed pretty much uniformly by increasing prices, demand is naturally being mitigated therefore companies are seeing reduced revenue.
Meanwhile, most of the world's central banks have adopted a tightening posture (talk is different from action as we'll see). Some, like the ECB already have tightened. While BOE and The Fed fritter, the Developing Nations have jumped into the breech with China, India, Vietnam, Mexico and Brazil sharply increasing interest rates, pushing base rates in some instance well into the double digits (e.g., Vietnam at 12%).
These hikes are necessary as some dead tree research I saw recently (Morgan Stanley) indicates that over 25% of the worlds economies, comprising some 42% of global population are now experience double digit inflation. Vietnam is a good example, with June's CPI numbers showing inflation raging at 26.8% per annum. Not good.
So, globally, interest rate increases are definitely on the table however the intentions of BOE and The Fed aren't that clear (let's talk more about that later).
As interest rates rise equity prices damn well better otherwise investors are better off holding Government Securities.
Share prices reflect investor expectations of future profits.
Increasing interest rates rise slow business activity, thus mitigating profits.
So in this troublesome economic environment (spiking commodity prices, increasing interest rates, slowing consumer demand) it's a no brainer to predict lower share prices.
But a crash is totally different.
Tlogmer -- "If you're looking for a more famous source, here's Barron's take."
Thanks for the Barrons link, but its behind a paywall. Is there another source?
Also I'm not aware of any academic studies, either on this indicator or citing it.
I've been aware of this indicator for a while now (it has in fact signaled many times in the past) and have looked for academic studies. Just checked before I commented, and there are none out there.
posted by Mutant at 3:15 AM on July 7, 2008 [9 favorites]
Well, at first glance one might first ask where is the market gonna crash from? After all, its not exactly overvalued at current levels - not really dirt cheap either - and in fact has been beat to hell as anyone with any type of market exposure already knows.
Ok here is the backdrop as I see it. I'm not gonna put out a lot of numbers and analysis as that would be counterproductive and just not polite.
We've got lots of inflationary spikes working their way through the system right now. Oil is just the tip, pretty much any other commodity has been hit and those price increases - PPI shocks - previously absorbed largely by the intermediaries are now being passed through to the consumer, so we're seeing CPI shocks.
So at the same time the consumer is getting squeezed pretty much uniformly by increasing prices, demand is naturally being mitigated therefore companies are seeing reduced revenue.
Meanwhile, most of the world's central banks have adopted a tightening posture (talk is different from action as we'll see). Some, like the ECB already have tightened. While BOE and The Fed fritter, the Developing Nations have jumped into the breech with China, India, Vietnam, Mexico and Brazil sharply increasing interest rates, pushing base rates in some instance well into the double digits (e.g., Vietnam at 12%).
These hikes are necessary as some dead tree research I saw recently (Morgan Stanley) indicates that over 25% of the worlds economies, comprising some 42% of global population are now experience double digit inflation. Vietnam is a good example, with June's CPI numbers showing inflation raging at 26.8% per annum. Not good.
So, globally, interest rate increases are definitely on the table however the intentions of BOE and The Fed aren't that clear (let's talk more about that later).
As interest rates rise equity prices damn well better otherwise investors are better off holding Government Securities.
Share prices reflect investor expectations of future profits.
Increasing interest rates rise slow business activity, thus mitigating profits.
So in this troublesome economic environment (spiking commodity prices, increasing interest rates, slowing consumer demand) it's a no brainer to predict lower share prices.
But a crash is totally different.
Tlogmer -- "If you're looking for a more famous source, here's Barron's take."
Thanks for the Barrons link, but its behind a paywall. Is there another source?
Also I'm not aware of any academic studies, either on this indicator or citing it.
I've been aware of this indicator for a while now (it has in fact signaled many times in the past) and have looked for academic studies. Just checked before I commented, and there are none out there.
posted by Mutant at 3:15 AM on July 7, 2008 [9 favorites]
I'd think there are more solid reasons to worry about a stockmarket crash
Centrum can no longer cost-effectively include zinc in their vitamins...
posted by maxwelton at 3:22 AM on July 7, 2008
Centrum can no longer cost-effectively include zinc in their vitamins...
posted by maxwelton at 3:22 AM on July 7, 2008
Re - Spammy.
Crap. Sorry about that. I was posting a swathe of links. If they're too profit oriented, please nuke.
I thought it was an interesting phenomenon, not a "ZOMG MAKE MONEY FAST "moment. Stupid stock market.
posted by Lord_Pall at 3:32 AM on July 7, 2008
Crap. Sorry about that. I was posting a swathe of links. If they're too profit oriented, please nuke.
I thought it was an interesting phenomenon, not a "ZOMG MAKE MONEY FAST "moment. Stupid stock market.
posted by Lord_Pall at 3:32 AM on July 7, 2008
Lord Pall "I thought it was an interesting phenomenon, not a "ZOMG MAKE MONEY FAST "moment. Stupid stock market."
No doubt it is interesting, but the thing that does it for me with these indicators (and there are lots of them out there) is the lack of objective, academic study that either corroborates or refutes their validity.
For example, The Weekend Effect or The January Effect, - I previously inventoried a bunch of them - have in fact been studied extensively and widely corroborated and not by folks that are trying to sell you something, like this guy is (sidenote: what makes many of these indicators really interesting is 1) the fact that they persist, even after discovery, and 2) they currently and in the past have transcended national borders, even long before the markets became globally interconnected as they are now)
I've got an open mind towards these things ('cause I'm a helluva nice guy and I like to make money), but as far as I can tell, absolutely nobody has studied this indicator, nor cited it in some other area of research. Not saying it's bad, just I'd be careful of trading this indicator.
That being said, these are good times to make money in the markets. Participants are to a large extent driven by emotions (much as we try to eliminate them), and whenever we're deep in fear territory opportunities are rich.
I've got some friends that do distressed debt, and their funds are returning between six and eight percent a month. That's hefty.
So even if the markets crash, it doesn't mean the end of the world. If you think a crash is due move to cash, then post crash, pick up the shares you always wanted, dirt cheap. Of course timing the bottom is a bitch but there you go.
Alternatively, move into govvies and ride it out. If one if particularly financially astutue you could always purchase index put options, but why bother when there are ETFs out there that will let you assume that market exposure, with leverage no less? Proshares Ultrashort 30 -Ultrashort as they use leverage to magnify their short position - is but one.
Markets rise and fall. Markets can and will crash. No need to panic as there are lots of ways to profit from a market crash.
posted by Mutant at 4:01 AM on July 7, 2008 [1 favorite]
No doubt it is interesting, but the thing that does it for me with these indicators (and there are lots of them out there) is the lack of objective, academic study that either corroborates or refutes their validity.
For example, The Weekend Effect or The January Effect, - I previously inventoried a bunch of them - have in fact been studied extensively and widely corroborated and not by folks that are trying to sell you something, like this guy is (sidenote: what makes many of these indicators really interesting is 1) the fact that they persist, even after discovery, and 2) they currently and in the past have transcended national borders, even long before the markets became globally interconnected as they are now)
I've got an open mind towards these things ('cause I'm a helluva nice guy and I like to make money), but as far as I can tell, absolutely nobody has studied this indicator, nor cited it in some other area of research. Not saying it's bad, just I'd be careful of trading this indicator.
That being said, these are good times to make money in the markets. Participants are to a large extent driven by emotions (much as we try to eliminate them), and whenever we're deep in fear territory opportunities are rich.
I've got some friends that do distressed debt, and their funds are returning between six and eight percent a month. That's hefty.
So even if the markets crash, it doesn't mean the end of the world. If you think a crash is due move to cash, then post crash, pick up the shares you always wanted, dirt cheap. Of course timing the bottom is a bitch but there you go.
Alternatively, move into govvies and ride it out. If one if particularly financially astutue you could always purchase index put options, but why bother when there are ETFs out there that will let you assume that market exposure, with leverage no less? Proshares Ultrashort 30 -Ultrashort as they use leverage to magnify their short position - is but one.
Markets rise and fall. Markets can and will crash. No need to panic as there are lots of ways to profit from a market crash.
posted by Mutant at 4:01 AM on July 7, 2008 [1 favorite]
No need to panic as there are lots of ways to profit from a market crash.
That's the message of at least half the links in this FPP.
posted by rokusan at 4:25 AM on July 7, 2008
That's the message of at least half the links in this FPP.
posted by rokusan at 4:25 AM on July 7, 2008
I know that lots of intelligent people defend technical analysis, but I agree with those that compare it to numerology. I think technical analysis very effectively taps into flaws in the way our brains are hardwired to process information.
That said, I wouldn't be at all surprised if the stock market crashed.
posted by diogenes at 5:12 AM on July 7, 2008
That said, I wouldn't be at all surprised if the stock market crashed.
posted by diogenes at 5:12 AM on July 7, 2008
One interesting thing I've seen is that this Hindenburg Omen presages a market decline about 75% of the time, while only crashing 25% of the time. But in order to calculate it, one of the indicators is "a bunch of stuff are at yearly lows" -- so yeah, some things are definitely in decline. And another indicator is that "things are at yearly highs", so there's some bubble there to be accounted for as well.
In other words, isn't this almost a tautology?
posted by taumeson at 5:15 AM on July 7, 2008
In other words, isn't this almost a tautology?
posted by taumeson at 5:15 AM on July 7, 2008
It would save time if mutant got to moderate all financial submissions.
There may be an outbreak of financial numerology. There is some talk about the Kondratiev wave out there, which is something like biorhythms, but for markets.
Are there any other popular ones?
posted by sien at 5:34 AM on July 7, 2008
There may be an outbreak of financial numerology. There is some talk about the Kondratiev wave out there, which is something like biorhythms, but for markets.
Are there any other popular ones?
posted by sien at 5:34 AM on July 7, 2008
behind a paywall
Oops -- I got there from google news -- here's the gnews link
posted by Tlogmer at 5:36 AM on July 7, 2008
Oops -- I got there from google news -- here's the gnews link
posted by Tlogmer at 5:36 AM on July 7, 2008
Aw. Crud. Nevermind. Well, it was free a couple hours ago.
posted by Tlogmer at 5:36 AM on July 7, 2008
posted by Tlogmer at 5:36 AM on July 7, 2008
Oh god, more "technicals" analysis, where people look at the patern of the a stocks (or markets) ups and downs and try to make predictions based entirely on that. It's about as psudosciencey as you get get, these people would have better luck with a ouija board.
posted by delmoi at 5:57 AM on July 7, 2008
posted by delmoi at 5:57 AM on July 7, 2008
Blazecock Pileon:
Under your assumed scenario of 160 predictions of "crash" followed by 80 crashes and 80 non-crashes, yes, that's a 50% false positive rate, and yes, that's the same false positive rate that flipping a coin would give you.
But that's not all there is to the story. A coinflip would also give you a 50% false negative rate. Presumably, this "omen" would not. That is a very important distinction.
For purposes of demonstration, let's say that this "omen" makes a prediction once a day, for the next market day (I understand that it does not, and see the bottom of this post for more detail, but let's make this assumption for, again, the purposes of demonstration).
There have been something like, what, roughly 5000 trading days in that period. Let's assume that few if any, other than those preceded by this "omen", were significant crashes.
Then a coin would say:
Meanwhile, this "omen" would say:
Such an "omen", were it real, would be extremely valuable, despite the fact that it would get it wrong in half the cases that it says "crash coming". By following its advice, you would miss every major crash, at a very minor cost (i.e. you miss about 1.6% of normal market days).
In fact, it could be extremely valuable even if it were significantly worse than a coin in the case that it says "crash coming".
What I am more wary of, regarding this supposed omen, is this line from Wikipedia:
Perhaps the real author actually makes a similar statement in a well-defined manner, but the statement, as it is in Wikipedia, is phrased in a manner that is common to a lot of poor probabilistic arguments.
Plus, even if it's just "40 days", if Blazecock is right about that "160 predictions in 20 years", it's entirely possible (depending upon how those 160 predictions are distributed within those 20 years) that this "omen" is predicting that you should always stay out of the market.
posted by Flunkie at 6:21 AM on July 7, 2008 [1 favorite]
So I would ask, how many false positives are there from this algorithm Omen: from how many of the 160 separate days on which signal is observed between 1985 and 2006 are there "true" stock market crash or panic events?I agree with your overall sentiment, but this is shoddy mathematics.
Five? Ten? 80? 160?
Assuming, at best, 80 stock market crashes between 1985 and 2006, that's a 50% false positive rate: the reliability of that signal Omen is no better than flipping a coin and hoping baby gets a new pair of shoes.
Under your assumed scenario of 160 predictions of "crash" followed by 80 crashes and 80 non-crashes, yes, that's a 50% false positive rate, and yes, that's the same false positive rate that flipping a coin would give you.
But that's not all there is to the story. A coinflip would also give you a 50% false negative rate. Presumably, this "omen" would not. That is a very important distinction.
For purposes of demonstration, let's say that this "omen" makes a prediction once a day, for the next market day (I understand that it does not, and see the bottom of this post for more detail, but let's make this assumption for, again, the purposes of demonstration).
There have been something like, what, roughly 5000 trading days in that period. Let's assume that few if any, other than those preceded by this "omen", were significant crashes.
Then a coin would say:
- There will be 2500 crashes (and would be correct in 40 cases)
- There will be 2500 non-crashes (and would be correct in 2460 cases)
Meanwhile, this "omen" would say:
- There will be 160 crashes (and would be correct in 80 cases)
- There will be 4840 non-crashes (and would be correct in 4840 cases)
Such an "omen", were it real, would be extremely valuable, despite the fact that it would get it wrong in half the cases that it says "crash coming". By following its advice, you would miss every major crash, at a very minor cost (i.e. you miss about 1.6% of normal market days).
In fact, it could be extremely valuable even if it were significantly worse than a coin in the case that it says "crash coming".
What I am more wary of, regarding this supposed omen, is this line from Wikipedia:
Looking back at historical data, the probability of a move greater than 5% to the downside after a confirmed Hindenburg Omen was 77%, and usually takes place within the next forty-days."Usually takes place within the next forty days" smacks of vagueness, which is never a good sign for a probabilistic argument. "Usually"? Well what about the other cases? How long after the "omen" does a crash have to occur before the author of that statement considers the crash to be unrelated to the omen?
Perhaps the real author actually makes a similar statement in a well-defined manner, but the statement, as it is in Wikipedia, is phrased in a manner that is common to a lot of poor probabilistic arguments.
Plus, even if it's just "40 days", if Blazecock is right about that "160 predictions in 20 years", it's entirely possible (depending upon how those 160 predictions are distributed within those 20 years) that this "omen" is predicting that you should always stay out of the market.
posted by Flunkie at 6:21 AM on July 7, 2008 [1 favorite]
For Blazecock and Flunkie. It appears the difference between what is being claimed 160 predictions in 20 years and the probability of a crash (etc) is whether or not it is a "confirmed Hindenburg Omen." The Hindenburg Omen appears frequently, the confirmed one is rare. I'm not trying to make the large argument that this omen is any good in predicting, I'm just adding what I understood from the reading.
posted by dances_with_sneetches at 6:33 AM on July 7, 2008
posted by dances_with_sneetches at 6:33 AM on July 7, 2008
There is something inherently wrong with an economic system so fraught with manipulation and corruption that those playing with the fire are seldom the ones who end up getting burned.
posted by any major dude at 7:09 AM on July 7, 2008
posted by any major dude at 7:09 AM on July 7, 2008
When was the Hindenburg Omen formulated? I submit that with the millions of possible metrics that might be formulated from stock market data, it's easy to find one that appears to have good correlation with past events, and thus, only its results since the time it was published should be used to measure its accuracy. (And if you "tweak" it, you have to start over at the time the tweak is published.)
Around November 2004, I thought I had found a good metric for picking NFL games--looking back at the past few seasons, I found this metric would have picked 75-80% of games it picked accurately, against the spread, from 2001 through that point in 2004. (It only picked a small fraction of games each week, but that was part of the metric.) Rather than going out and betting my life savings on games, I decided to watch it a while longer and see if it held up. Since then, it's picked about 55% of games correctly, and since it only makes picks on a smallish number of games to begin with, it's not statistically significant.
Whether this is happening with the Hindenburg Omen, I'm not sure, but it's something to consider.
posted by DevilsAdvocate at 7:10 AM on July 7, 2008
Around November 2004, I thought I had found a good metric for picking NFL games--looking back at the past few seasons, I found this metric would have picked 75-80% of games it picked accurately, against the spread, from 2001 through that point in 2004. (It only picked a small fraction of games each week, but that was part of the metric.) Rather than going out and betting my life savings on games, I decided to watch it a while longer and see if it held up. Since then, it's picked about 55% of games correctly, and since it only makes picks on a smallish number of games to begin with, it's not statistically significant.
Whether this is happening with the Hindenburg Omen, I'm not sure, but it's something to consider.
posted by DevilsAdvocate at 7:10 AM on July 7, 2008
The more I look for the number 5, the more I keep finding it.
posted by Nelson at 7:19 AM on July 7, 2008
posted by Nelson at 7:19 AM on July 7, 2008
I am not worried. Today's stock market is inflated with helium instead of hydrogen.
posted by weapons-grade pandemonium at 7:46 AM on July 7, 2008 [2 favorites]
posted by weapons-grade pandemonium at 7:46 AM on July 7, 2008 [2 favorites]
I heard about this last year, but I'm surprised it didn't make MeFi back then. The one in October was followed by a ~9.5% drop in the Dow over the next 40 days. This one, according to the Seeking Alpha buys, was first seen June 6, and since then the Dow has dropped ~9.5%.
So, if this is a real measure and not caught in some sort of confirmation bias, then it suggests it's a good measure that the market is about to slump for a 40 day period. Not crash, slump.
I guess I'd compare it to NOAA issuing a tornado watch due to favorable conditions. Conditions are ripe for tornadoes. It neither means a tornado will hit your house nor will there be an F5 causing cataclysmic destruction. There may not be any tornadoes. But it's something to watch for.
posted by dw at 7:52 AM on July 7, 2008
So, if this is a real measure and not caught in some sort of confirmation bias, then it suggests it's a good measure that the market is about to slump for a 40 day period. Not crash, slump.
I guess I'd compare it to NOAA issuing a tornado watch due to favorable conditions. Conditions are ripe for tornadoes. It neither means a tornado will hit your house nor will there be an F5 causing cataclysmic destruction. There may not be any tornadoes. But it's something to watch for.
posted by dw at 7:52 AM on July 7, 2008
And of course, the Dow != the market.
Kinda how oil != oil company stock anymore. Which is all I can say about how oil has risen nearly 40% since the start of the year, but oil company stocks have been flat to negative. Arrgh.
posted by dw at 7:58 AM on July 7, 2008
Kinda how oil != oil company stock anymore. Which is all I can say about how oil has risen nearly 40% since the start of the year, but oil company stocks have been flat to negative. Arrgh.
posted by dw at 7:58 AM on July 7, 2008
dw makes a great point in that The Dow isn't the market; in fact The Dow comes under a lot of criticism, as it is a price weighted average of 30 shares, while the S&P 500 is a market capitilisation weighted average of 500 shares. The S&P 500 is representative of the broader market, if you will.
The Dow is very interesting because of it's volatility; that same price weighting gives rise to interesting trading opps. I used to trade slightly out of the money short term options on The Dow, just because all you needed was a sharp movement in one or more of the higher priced components to see a nice return.
Anyhow, looking at Q1 data, we see that operating earnings were hammered, declining some 25.9%. But that's deceptive, as if we exclude financials we can see that operating earnings were actually up 8.8% YOY. Not too shabby.
Last year one of the larger contributing sectors to the S&P 500 earnings were financials (29.7%), now, however, we're seeing a shift to energy which contributed to 23.3% of overall earnings.
The takeaway: just as The Dow isn't the market, the market isn't the market and shouldn't be viewed monolithically. There are well performing subsectors, now and especially when markets crater. Some sector, subsector of grouping of shares will offer positive returns no matter how the overall market is performing; this is almost always true.
When folks get scared they bail on the entire market, dumping everything in their fear to protect their wealth, thus giving rise to opps that more disciplined investors can exploit.
Emotion is a stone cold killer in the equity markets.
posted by Mutant at 8:32 AM on July 7, 2008 [2 favorites]
The Dow is very interesting because of it's volatility; that same price weighting gives rise to interesting trading opps. I used to trade slightly out of the money short term options on The Dow, just because all you needed was a sharp movement in one or more of the higher priced components to see a nice return.
Anyhow, looking at Q1 data, we see that operating earnings were hammered, declining some 25.9%. But that's deceptive, as if we exclude financials we can see that operating earnings were actually up 8.8% YOY. Not too shabby.
Last year one of the larger contributing sectors to the S&P 500 earnings were financials (29.7%), now, however, we're seeing a shift to energy which contributed to 23.3% of overall earnings.
The takeaway: just as The Dow isn't the market, the market isn't the market and shouldn't be viewed monolithically. There are well performing subsectors, now and especially when markets crater. Some sector, subsector of grouping of shares will offer positive returns no matter how the overall market is performing; this is almost always true.
When folks get scared they bail on the entire market, dumping everything in their fear to protect their wealth, thus giving rise to opps that more disciplined investors can exploit.
Emotion is a stone cold killer in the equity markets.
posted by Mutant at 8:32 AM on July 7, 2008 [2 favorites]
What, you mean anyone has enough confidence to still own American stocks? LOL!
posted by Goofyy at 8:35 AM on July 7, 2008
posted by Goofyy at 8:35 AM on July 7, 2008
What, you mean anyone has enough confidence to still own American stocks? LOL!
Yeah..I'm going to be laughing too when great stocks become cheap because of attitudes like yours and I buy a whole crap load of them and 10 years from now when the cycle spins the other way I retire to the Bahamas.
posted by spicynuts at 9:02 AM on July 7, 2008 [1 favorite]
Yeah..I'm going to be laughing too when great stocks become cheap because of attitudes like yours and I buy a whole crap load of them and 10 years from now when the cycle spins the other way I retire to the Bahamas.
posted by spicynuts at 9:02 AM on July 7, 2008 [1 favorite]
According to McHugh, the omen only created a signal on 160 separate days, or 3.2% of the approximate 5,000 days that he studied.
I only read the first half of the links, so maybe this is cleared up later, but isn't it likely that some of those signals were days or weeks apart, and could therefore both be pointing to the same event? You might not need 120 crashes for a 75% success rate. For all I know, you would only need 40. Do the signals cluster like that, or are the all distinct events?
posted by Pater Aletheias at 9:13 AM on July 7, 2008
I only read the first half of the links, so maybe this is cleared up later, but isn't it likely that some of those signals were days or weeks apart, and could therefore both be pointing to the same event? You might not need 120 crashes for a 75% success rate. For all I know, you would only need 40. Do the signals cluster like that, or are the all distinct events?
posted by Pater Aletheias at 9:13 AM on July 7, 2008
Emotion is a stone cold killer in the equity markets.
Which is exactly why I go for passive management and asset allocation, cause I can't tie my shoes without getting emotional.
posted by slickvaguely at 9:29 AM on July 7, 2008 [1 favorite]
Which is exactly why I go for passive management and asset allocation, cause I can't tie my shoes without getting emotional.
posted by slickvaguely at 9:29 AM on July 7, 2008 [1 favorite]
A rare signal or event in the Shannon’s information theories (the backbone of the modern day digital communications) is considered to contain a higher amount of information.
Man, I love information theory. In biology, it's always good for some breathless statement by someone trying to grapple with the behavior of really complex systems. In the wrong hands, it's kind of like astrology for statistics. And it rarely seems to be in the right hands. Is that also true for financial systems, or is it a better tool there?
posted by Tehanu at 9:32 AM on July 7, 2008
Man, I love information theory. In biology, it's always good for some breathless statement by someone trying to grapple with the behavior of really complex systems. In the wrong hands, it's kind of like astrology for statistics. And it rarely seems to be in the right hands. Is that also true for financial systems, or is it a better tool there?
posted by Tehanu at 9:32 AM on July 7, 2008
Then a coin would say:
* There will be 2500 crashes (and would be correct in 40 cases)
* There will be 2500 non-crashes (and would be correct in 2460 cases)
For an overall predictive ability of 50%.
Meanwhile, this "omen" would say:
* There will be 160 crashes (and would be correct in 80 cases)
* There will be 4840 non-crashes (and would be correct in 4840 cases)
For an overall predictive ability of 98.4%.
I think you're badly misunderstanding the coin metaphor. I also think your understanding of the math is perhaps not entirely correct here.
1. It is unlikely that there is a 0% false negative rate. Unless this model is perfect, and by all appearances it is not, by some non-zero chance some of the non-crash "predictions" will in fact be wrong, i.e. some of the "non-events" as such will indeed be crashes.
2. It's not that, of 5000 trials, a coin would say 2500 crashes and 2500 non-crashes. That assumes the model is a coin toss, which I'm not saying. I don't even know or care what the model is, and it doesn't even really matter.
3. It is that, of the 160 signals used to predict a crash (assuming that signal-implies-crash-within-a-time-frame), only a handful of those signals coming out of the model seem actually predictive, given empirical knowledge (such as a list of crash events on Wikipedia). It would require 80 market crashes over 1985 to 2006 for the model gain the predictive power of a flipped coin. Or: Granting 80 crashes over that 20-year time period, if I told you that with certainty the market was now going to crash, you could flip a coin to decide if my crash prediction was true or false.
4. The Omen is not used to predict non-crashes. It predicts crashes. It's not informative to predict non-crashes, because that is the preexisting base condition or null distribution. You're using "signal" to discern crash conditions and move your investments around to protect your money. Or to put it in different terms: You don't test a drug for non-efficacy, you test it to work better than the previous, existing drug (or placebo).
Here's another example. Of a survey of 1000 genes, looking for interesting genes that might be tied to breast cancer, if your statistical test says that 50 genes are worth spending your lab's funding, equipment and staff time on, but the test has a 50% failure rate, then only 25 of those genes were probably of any interest to begin with. Missing the target can potentially be a massive failure: you don't publish and lose all your funding. Having a good test is critical.
For an investor who ties his or her money to the outcomes predicted by the Omen (or by any Omen, German or otherwise) it is much more important that positive signals have the power to be predictive. No one cares when the model doesn't predict anything -- by all appearances, the Omen seems to have kept stumm when the housing market crashed, for example.
posted by Blazecock Pileon at 9:38 AM on July 7, 2008
* There will be 2500 crashes (and would be correct in 40 cases)
* There will be 2500 non-crashes (and would be correct in 2460 cases)
For an overall predictive ability of 50%.
Meanwhile, this "omen" would say:
* There will be 160 crashes (and would be correct in 80 cases)
* There will be 4840 non-crashes (and would be correct in 4840 cases)
For an overall predictive ability of 98.4%.
I think you're badly misunderstanding the coin metaphor. I also think your understanding of the math is perhaps not entirely correct here.
1. It is unlikely that there is a 0% false negative rate. Unless this model is perfect, and by all appearances it is not, by some non-zero chance some of the non-crash "predictions" will in fact be wrong, i.e. some of the "non-events" as such will indeed be crashes.
2. It's not that, of 5000 trials, a coin would say 2500 crashes and 2500 non-crashes. That assumes the model is a coin toss, which I'm not saying. I don't even know or care what the model is, and it doesn't even really matter.
3. It is that, of the 160 signals used to predict a crash (assuming that signal-implies-crash-within-a-time-frame), only a handful of those signals coming out of the model seem actually predictive, given empirical knowledge (such as a list of crash events on Wikipedia). It would require 80 market crashes over 1985 to 2006 for the model gain the predictive power of a flipped coin. Or: Granting 80 crashes over that 20-year time period, if I told you that with certainty the market was now going to crash, you could flip a coin to decide if my crash prediction was true or false.
4. The Omen is not used to predict non-crashes. It predicts crashes. It's not informative to predict non-crashes, because that is the preexisting base condition or null distribution. You're using "signal" to discern crash conditions and move your investments around to protect your money. Or to put it in different terms: You don't test a drug for non-efficacy, you test it to work better than the previous, existing drug (or placebo).
Here's another example. Of a survey of 1000 genes, looking for interesting genes that might be tied to breast cancer, if your statistical test says that 50 genes are worth spending your lab's funding, equipment and staff time on, but the test has a 50% failure rate, then only 25 of those genes were probably of any interest to begin with. Missing the target can potentially be a massive failure: you don't publish and lose all your funding. Having a good test is critical.
For an investor who ties his or her money to the outcomes predicted by the Omen (or by any Omen, German or otherwise) it is much more important that positive signals have the power to be predictive. No one cares when the model doesn't predict anything -- by all appearances, the Omen seems to have kept stumm when the housing market crashed, for example.
posted by Blazecock Pileon at 9:38 AM on July 7, 2008
> a move greater than 5% to the downside [a crash]
There's the joker in the deck. If you define a crash as a drop greater than 1% then ZOMG the market crashes all the time, but that's a silly definition. 5% is silly also.
Three groups don't think it's silly. These are 1. Doom-peddling websites; 2. How-To-Prosper-In-The-Coming-Holocaust book peddlers; and 3. economists who have been overcome by their vested interest in trying to make the Dismal Science thrilling. These love to steal drama from the common-language meaning of "crash" (which covers things like flaming jetliner crashes--"Hindenburg Event," forsooth) and glue the stolen drama onto some statistic that would otherwise have a two-paragraph My Eyes Glaze Over description instead of a name. Nobody else keeps a 5% market drop in the same mental cupboard as the Concorde coming down. (Well, maybe the folks who bought in with Other People's Money and got margin calls.)
A crash, properly so called, is a sudden drop in the value of all the market's imaginary money that is expressible as a simple fraction like "one half" or "three fourths" and is followed by extended economic misery. 1929 was a crash. 1987 may have qualified in other markets but did not in the US (market down only 22% on Black Monday, and followed by no depression--the long, steady rise picked up right where it left off.)
So: from the 1920s to today, one crash.
posted by jfuller at 9:43 AM on July 7, 2008
There's the joker in the deck. If you define a crash as a drop greater than 1% then ZOMG the market crashes all the time, but that's a silly definition. 5% is silly also.
Three groups don't think it's silly. These are 1. Doom-peddling websites; 2. How-To-Prosper-In-The-Coming-Holocaust book peddlers; and 3. economists who have been overcome by their vested interest in trying to make the Dismal Science thrilling. These love to steal drama from the common-language meaning of "crash" (which covers things like flaming jetliner crashes--"Hindenburg Event," forsooth) and glue the stolen drama onto some statistic that would otherwise have a two-paragraph My Eyes Glaze Over description instead of a name. Nobody else keeps a 5% market drop in the same mental cupboard as the Concorde coming down. (Well, maybe the folks who bought in with Other People's Money and got margin calls.)
A crash, properly so called, is a sudden drop in the value of all the market's imaginary money that is expressible as a simple fraction like "one half" or "three fourths" and is followed by extended economic misery. 1929 was a crash. 1987 may have qualified in other markets but did not in the US (market down only 22% on Black Monday, and followed by no depression--the long, steady rise picked up right where it left off.)
So: from the 1920s to today, one crash.
posted by jfuller at 9:43 AM on July 7, 2008
To understand how markets work and and know how to predict them, go swim in the ocean for awhile. Better yet, surf.
You can see the waves coming. You sit there on your board and you feel the waves push you up and down. Sometimes the waves are good, and sometimes not so good. You can see that. You can see a good wave coming, and it's timed just right with your position, so that with a good stroke you can drop into it, get up on the board, and you are surfing.
Every now and then that good wave turns out not so good, or it develops faster than you thought, and you get pummeled, or it develops slower, and you miss it. It's no big deal though, cause another wave is coming, and you'll get one. There's definitely a rhythm to the waves, you can see it, you can feel it. The waves are entirely rational. They are predictable. There are good places to surf, and good conditions to surf in.
Now, try to predict exactly when a good wave will come, a good catchable wave that you can just drop right in on - exactly one minute into the future. One minute. Good luck.
posted by Xoebe at 10:01 AM on July 7, 2008 [3 favorites]
You can see the waves coming. You sit there on your board and you feel the waves push you up and down. Sometimes the waves are good, and sometimes not so good. You can see that. You can see a good wave coming, and it's timed just right with your position, so that with a good stroke you can drop into it, get up on the board, and you are surfing.
Every now and then that good wave turns out not so good, or it develops faster than you thought, and you get pummeled, or it develops slower, and you miss it. It's no big deal though, cause another wave is coming, and you'll get one. There's definitely a rhythm to the waves, you can see it, you can feel it. The waves are entirely rational. They are predictable. There are good places to surf, and good conditions to surf in.
Now, try to predict exactly when a good wave will come, a good catchable wave that you can just drop right in on - exactly one minute into the future. One minute. Good luck.
posted by Xoebe at 10:01 AM on July 7, 2008 [3 favorites]
1937-38 - Dow drops 49%, years of Depression continue.
1973-74 - Dow drops 45%, years of stagflation follow.
1939-42 - Dow drops 42%, years of War-associated hardship follow.
posted by storybored at 10:08 AM on July 7, 2008 [1 favorite]
1973-74 - Dow drops 45%, years of stagflation follow.
1939-42 - Dow drops 42%, years of War-associated hardship follow.
posted by storybored at 10:08 AM on July 7, 2008 [1 favorite]
So: from the 1920s to today, one crash.
To elaborate, no-one remembers the Crashes of yesteryear. Aside from the biggie in '29.
I'd like to propose the Santayana indicator - the probability of a Crash increases with the financial amnesia of the investing population.
posted by storybored at 10:13 AM on July 7, 2008
To elaborate, no-one remembers the Crashes of yesteryear. Aside from the biggie in '29.
I'd like to propose the Santayana indicator - the probability of a Crash increases with the financial amnesia of the investing population.
posted by storybored at 10:13 AM on July 7, 2008
1973-74 - Dow drops 45%, years of stagflation follow.
Actually, we were already in stagflation by then. And two years later, the market had recovered most of the losses in '74.
We're entering what may be a generation-long bear market. But that doesn't mean the market will just keep dropping. What it means is that the opportunities to make money will be different. The big question is whether this generation of financial advisors, born in the 1970s and 1980s and raised during a long period of incredible economic growth, will understand how a bear market works well enough to identify these opportunities.
Mutant is right -- those without emotion will reap the rewards. Part of the reason why I'm considering getting an actual financial adviser. I'm too emotional when it comes to the markets.
posted by dw at 11:20 AM on July 7, 2008
Actually, we were already in stagflation by then. And two years later, the market had recovered most of the losses in '74.
We're entering what may be a generation-long bear market. But that doesn't mean the market will just keep dropping. What it means is that the opportunities to make money will be different. The big question is whether this generation of financial advisors, born in the 1970s and 1980s and raised during a long period of incredible economic growth, will understand how a bear market works well enough to identify these opportunities.
Mutant is right -- those without emotion will reap the rewards. Part of the reason why I'm considering getting an actual financial adviser. I'm too emotional when it comes to the markets.
posted by dw at 11:20 AM on July 7, 2008
Thank you for posting an unfathomable economic indicator that makes me feel like I'm not missing anything.
The stock market's still what, 5x it's size in 1990?
Personally, I don't think that consumer culture can withstand a SERIOUSTATO depression, but I look forward to being proven wrong. It should build character, at least.
I miss the perspective of people who lived through the great depression.
I feel that a sense of entitlement has permeated all our thoughts on economic policy.
posted by Busithoth at 11:38 AM on July 7, 2008
The stock market's still what, 5x it's size in 1990?
Personally, I don't think that consumer culture can withstand a SERIOUSTATO depression, but I look forward to being proven wrong. It should build character, at least.
I miss the perspective of people who lived through the great depression.
I feel that a sense of entitlement has permeated all our thoughts on economic policy.
posted by Busithoth at 11:38 AM on July 7, 2008
* That the daily number of NYSE new 52 Week Highs and the daily number of new 52 Week Lows must both be greater than 2.2 percent of total NYSE issues traded that day.
* That the smaller of these numbers is greater than 75. (this is not a rule but a function of the 2.2% of the total issues)
* That the NYSE 10 Week moving average is rising.
* That the McClellan Oscillator is negative on that same day.
* That new 52 Week Highs cannot be more than twice the new 52 Week Lows (however it is fine for new 52 Week Lows to be more than double new 52 Week Highs). This condition is absolutely mandatory.
* That the smaller of these numbers is greater than 75. (this is not a rule but a function of the 2.2% of the total issues)
* That the NYSE 10 Week moving average is rising.
* That the McClellan Oscillator is negative on that same day.
* That new 52 Week Highs cannot be more than twice the new 52 Week Lows (however it is fine for new 52 Week Lows to be more than double new 52 Week Highs). This condition is absolutely mandatory.
- If the tertiary weighted averages are currently subject to trunkchecking, it is only a Double Secret Probationary Hindenburg Omen subject to future unpublishing.
- Flip a Kennedy half-dollar 10 times. If 7 heads or more, the Omen is not in effect. DO NOT flip a Susan B. Anthony dollar 10 times except on Leap Day.
- 2nd Zeppelin Omen: If everything still turns to gold but you can listen very hard, maintaining rock-like stability as the markets roll, it may be time to invest in stairway companies.
Thank you for posting an unfathomable economic indicator that makes me feel like I'm not missing anything.
Seconded.
I was all expecting to say "I don't follow the stock markets, but that was a cool analysis."
Unfortunately, that was about as far from the case as possible. BO-RING!
I think Wikipedia hit it squarely on the head: "if one backtests through a large data set and tries enough different variables, eventually correlations are bound to be found that don't really have any predictive significance." YES.
posted by mrgrimm at 12:12 PM on July 7, 2008
Seconded.
I was all expecting to say "I don't follow the stock markets, but that was a cool analysis."
Unfortunately, that was about as far from the case as possible. BO-RING!
I think Wikipedia hit it squarely on the head: "if one backtests through a large data set and tries enough different variables, eventually correlations are bound to be found that don't really have any predictive significance." YES.
posted by mrgrimm at 12:12 PM on July 7, 2008
I read these threads just to get Mutant's contribution. [/mancrush]
posted by bumpkin at 12:56 PM on July 7, 2008
posted by bumpkin at 12:56 PM on July 7, 2008
Oh, the Humanity!
Man, I just came in here for "Oh, the economy!". Thanks for the jinx.
posted by goodnewsfortheinsane at 1:32 PM on July 7, 2008
Man, I just came in here for "Oh, the economy!". Thanks for the jinx.
posted by goodnewsfortheinsane at 1:32 PM on July 7, 2008
I am not convinced by all this technical stock analysis. Technical analysis, Hindenburg Signals, seems random to me. In the long run most stocks will climb due to inflation. Would be interesting to see the S&P500 or DOW in Gold and how traders perfom if measured in gold.
This is a better information ressource about stock markets:
The (Mis)behaviour of Markets by Benoit B. Mandelbrot
posted by yoyo_nyc at 7:12 PM on July 7, 2008
This is a better information ressource about stock markets:
The (Mis)behaviour of Markets by Benoit B. Mandelbrot
posted by yoyo_nyc at 7:12 PM on July 7, 2008
"No panic sell-off occurred over the past 21 years without the presence of a Hindenburg Omen."
No panic sell-off occurred over the past 21 years without the presence of a man wearing red socks.
(I am aware of the false analogy, but the way this is written, it doesn't do too good of a job at differentiating correlation with causation. Just sounds kinda... half-assed.)
posted by tehloki at 8:31 AM on July 8, 2008
No panic sell-off occurred over the past 21 years without the presence of a man wearing red socks.
(I am aware of the false analogy, but the way this is written, it doesn't do too good of a job at differentiating correlation with causation. Just sounds kinda... half-assed.)
posted by tehloki at 8:31 AM on July 8, 2008
« Older Expeditions | Truly Brutal Newer »
This thread has been archived and is closed to new comments
posted by Happy Dave at 1:49 AM on July 7, 2008