Pro Golf, the Stock Market, and Herding

So I am a pretty awful golfer. I spend most of my time on the course, or rather off the course in the sand and fishing golfballs out of the water hazards, so naturally I do not hit many birdies or pars. However, pro golfers get paid to sink pars and birdies consistently. They don’t get paid for playing poorly, and so they only take risky shots when they are in a good position to do so. In a recent Moneyweb article, Cass R. Sunstein brought to light an interesting fact about pro golfers: when faced with a putt of equal distance and degree of difficulty, pro golfers sink the putt with a better percentage when the putt is for par rather than for birdie. This happens because pro golfers are loss-averse. They prefer to lay the ball up closer to the hole rather than risk missing the birdie putt, having the bao-golf-tips-facebookll roll far from the hole, and then ending up with a bogey.

Sunstein made a clever connection between the volatility of the stock market and this behavior among pro golfers. He exclaims that a loss-averse participant in the stock market behaves a lot like a pro golfer in circumstances when market prices start dropping. These loss-averse investors will remove some of their holdings in order to limit their losses in the short run (like a golfer limiting their strokes on a given hole), but over time, these decisions to pull money out of investments hurt, as the small losses add up.

This behavior can spark stock market panics, like the one that occurred in February this year. Many investors start dropping stocks based on the behavior of other investors (by receiving “sell” signals via communication) even when metrics and the probability of gain are in the investor’s favor, resulting in the continued loss of value for a single stock or the stock market as a whole. Massive stock sell offs are an example of negative informational effects in cascading behavior. By following the information of others and becoming part of a cascade rather than following reason, the sellers often lose in the long-run, as more savvy investors, those who do not fall into the trap of a cascade, start buying the stocks based on fact and the stock prices go up. These informed investors are important for balancing out the market and preventing the cascade from causing a complete market failure.

We should not just take away from this the lesson that we need to just recognize informational cascades and rely solely on our own information. Rather, we, as future leaders, especially aspiring political leaders and economic experts, need to recognize the power that credentials have over the masses, and use this to not spread extreme emotions among the public, as heightened fear or excitement in a financial market can lead to crashes. President Franklin D. Roosevelt famously coined the phrase, “We have nothing to fear but fear itself”, recognizing that public confidence in the market was essential to avoiding detrimental informational cascades. However, overconfidence can lead to market bubbles, which can be equally harmful to the economy– for example, the housing bubble that burst in 2008.

 

 

 

Advertisements
Posted in Uncategorized | Leave a comment

Information Cascades Against Clusters

Looking at it from a distance, information cascades seem fairly simple. If the two people going before you picked a ball out of a bucket and declared that both the balls were blue, you go and state that the bucket is “majority blue” even if you pick a red ball. It’s simple right? Although we have used cascades for primitive examples for the scope of our class, information cascades and the herding effects that come with it impact our world drastically.

 

arabspring-tweeter.jpghttp://www.economist.com/blogs/democracyinamerica/2011/10/mass-movements

In 2011, an Economist article aimed to tackle the Arab Spring through herding. The author of the article quoted a paper written by Oregon Economics Professors Chris Ellis and John Fender, where they described revolutions and civil unrest through information cascades. They defined these cascades as a phenomena that forced people to “make decisions on the basis of their observations of other peoples’ actions” and observing signals through their behaviors. They believed that workers decided whether to rebel or not by observing their counterparts actions and though of other workers rebellions as a sign of the regime’s weakness. If enough of them organized and rebelled, then a good shot was born for the regime to be overthrown.

The Economist used this bit to explain the Arab spring, where information cascades both within countries and between countries, created an avalanche effect that took the whole world even the protestors themselves by surprise. Events like these are actual testimonies to just how effective and sudden information cascades can take over.

But we know that virtually every country on Earth has protestors and citizens that want to topple the government or establish the rule of a new leader or new ideology. So why can’t all these movements take off and use the avalanche effect created by cascades? Clusters.

Clusters are closely knitted communities of people that typically share the same beliefs about political issues. And each informational cascade has an innate threshold. A threshold is a magical ratio between the nodes in a person’s network that believe in a given ideology and the total number of nodes in that persons network. For an information cascade to influence an individual this threshold has to be surpassed and significant majority of the nodes in a person’s network should believe in that ideology. However clusters characteristically force people to share a tight group of neighbors and have densities that prevent these cascades from claiming one of their members. And the natural by-product of this clash between cascades and clusters is polarization.

Several posts in this blog have explained the polarization between conservatives and liberals in the US growing due to these clusters; and they are right. However, this phenomenon represents a more global trend.

https://www.washingtonpost.com/news/monkey-cage/wp/2013/10/10/maps-of-global-polarization/ The Washington Post published an article that described polarization in the world and in between the different cultures and governments of the world.

imrs.php

We don’t need to go into the underlying details of the map, but it basically depicts different ideologies with different colors. And the vast difference between the colors of our world is striking. This is the result of a battle between information cascades that aim to create global revolutions and national clusters that want to protect their own ways of life. It will be exciting to see how this will pay out.

Posted in Uncategorized | Leave a comment

Polarization in US Politics

In mid 2014, the Pew Research Center published an article titled, “7 things to know about polarization in America”. The findings were the following: First, over the past twenty years, the percentage of Americans who express consistent ideological views has doubled. Second, partisan animosity has increased. Third, people surround themselves mostly by people of the same political views. Fourth, communities from the left and right differ drastically in terms of religious and ethnic affiliations. Fifth, there are fewer moderates. Sixth, the most ideologically extreme people get most involved in US politics. Seventh, compromise means getting more instituted from one’s own ideology.

Unsurprisingly, the 2016 election has so far exhibited an affinity for extremes. Yes- primaries have always been less moderate than the general election, yet presidential candidates have not before called to ban all Muslims from entering the country…. While many sociologists have blamed factors such as the two party system and the media for the phenomenon of polarization, Jon Evans from TechCrunch has offered us a new way to think about this polarization: power laws.

As we learned in Chapter 18 of the class, empirical studies on web pages have shown popularity follows a power law due to correlated decision making across our populations. We like to copy the decisions of others because doing so may be the rational thing to do. This same logic can apply to political choice, which is clearly correlated across populations.

So why has polarization increased then? In an age of social interconnectedness, social networks have facilitated the display of other peoples’ beliefs and ideas. As the article emphasizes, Facebook has greatly contributed to this divide. Since people of the same regions and political parties tend to be friends, “conservatives see conservative stories, and liberals see liberal ones […] liking opinions that tell us we’re right instead of engaging with viewpoints that make us question our assumptions”. In so, “filter bubbles” are created and precipitate extremes.

Whether this is good or bad is still up for debate. However, an awareness of the perpetuation of power laws associated with social networks will hopefully give us greater insight into how to deal with the changing face of US politics (and other spheres) in the upcoming years.

Posted in Uncategorized | Leave a comment

Ant social network is spatially organized in ant nest network

This post is the third in “the ant network” trilogy; the first in the series discussed how the architectural topology of an ant nest can influence the collective behavior of ants, and the second post showed how having few talkative ants can increase the information spread within a colony. The paper cited for this last post lays down the bridge between the architectural network and the social network of an ant colony showing how the social network of ants is organized spatially within the architectural network of their nest. The authors of the paper have tracked all movements of the ants in six different colonies for six weeks without losing individual identities of ants. They attached a small piece of QR code on the back of an ant and took time-lapse photos, which were later analyzed to retrieve the spatio-temporal location and activity of each individual ant. They retrieved more than 9 million interactions and found that an ant is more likely to interact with another ant if both of them share the same ‘occupation’ (e.g. there are nursing ants, foraging ants, cleaning ants, etc). The interaction network shown below visually shows this task-biased giant components.

Screen Shot 2016-05-30 at 10.08.04 PM

The study further showed that not only did the same task group ants interact with each other more frequently, but also different task groups were occupying different locations within a nest.

Screen Shot 2016-05-30 at 10.07.38 PM.png

This implies that the spatial locations of ants within a nest make spatially closer ants to interact more frequently, which in turn made ants in proximity perform a similar ‘job’. Because spatial locations influence the interaction patterns within an ant society, we can imagine that under different nest architectures, ants will form different spatial distributions, which in turn may change the ‘job’ structure in their society. Here, we must remind ourselves that ants build their nests whose architecture influences their social structure; thus, in a way, ants are collectively determining their own society’s fate.

Posted in Uncategorized | Leave a comment

How Herding in Iowa and in the Democratic Primary Has Shaped the Presidential Race

The 2016 Presidential Election is receiving more news and social media coverage than any in our nation’s history, which is resulting in an abundance of questions from first-time voters about the logistics of a caucus and the role of superdelegates in the Democratic Primary race. These two formalities of the United States election process have had a substantial impact in influencing both party’s races in 2016. The caucuses and superdelegates have been working effectively for two parties that have tried to cement the establishment candidate while fighting off the rise of their anti-establishment counterparts. This article written in early March as an effort to explain the importance of superdelegates to Bernie Sander’s supporters serves as an example of how systematic herding of would-be voters has influenced the 2016 election up to today. If you were to google the Presidential Primary Results for the Democratic Party sometime in February, this is what you would see:

CbtNmYXUAAEmb7d

At first glance, it appears that Hillary Clinton is running away with the nomination and will win in a landslide. However, uninformed voters are not aware that the dark blue portion of the bar graph indicates the superdelegates that each candidate has garnered, instead of delegates awarded based on the popular vote. Superdelegates are individual who are free to support any candidate regardless of their results in each state’s primaries, many of which claim they will support whichever candidate wins the popular vote. So, while it appears that Bernie Sanders is at a nearly 450 vote disadvantage early in the race, he is, in reality, down by less than 50 delegates. This is, in effect, the democratic establishment using its authority to herd its voters even amidst the anonymous voting process.

Another interesting component of the 2016 primary elections is the presence of several caucuses. At a caucus, instead of voting anonymously in a private booth, members of a specific location deliberate the views of the candidates and share their beliefs, ultimately gathering in specific parts of the room to show their affiliation with a specific candidate. This introduces the possibility of herding and groupthink that could skew the results of individual locations. At the time of the Iowa caucus, there were candidates in both parties that appeared to be the de facto choice (Clinton and Cruz/Rubio) and candidates in both parties that appeared slightly too radical with which to be openly affiliated (Sanders and Trump). Interestingly, in the Republican side of the race, Trump appeared to be in line for a win in Iowa over Cruz and Rubio but ended up placing second to Cruz. This begged the question of whether or not the voting format herded people away from declaring their support of Trump due to the negative stigma attached to agreeing with his platform.

Both of these situations present different methods by which the established system of the Primary Presidential elections could have influenced the early voting results. One could argue that the existence of superdelegates is the catalyst that will eventually spur the nomination of Hillary Clinton as the candidate for the Democratic party. On the republican side, Donald Trump’s results improved dramatically when shifting from the caucus format to the conventional voting style in the following primaries. Though the primary elections are not completely over, these herding techniques undoubtedly played a significant role in manipulating the people’s votes in determining the candidates for the next President of the United States of America.

Posted in Uncategorized | Leave a comment

Retrospective Answers in Pandemic Diseases

The spread of diseases has, without a doubt, been an extremely lethal yet frustratingly unpreventable phenomenon throughout the history of mankind. The black plague, small pox, and most recently, Zika and Ebola outbreaks have decimated populations around the globe, eventually dying down after a fatal international spree; for such reason, scientists have continually tried to find new and creative ways of counter-attacking the aforementioned outbreaks. Fortunately, Dirk Brockmann – a theoretical physicist and professor of complex systems at Northwestern University – thinks he’s on the verge of finding a solution, particularly with regards to modelling and estimating the magnitude of outbreaks.

In the article, “We’ve Been Looking at the Spread of Global Pandemics All Wrong”, published by The Atlantic, the author, Emily Badger, discusses the recent progress that Professor Brockmann has had, and the present/future impacts it will have for international prevention efforts. In particular, Brockmann claims that scientists have been “looking at the map of the world all wrong”. To illustrate this, consider the following two scenarios – both models that we saw during our class on “Epidemics”. If a disease spreads solely through local traffic, then it can be depicted as a perennially growing circle of sorts – such scenario would have been the case hundreds of years ago. On the other hand, if the disease could also spread through air travel, then the outbreaks could be depicted by many of the aforementioned growing circles appearing at arbitrary locations and rapidly covering a map or a screen. Thus, analyzing the initial location of the outbreak and the riskier regions of contracting the disease next can be an arduous task for even the best data scientists. Brockmann’s solution is based on simplifying that second model in order to make it much easier to work with and comprehend.

By redefining distance between two places to be “defined by the flow of air travel between them”, Brockmann is able to depict what the world looks like from different arbitrary locations. For instance, consider the following image that depicts the world from Cyprus with the most probable air routes:

Screen Shot 2016-05-28 at 1.26.25 PM.png

Given the above representation of outbreaks, scientists can now work with a model that resembles the perennially growing circle pattern that was believed to occur hundreds of years ago. As Badger discusses close to the end of the article, the true power of this new model is the ability to pinpoint the origin of new diseases simply from looking at ‘where the ripples began’. Additionally, with respect to the basic reproductive number (R0), utilizing this model can help governments reduce both p (probability of infection) and k (amount of neighbors) – both components that make up the R0 – by employing the use of extraordinary sanitary measures and quarantines in order to bring about an end to fatal diseases.

Funnily enough, Brockmann remarks “Why didn’t we think of all this earlier?” After all, the basic premise of his innovative model is actually to look at pandemics in a similar way as before.

Attached is the link to the article and videos that depict outbreaks in Brockmann’s new model.

We’ve Been Looking at the Spread of Global Pandemics All Wrong

Spread of Pandemic Influenza – Traditional Map

Spread of Pandemic Influenza – Brockmann Model

 

Posted in Uncategorized | Leave a comment

New Network Analysis in Zika Outbreak Discussion

With the 2016 Rio Olympics only a few months away there has been much discussion about the outbreak of the Zika virus in Brazil and it’s impact on the spread across the country. This has lead to a discussion and a petition of over 150 professors at top schools such as Harvard, Yale, Oxford, University of Tokyo, Columbia, New York University, University of Pennsylvania, Princeton and others calling for a postpone in the games until the outbreak is controlled. In their petition to postpone the Rio Olympics they list such issues as:

“The evidence shows: (i) that Brazil’s Zika virus strain has more serious medical consequences than previously known, (ii) that Rio de Janeiro is one of the most affected parts of Brazil, and (iii) that Rio’s mosquito-killing efforts are not meeting expectations, but rather mosquito-borne disease is up this year.”

Zika virus disease is a disease caused by the Zika virus, which is spread to people primarily through the bite of an infected Aedes species mosquito. The most common symptoms of the disease are fever, rash, joint pain, and conjunctivitis (red eyes). The illness is usually mild with symptoms lasting for several days to a week after being bitten by an infected mosquito. People usually don’t get sick enough to go to the hospital, and they very rarely die of Zika and for this reason, many people might not realize they have been infected. However, Zika virus infection during pregnancy can cause a serious birth defects such as microcephaly, where a baby is born with a head much smaller than expected, as well as other severe fetal brain defects.

The first confirmed Zika virus infection in Brazil was found in May 2015 and on February 1, 2016, the World Health Organization declared Zika virus a Public Health Emergency of International Concern as local transmission has been reported in many other countries and territories.

Despite all of this the World Health Organization decided to reject the call to move or pospone the Rio Olympic Games over the Zika outbreak. In their response they list “this would “not significantly alter” the spread of the virus”. This comes out the same day as a new network analysis of the situation was released in Japan.

A new study conducted by a team of researchers from the University of Tokyo, Hokkaido University, and the Japan Science and Technology Agency in Japan have developed a tool for predicting the risk of Zika virus importation and local transmission for 189 countries.

Network Photo 1

Global distribution of the risk of local transmission with Zika virus.
The risk is given as the percentage of observing local transmission by the end of 2016, clored by intensity (0-15, 15-30, 30-45 and 45-60%, respectively). The origin country Brazil and countries that have already experienced case importation prior to importation event in Brazil are colored by grey.

The team predicted the virus’ potential of importation and local transmission by the end of 2016. In it they used a survival analysis model, information about airline transportation networks, and transmission data for dengue and chikungunya viruses, which are also transmitted by the same mosquito species. The model they use is similar to the SIR epidemic model described to us in class. The model goes through different accounts of country susceptibility, it’s level of infectiousness, and also emits (removes) the nodes that have already experienced case importation. This is done at a much more in depth and technical level than that of the course, however what we have learned is core to the research on the outbreak.

Network Photo 2

(A–B) List of top 30 countries with the estimated highest risks. (A) shows the risk of importation, while (B) shows the risk of local transmission. The risks shown on horizontal axes represent our estimates by the end of 2016 (week 92). Bars filled with grey represent countries that have already experienced importation of ZIKV infected case(s) by 31 January 2016 (week 46).

As this information is so recent it is unclear of the effect that this new study will have on the Rio Olympics situation. However it is very interesting seeing a epidemic network analysis being done on a global scale that is relevant in all countries. Especially a day after we went over Epidemics in class.

 

Posted in Uncategorized | Leave a comment