Behavioral finance FAQ / Glossary (Cascade)
This is a separate page of the C section of the Glossary
Dates of related message(s) in the
Behavioral-Finance group (*):
Year/month, d: developed/ discussed,
(Information / informational) Cascade, cascading
00/7i - 01/4i,9i,10i,12i - 02/1i,
8i,9i - 03/6i,11i,12i - 04/10i -
06/2i,3i + see (rational)
expectations, rational bubbles,
crashes, trend, signal, noise,
reflexivity, beauty contest,
percolation + bfdef2
"If they do it, they must know something I don't know"
=> So let us jump with them in the waterfall !
Definition (in general) :
In general, an information cascade is a collective behavioral bias by which a
superficial (and often emotional) interpretation (availability cascade) of
a minor event get spread, accelerated and amplified.
Definition (in markets)
A market-type information cascade materializes in a snowballing /
accelerating/ cumulative price move.
It starts with an initial price change that players
interpret as a signal that other players have relevant information.
It persists in a series (cascade) of self-reinforcing information /
How does the market start the flow?
Do those guys really know something I don't know?
Or is it just what water-tasting / cascade-rafting
An information cascade can start from a tiny buying or selling surge
by some traders, for good or bad reasons.
They are imitated by other players who suppose
the first ones to be better informed than them.
This first wave can launch repetitive moves in which most players
imitate one another, instead of taking into account
their own private information.
This gives a series (cascade) of self-reinforcing
information / self-feeding behavior
This mimetic phenomenon has some similarities with (see those words):
The "beauty contest", but without the divinatory element. And it might
be called ...ugliness contest when it sends price spiraling downwards,
"Herding" (but without the emotional content).
Other social epidemics and self-replicating phenomena found in social
communication : viral communication, rumor, buzz, meme....
Short term and long term market cascades
How fast can the cascade get dry again?
A continuous trickle or avalanche (depending on its strength) of buying or
selling irrupts sometimes in financial markets.
It sustains a price trend, either upward or downward, temporary
or persistent, .
We have here a serial "positive feedback loops" (see the word "feedback"),
in which events get farther and farther from the mean, leading to a vicious (or
sometimes virtuous) spiral.
Actually, there are:
Many short term cascades that usually abort
soon (often the day they started),
But also some medium / long term
ones that can reach extreme price lows and highs.
This happens when investor's emotions enter the game and reinforce what
was at the start only a feeble cognitive phenomenon (see below the section
on persistent cascades).
If the move becomes quite strong, it can even influence fundamentals
in a self-predicting way (see reflexivity, self fulfilling prophecy).
In financial markets,
how is an information cascade is triggered
and how does it evolve into a spiral?
An ear on the rail, in case an incoming financial train
splashes a steam cascade
At its start, a financial information cascade can be triggered by a minor
signal, which breaks through the surrounding "noise" and
comes either from outside (exogenous) or from inside (endogenous)
1) Cascade activated by (positive or negative)
(= coming from the economic sphere)
After this preliminary signal, market prices start moving,
The cascade starts if this price move is taken as an exogenous
signal (= one coming from the financial sphere, in this case the
This convergence between exogenous and endogenous signals might
start the spiral - even in the lack of new confirmation from the
By the way, the economy also can be hit by self reinforcing
They lead to economic development or decline in the long term, or
to cyclical expansion or depression in shorter periods.
Inflation spirals are another example, when people buy and hoard
goodsby fear of inflation, contributing themselves to the price rise.
2) Cascade proceeding directly from "endogenous"
info (from the financial market itself, a noticeable price
In this case:
1) The first information is a conspicuous market price rise or fall,
in the lack of any visible economic event than would explain it.
2) Some players take it as a signal that some other traders
are buying or selling because "they must know something
that I don't know",
This is typical "noise trading" (see that phrase).
Somebody raises a little its voice, everybody does.
2b) Thus, they join the party and buy or sell in their turn.
Those actions reinforce the rise or fall, adding a second wave.
3) This new move brings a feeling among investors that the trend is
It attracts even more people, who think there must be a good
reasonfor it, and tend to believe and spread the associated
chatter, gossip and noise.
Those additional players start a third wave of rises or falls.
4) - 5) - X) Another wave of people sees that
and act on their turn, and things cascade on and on.
Thus, even a lack of initial information by the first people
who started the trend can solidify into an information
for the next ones who saw their moves.
It shows that the trend can start out of nothing, when a pure
noise is taken as a signal (see noise, signal).
Technical analysis might play a part, when traders wait for a
rise or fall,as a signal that the trend is reverting, before they start
to buy or sell.
Are information cascades rational?
Tactical followers sharing the shower.
In an information cascade, people decide consciously not to use their
private information and to act as "followers".
They deliberately consider the information that other players seems to know
as more relevant than what they know by themselves.
They feel under-informed on fundamental facts. Thus they
have more trust in market price information, even
if it is just a noise or a slight adaptive market move.
Cascades vs. herding.
And how persistent are they?
Flowing calculators? Or flowing cattle?
Cascades are sometimes called "rational herding".
Is that definition justified ?
There is some rationality,
cool calculation and
consciousness in information
cascades, even if it can be
called a perverse
Some even talk about
rational bubbles, rational
crashes, rational herding
(hmm, why not rational
But if the momentum reaches
extremes, greed, fear, pride,
envy and other
emotions come into play and
A typical emotional herding
(see that word).
To call cascades "rational
herding" can be justified in a
first phase, but no more once
emotions take control.
Actually, without that emotional support,
the cascade does not reach the herding critical
threshold (see percolation) and aborts rapidly.
Information cascade that are not fed by emotion are fragile, as cold hand
players take soon the lead and start to check if there are some
fundamental realities behind it, or if in fact the king is naked. That is why,
People might obtain gains in that specific category of "trend-following",
based on the belief of hidden / private information.
This is one of the notions that is behind technical analysis.
But they might lose more than what they gained when the cascade
music stops and price trends start to revert or to go
Look at that, some cascades flow upwards!
Here are examples of two opposite long term cascades,
a continuous stock market rise and a continuous stock market fall:
In a first step, a stock (or other asset) price rise happens at random,
without a relevant exogenous event (an effect of noise traders activity
This rise is seen by some investors as a positive information.
They act on this belief. This initiates a new price rise, which in its turn
attracts the attention of other investors.
The steady arrival, every time there is a new rise, of new buyers who
think that those who bought just before them must know something,
fosters a bullish trend or even a bubble.
Cascades take place also in bearish trends and in crashes.
Some first investors get weary or cautious, even if no relevant new event
takes place to cause it.
Under such an impression, they leave the game.
They are then followed by others who take the exit.
This is how Crashes usually start out of the blue, without new
exogenous economic events.
This does not mean there are no reasons at all, as overpricing is
usually the real cause.
But the precise moment when the market will act on that reality is
rather unpredictable, there are often no discernible early warnings
that the party is over.
(*) To find those messages: reach that BF group and, once there,
1) click "messages", 2) enter your query in "search archives".
Members of the BF Group, please
vote on the glossary quality at BF polls