Storyboard

Since physicists began to model problems in our environment, it was increasingly necessary to include the behavior of the people who intervened. This started even with basic operating processes of machines in particular when they were used by people. For example, the way the machine works depends on how we press and / or orient it in space.

At the end of the Second World War, many physicists were laid off and sought their opportunities in industry but also in services. One of these items was the stock market in which economic processes are not modeled but rather as people perceive them. A stock rises or falls in value depending on whether the person believes that the company will do so in the future. This depends on the performance of the economy, on the sector in particular, on the new technologies that can be operated, but also on how the client will decide to buy the product or not. In the end everything is the behavior of people the key to forecast correctly.

>Model

ID:(1603, 0)

Image

>Top

One of the curious themes that I discovered working on modeling the buying behavior of Hilti customers was the fact that it was key that the machines were red. The color factor determined 40% of the purchase decision against a series of other parameters that had to do with the technical superiority and services that the company offered.

Curiously, a few years ago I discovered that in a television series the uniform that the band wears (the 'Paper House' series that can be seen on Netflix) are very similar to those that the Hilti technicians had when they went to client (I did not have the cap) 'Hilti red'. In the same series, Hilti equipment is repeatedly observed in the background, making it clear that it was a product positioning (technique in which one shows a product 'casually'). Even though I was working at the headquarters if I had to advise Hilti Spain and I discovered that the local team was highly motivated and creative so I was not surprised to find something like a photo of the signing of a contract online.

ID:(11910, 0)

Image

>Top

The conclusion that the color red was key was achieved by putting together an 'expert system' that was what artificial intelligence systems were called in the 80s of the last century. For this, various products are characterized by objective-numerical characteristics (example power, noise level, durability, etc.) and subjective (example color, economy, etc.). Then they look for 'clusters' that can be understood as agglomerations of data around product profiles that are associated with the probability of being selected in the purchase.

In this way the models are used to understand how they participate and in what proportion the different parameters that characterize the product. They can also be used to develop product profiles that do not even exist yet seeking to combine the most successful parameters.

In the case of the red color, it was discovered that this is associated with the fact that the most expensive Hilti machines are those with the most important supervisors, foremen and operators. So color is associated with status, with being someone.

ID:(11911, 0)

Image

>Top

When the second world war ended many physicists were laid off. They looked for opportunities in the industry but also in services such as banking and the stock market. In the latter case, they found a universe full of data and very few models that would allow us to infer how share prices would vary.
A book that describes this history of other physics is:

The physics of Wall Street : a brief history of predicting the unpredictable
James Owen Weatherall
Houghton Mifflin Harcourt, 2013

ID:(11912, 0)

Image

>Top

The activities can be structured today into three main themes

• find and analyze data
• hypothesis-based mathematical models and data analysis
• artificial intelligence type systems that allow modeling but also obtaining results

In the first step in the stock market, the evolution of stock prices was analyzed.

ID:(11913, 0)

Image

>Top

One of the first conclusions was that many distributions were close to Gaussian distributions. In many cases the behavior was quite random and could not be predicted other than as evolution probabilities. From that time is the notion that the job of stockbrokers of that time was nothing more than rolling dice, that is, completely random. Burton Malkiel spoke of how even a group of monkeys that throw dice could be better stockbrokers than the real ones that existed at the time.

Mas en:
Random Walk Down Wall Street
Burton G. Malkiel
W. W. Norton & Company, 1999

ID:(11914, 0)

Image

>Top

Once the data began to be analyzed and work was done on the first economic and personal perception models, the first algorithms that helped investors emerged. Over time the models became so reliable that they began to be automated. The first robots or bots emerged that autonomously compared and sold stocks.

ID:(11915, 0)

Image

>Top

Over time, the first catastrophes arose in which the models erred and the bots aggressively replicated the errors, creating catastrophic falls in the stock market. The search for the cause of these true financial catastrophes that affected the global economy began. Over time they realized that the information on which the models were built was too restricted and that there were events less likely to be ignored. Its emergence destabilized the models that were beginning to enter vicious circles type in which a drop in price leads to sale, panic to sale, panic to more sale, the largest sale to decline in the price etc.

ID:(11916, 0)

Image

>Top

The use of very limited data series meant that the systems were considered to have Gaussian distributions, which was not correct. Less likely events were found to cause nonzero values to exist far out of equilibrium. This came to be called fat tails and events were named by Taleb as unlikely events such as finding a black swan.

More about the events 'black swans' in the book that gave rise to the name
The black swan: the impact of the highly improbable
Nassim Nicholas Taleb
Random House, 2007

Note: There are people who speak that the pandemic is a black swan, that nobody could provide it. This is not correct, for years it was feared that there would be and even governments prepared. However in recent times these dismantled these preparations they consider that it was unlikely and not listening to the experts.

ID:(11917, 0)

Image

>Top

The key is to use the scientific method rigorously. That means:

• develop a model, conclude that it implies, verify that it fits with the measured data and, if so, structure a way of applying the model
• aggressively search for new data and verify that the model continues to reflect what has been observed
• review hypotheses before new information and thereby correct the models
• correct the systems in application so that they always reflect the state of the art.

ID:(11918, 0)

Image

>Top

Once the models are sufficiently precise, new strategies can be explored. This leads to innovate, to create new products and processes that have the potential to generate profound changes in society.

ID:(11919, 0)

Image

>Top

An example of innovating and which meant that one of the contributors to 'String Theory' in physics became a billionaire is the Hedging model to which Jim Simons contributed. Hedging means insuring and what the method does is to look for groups of actions that among these are 'insured' that always win. In general, the mix compensates for fluctuations in the economy so that in every situation there is always a group of shares that yields enough profit to compensate for what happens with the other shares.

A simple example could be buying shares of a bus company and an airline. In good times everyone flies, the airline is doing very well while the bus company is not doing so well. The pair is appropriate if the additional gain from the airline is much greater than the loss on the bus line. When the economy goes through a bad time, people travel on the bus line and those are the actions that now dominate. Again they must give more income than the airline means in losses. In this way one never loses, it is said that the fund is insured or compensated or 'hedged'.

ID:(11920, 0)

Image

>Top

Another simple example was implemented by my son Klaus Stefan Gerber who sadly passed away a year ago. He found that fluctuations similar to the stock market occur in sports betting. There are minutes that fans of a team dominate the market by offering bets in which one wins more than they lose (example, you pay 3 against 1 - if you win you receive 3, if you lose you pay 1). Over time the opposing team also goes through a phase of optimism in which they also offer bets in which one wins more than they lose. If one opportunely closes both bets, he ends up in a situation where he always wins (in which he wins, he gets 3 and in the other 1 is lost with what would be effectively won 2).

For this type of bet, however, one cannot be permanently pending, you must use bots, robots that permanently observe the market and take the bets every time the pairs that allow hedging are generated.

ID:(11921, 0)

Image

>Top

One of the observations my son made during the process of building and operating the bots was that there were sports in which one observed parameters that are associated with the ability of the players, such as tennis. There are others in which the player consciously decides to reduce his capacity, such as jockies in horse riding. The owner of the horse decides to 'take care' of the horse, that is, he lets it participate but does not seek to win by waiting for a better time when he bets on his horse. In soccer the subject is even more complex since there is a dependency on some players who can be sent off. In all these cases, true black swans appear, which means that there are dynamics that have not been observed before.

In summary, it becomes necessary to model human behavior in order to obtain useful modeling results.

ID:(11922, 0)

Image

>Top

One of the few more systematic models that tries to describe how we make decisions comes from Daniel Kahneman, the only sociologist with a Nobel Prize (Nobel Prize in Economics because his theories help to develop the economy). In Kahneman's work there are two themes that stand out:

• Human beings think on two levels, a fast impulsive one without further meditation and a slow second in which analysis predominates. This is much used in sales processes that seek to pressure the client to make the decision impulsively and not meditate on their decision. The mechanism is probably evolutionary and designed so that for stressful situations we can react quickly to avoid danger.

• In general we feel much more the loss that we are happy about the gains. This means that we tend to avoid risk when it comes to winning, that is, we prefer a sure profit to risk to win more. However we have difficulties to assume losses and we tend to take more risks to try to avoid losing even though the possibilities of avoiding the loss are remote.

The recommended book is:
Thinking, Fast and Slow
Daniel Kahneman
Farrar, Straus and Giroux, 2011

ID:(11944, 0)

Image

>Top

Today physics has become accepted within economics as a key part of what advanced economics is becoming. In its origins the economy worked in a very similar way to how it worked in physics in thermodynamics. Variations of the different parameters and their relationships were studied.

Today there is a more general theory that as always application delivers all the laws of thermodynamics. This is statistical mechanics and its main tool is the so-called partition function, which is a function that is calculated from the microscopic equations of state, obtaining all the macroscopic properties of the system. Something similar is happening in the economy where partition functions are calculated today from which the factors that describe the economic situation are obtained. The discipline is called within physics as 'econophysics'.

Additionally, it is possible to describe the behavior of people and their behavior in what is called 'sociophysics'. In this case, as Feyman said, sociology does not conform a science with a theoretical body with universal laws, as economics does. It is for this reason that sociophysics in this case constitutes the first theory as such and that it can engage with the broad empirical system that sociologists manage and model phenomenologically.

Some books are:

Econophysics and Sociophysics: Trends and Perspectives

Bikas K. Chakrabarti, Anirban Chakraborti, Arnab Chatterjee

Wiley-VCH Verlag GmbH & Co. KGaA, 2006

Classical Econophysics

Allin F. Cottrell, Paul Cockshott, Gregory John Michaelson, Ian P. Wright, Victor Yakovenko

Routledge, 2009

An introduction to econophysics: correlations and complexity in finance

Rosario N. Mantegna, H. Eugene Stanley

Cambridge University Press, 2000

Sociophysics: A Physicist's Modeling of Psycho-political Phenomena

Serge Galam (auth.)

Springer-Verlag New York, 2012

ID:(11923, 0)

Image

>Top

Today physics has become accepted within economics as a key part of what advanced economics is becoming. In its origins the economy worked in a very similar way to how it worked in physics in thermodynamics. Variations of the different parameters and their relationships were studied.

In the past, data and models formed the basis for working with physics tools on topics that are not associated with traditional physics. However, there are two situations that make it necessary to add a third element: artificial intelligence systems. The reasons are:

• The existence of areas to model that cannot be taken to numbers and that must remain in the subjective area related to how people act. For this, neural networks allow to model and use algorithms that can be integrated with objective models.

• The need to explore attractive solutions that allow models to be used to their advantage. In the past the models were simple and the relationships well defined so that the optimal ones were easy to detect. Current models are not easy to interpret and require the tireless search of AI systems to shorten the interesting alternatives.

Note: Google's Tensorflow tool is indicated here, which is one of the most powerful tools for structuring AI systems.

ID:(11924, 0)

Image

>Top

One of the unfortunate situations in which physicists participated was in how those who favored the United Kingdom leaving the European Community won. At the time, Boris Johnson's advisor, Domic Cummings, hired this group, which apparently formed the AIQ company, to study the electorate and how it could be influenced. They probably obtained data from Facebook via the Cambridge Analytica company and directed not necessarily true messages to easy-to-manipulate groups to achieve a sufficient majority. The profiles of insecure people, their preferences were analyzed and the appropriate message was sent to trigger their adherence to the cause of leaving. For this, an artificial intelligence system was modeled and used to detect and then direct the attack.

ID:(11925, 0)

Image

>Top

The artificial intelligence system is created

• structuring the data you are going to learn from including multiple support planes,
• defining the type of result you are looking for
• establishing a criterion so that the system knows its performance
• know the rules with which it operates

With this information the system begins to form the network that links the different planes (neural network) creating a system in which information at the base is brought to the result that is considered correct performance. The multiple planes are described as deep learning.

Once the system has been designed and tested with a second set of data, it can be used to directly advise or explore alternative solutions.

ID:(11926, 0)

Video

>Top

The video shows an artificial intelligence system that learns to play the ATARI Breakout game. The system only has the rules of the game in the sense that it must destroy the wall hitting the ball that it can propel with the racket. The advance is:

• after 100 games she knows how to use the palette and manages to take the wall apart
• after 300 games he does it better than a human being
• after 500 games he discovers a new strategy that was unknown: create a tunnel and make the ball pass to the space behind the wall where it is by multiple bounces effectively disarming the wall

ID:(11927, 0)

Image

>Top

In this way we have a method that includes working with data, formulating models complemented by the AI system and finally using it to advise or find an optimal solution.

ID:(11928, 0)

0

Video

Video: The Other Physics: human behavior