This happens to be a very special blog post for me. An Italian version of the present article can be found here. A tribute in prose has been written on a whim elsewhere by Yours. Finally, a beautiful tribute in verses was penned by Justin Bold, another participant and is available on this page of the blog.

The beginning

My interest in risk and extreme events dates back to 2014, when I first came into contact, a short while after emigrating to Poland to pursue a scientific career in physics, with Nassim Taleb’s books on uncertainty and risk management.
Beside being an accomplished former trader, Nassim comes across as a talented and provocative writer. I was sold to his ideas as soon as I was done reading the first book (not in chronological order), The Black swan, which is the one who gave him his notoriety.
It wasn’t long before I started to want more, so I began browsing the web, typing his name in the Google search bar and following the links it returned in order to learn more about the man and explore his intellectual world. It wasn’t long before I came across the website of the Real World Risk Institute (short version for later: RWRI), an educational institution created by Nassim himself and people of the like of Robert Frey from Renaissance Technology and Raphael Douady, formerly Frey Family Chair Professor in the applied mathematics department at Stony Brook university and founder of the risk software firm RiskData, a name retained by his present consulting company.

The educational program, held in new York, bursts with insights on the conceptual content of all facets of risk, from finance and insurance to cybersecurity and climate change, with a particular focus on the much and sadly neglected extreme events, popularly known as Black Swans thanks to Nassim’s literary effort. Such occurrences can be defined, broadly speaking, as the class of happenings which escape the 101 modelling through the so-called normal distribution which is taught in our schools and is too widely used even at the institutional level (see large banks), only because it provides the illusion of being able to spit out quantitative predictions which are only occasionally wrong…except for the fact that, even forgetting about theoretical intricacies, when such events which are deemed impossible by the naïve approach do occur, they cause losses and distresses it is hard to recover from for those who had not anticipated them: again, see American banks in the financial crisis of 2008.
So, the key “thought-rewiring” one has to go through when or after attending the program, as I see it, is that in real life it is not probability of events that matters, but their outcomes. We are far too mind-numbed by school examples of dice games, where all outcomes are equally likely and equally impactful, to realise that real life is a different story, much more irregular in probabilities and more brutal in results.

As otherwise put, extremely effectively:

Never cross a river if it is only 5 feet deep on average!

Nassim Nicholas Taleb

So I clearly retained this key teaching inside my mind after June 2017 and it kept brewing inside of me for a long time afterwards, even having at first lost touch with this community.

But then, in June 2021, an email from the RWRI program coordinator popped up in my mailbox. It sketched the plan of a deep-dive program on the more mathematical aspects of risk theory, taught by Raphael Douady and also featuring Pasquale Cirillo for one lecture on so-called fat-tailed statistical distributions; I was unable to attend it that summer because of my work duties, which customers and reader of this blog may well imagine, but I subscribed to the following edition of the program, held last February.

Little did I know that the RWRI folks had started consolidating in a side community, with the new trademark brand R+, on a bunch of various thematic Signal groups moderated by Raphael and focused on investing, geopolitics, cryptos and the like, of which I became an enthusiastic member upon attendance of the program, after tasting the unusually high quality of the discussions held therein; neither did I know that Raphael had plans to start an yearly program of in-persons meetings and that the first edition would be held this very October at Tenute Carrisi, a 1-hour drive away from Torre dell’Orso!

The meeting

That was, in a nutshell, how I found myself helping the organizers set up the event and how Dentoni itself became a topic of the series of presentations we listened to during this exciting week: the presentation of our company, which I had the privilege to deliver in the wake of our delighting visit to Grotta della Poesia, led by Mimmo Lorusso, made it to the Saturday morning slot and was focused on the theme of opportunities and risks of family companies, of which we have been a case study for 41 years now.

Let me try to give you just a small mouthful of the topics we listened to. I’ll try to illustrate them concisely in the following sections. However, if science, cryptocurrencies and insurance are not your thing, you can skip directly to the description of Vivarch’s contribution to our municipality, which is certainly worth your time as a citizen and is the last part of this post.

Crypto assets, parallel societies and deep learning

My talk was preceded by much more impactful topics, which it is worth understanding for anybody wishing to better navigate a complex world like ours, possibly catching investment opportunities: Juraj Bednár, an accomplished Slovak hacker, taught us about his experience with crypto assets and the parallel society he and his peers are trying to build. This is a community centered around a crypto-based economic system and sharing the libertarian values which informed Satoshi Nakamoto’s motivations to propose Bitcoin (BTC).
BTC is the first cryptocurrency and about it very much has been and is being said all the time in the media, but very little is understood by either journalists or the general public. Their project, called Paralelni Polis, is extremely attractive and futuristic, certainly ahead of our present economy in terms of digitalization and use of blockchain technology, but it was enough to have me buy his book and have him teach me how to perform a fast Bitcoin transaction on the Lightning Network, on which I surely plan to learn more about.

Staying in the crypto world, but with a more artistic tilt, Andrea Marec from Reasoned Art taught us about the digital transition of art to the NFT (Non Fungible Token) realm, which has just begun but will certainly progress further, much like the transition of our economy to one where the role of crypto assets in daily life is sensibly higher than now. At least so I convinced myself after listening to him and Juraj…
Digital art is a subject worth at least one book in itself; however, if you are interested in getting the feeling of it instead of gobbling countless words, take a look at the Dreamscope website: these are all pictures of the likes we take with our smartphones, fed into a computer algorithm of the Deep Learning class. Buckle up, since what I am about to say will provide me with a natural connection to the topic of another talk.
Deep learning is a branch of computer science, more emphatically known as Artificial Intelligence, concerned with the task of having programs learn to recognize patterns within tables of data…well, roughly speaking. But since digital photographs are -for a computer- nothing but tables of colored pixels, one can feed the paintings of a given artist into such an algorithm and have it learn how this painter used to deal with color hues and shapes; once the learning task is done with, the algorithm can modify the said borders and colours of new images in such a way as to represent their subject with the style it has “learned”. Let me post here an example of a picture of mine with my cat on my lap, while I am drinking coffee…it was done with Dreamscope, asking it to apply the pre-packaged “seaside” style, on which the company has trained its servers, to the original picture on the left:

Now tell me again that digital art has no potential ! And this was just a trivial example…

I must say that it is always extremely interesting to be challenged to think about such a deep rewiring of a human expression we have been used to associate to specific communication vehicles over the centuries. Visual arts have always been associated to canvasses (painting), blocks of marble or other “carvable” materials (sculpture) or buildings (architecture)…in a word, to physical objects directly perceivable through our eyes without the support of an electricity-fueled machine.
Can we, now, re-think this as a non-necessary assumption, in the context of the anthropological transition brought about by the digital age? Can we stop and reflect on what it means for us the difference between an original painting and an almost identical copy of it? How does this change the function of art as a store of value, as opposed to just a mean of eliciting emotions in the observer? This is, broadly speaking, the problem that the NFT certificate solves once for all through the blockchain: immutable, certified uniqueness.
Somehow, this must be similar to the kind of debate which certainly occurred among our proto-historical ancestors when writing was invented and the knowledge that used to be transmitted through purely oral means was slowly entrusted to the written word, surely enraging the back-then purists of one-to-one human interaction. We all know how it went and now oral teachings now coexist with tons of books. And luckily so…

So I can now imagine a future where strictly physical art will not disappear, but where the importance of creations from the digital realm will increase, especially as stores of value. In a sense, digital art emerging as an increasingly important store of value feels to me like books having become easily accessible storage devices of knowledge: they have the advantage of being readily available and immutable, though they cannot replace the live interaction with a real master of the discipline they are concerned with; that is why universities still exist, just to give an example. In a similar way, digital art will not cancel, in my opinion, the importance of physical art, but will coexist with it, slowly gaining importance as store of value due to the immutable nature of a blockchain-based certificate, much like a book by a dead master is often much more worth trusting than a lecture by a copycat.

The coronation of the crypto triplet of talks came through a Q&A session with a world-class mathematician, Ricardo Perez Marco, who has devoted a lot of his research and papers to the mathematics of the Bitcoin network. The session had its highlight in the discussion of the differences and pitfalls of Proof of work and Proof of Stake, the two consensus algorithms which determine how blocks are added, by a majority agreement among the computers checking network transactions, to the immutable ledger which constitutes the blockchain itself. There are substantial differences in the protocols implemented by, for instance, Bitcoin and Ethereum, the two largest cryptocurrencies by market capitalization and number of users.
We cannot delve deeper here, but I trust you believe me when I say that Ricardo is a guy who knows his Bitcoins!

Deep learning and Data Science

Now something more about Deep Learning: Patrice Kiener delivered a riveting talk about neural networks. These stand to deep learning just as buildings stand to architecture: they are the tangible part of the discipline.
I should stress again that, since this is not a scientific blog, we do not have the space to dive into details, but these are among the most sophisticated algorithms ever produced by humans. You have probably already convinced yourself that it is the case by looking at what they can do through my picture above.
But there is a “backend beauty” to them which lies in the staggering conceptual simplicity of their engineering, featuring only elementary multi-variable calculus and matrix algebra. The capability of these programs to detect patterns in data is unprecedented in the history of computer science…even excessive, if overused, for they can easily spot regularities where there are just coincidences. That said,
I think this blog post by Andrej Karpathy, former director of AI research at Tesla, the pioneering electric cars company which invests a ton of money into self-driving cars technology, summarizes pretty well how, in principle, these class of algorithms could really trigger a revolution in the way we think about computing. And, if you are really curious about the perspectives of this fascinating branch of computer science, I recommend complementing Karpathy’s post with this video by Derek Muller from his Veritasium YouTube channel, which I highly recommend in itself to anyone seriously interested in learning some science through very entertaining explanations.
Back to the main plot: Kiener is a chemical engineer who moved to Artificial Intelligence research a while ago, founding his own consulting company, Inmodelia. As a former physicist turned manager, I cannot buy instinctually sympathize with career roamers, for I think their ability to leverage multi-disciplinary thinking is a definite plus.

Attention please: next few lines are a bit tough. If you think you may get discouraged and drop the article, there is no reason: jump over! Otherwise, if you are not too sensitive, go ahead! 🙃

Self-indulging jokes aside, the purpose of his talk was to explain how neural networks are in one-to-one correspondence with traditional, statistical non linear models, though they rely on a different parameter-tuning method, due to their very design. He also focused on the problem of training models with a very limited number of input data, and specifically so with data featuring some level of noise, which make the statistician’s task much harder. Let me put this straight: think about the original picture of mine above, forget about the restyling and imagine that some objects were plugged into it via Photoshop editing: could you teach an algorithm to spot such forgeries? The answer is that there are methods to do this, even if you have a limited set of data tables to train your machine. But disentangling noise from signal is generally pretty tough and requires a recursive series of attempts at framing the data initially…there would be much more to say, but computing has likely already overstayed its welcome for the time being.

(Non-)Insurance against systemic risk events

Another real killer was delivered by Arié Haziza, member and senior lecturer at RWRI and head of catastrophe Risk Analytics at AWAC, a reinsurance company. I did not even know, before listening to him, what a reinsurance company is. I learned, therefore, that it is nothing but a firm providing insurance to regular insurance companies, in an attempt of the economic system to soften the blow it suffers when a rare but devastating contingency happens: think of an insurance company with 50% market share in the domain of housing insurance in a geographical area which happens to be hit by a hurricane, as recently happened in Florida; this means that the rebuilding of half of the damaged houses must be paid by the insurance company, if the risk of devastation of the premise by extreme natural events is a provision of the contract.

Now, moving from this example, I will try to deliver Arie’s key message in three steps:

• First: the basic fact to understand is that insurance can work only if the insured events are reasonably independent: everybody pays a little bit to make sure that a few unlucky guys can get covered for their losses or the damages they incur when they are the ones hit by the unfortunate circumstance. This is easy to grasp.
  
• Next: what happens in the case of events such as the aforementioned hurricane? Well, since the insurance company surely takes a substantial loss, having insured many houses in the area and having, therefore, seen its de-correlation through diversification risk spreading fail, part of the loss is absorbed by the reinsurance companies.
  The latter also spread their risk: to put it very simplistically, they make sure to re-insure insurance companies which in turn insure people in various geographical areas. Likely, only one such area at most will be hit by a hurricane in each given year, so this will guarantee (or should, ideally) that neither the insurance company nor the re-insurance firm will go bust: in fact, the first will be compensated by the re-insurance company, the latter will make profits in all the other geographical areas not hit by a hurricane and, therefore, not generating a claim from insurance companies.
  
• Finally, and here is the key point: what occurs in such events as pandemics, where de-correlation of risks is totally lost and people dying in Wuhan from a virus can die at the same time in New York of the same ailment? Or when live events of the likes of concerts and conferences, which generate significant cash flow, are cancelled worldwide for the very same reason?
  There is only one answer: the system is unable to insure such losses. There is simply a limit to what the system can insure without going bust. Extreme risk can be hedged only so much.

Arié is a charismatic and enthusiastic speaker and I really got a booster of my understanding of the insurance world after listening to him. If you did not know about this subtleties, I hope I managed to effectively paraphrase the guy.

Raphael Douady on why Black Swans do happen

So finally we come to the boss’ presentation (thumbs up Raphael ! 🙃) . There were quite a few equations in there, but I am not going to bug you with those, so let me be more conversational.

I am not sure you want to read about quantum mechanics, but sure as hell I love talking about it, being a quantum physicist. Beside, quantum mechanics has an indirect connection, in what I want to tell you, to financial market crashes anyway.
Please find below an illustration of one of the many, many aspects of quantum mechanics which appear outright paradoxical to the untrained mind, dulled by this coarse, macroscopic world we spend all our lives in: if you have a classical ball rolling towards a hill -ideally with no dynamical friction, to keep the math simple- only two situations are possible in classical physics, i.e.

1. The ball has enough energy to compensate for the energy loss due to the gravitational pull it is subjected to and, thus, it can make it all the way up to the pinnacle of the hill and roll downhill on the other side.
   
2. It does not. Period.

That is not the case in quantum mechanics: in fact, transistors, which are among the most fundamental components of computer chips, are built on the very phenomenon that, in front of a potential barrier, some particles will come through and some won’t, even though their energy would not be enough to climb to the top of the energy barrier classically. This is due to the fact that particles are such only when an observer looks at them by making them interfere with a measuring device, something which somehow (but nobody knows precisely how) “collapses” their wave function.

The concept of wave function is really a game-changing concept, something that makes you grow overly pride of yourself when you study quantum physics. Roughly speaking, it means that, as long as you do not spot a particle with a measuring device, all you can say about it is that there is a probability cloud of sorts, spread all over space, where the particle may be located with varying degrees of probability, depending on the location of the speck of cloud where you might want to look for it with the said measuring device. Now, when a probability-cloud-still-to-turn-particle hits a potential barrier (see the red tunnel in the figure above, which is the illustrative equivalent of our cloud), it is spread all over space, including the other side of the barrier. Therefore, it might as well be that, once in a while, after the collision, your particle will be located in that left (as per the figure) tail of the cloud which made it past the barrier.
So, let’s say you throw a quantum particle onto a very, very high wall many times while it has an energy which is largely insufficient, classically, to make it through the wall. Quantum mechanics predicts that, once in a while, your particle will make it through the wall !
Wait, you might tell me: how come this never happens for a ping pong ball ???
I will be very hand-waving in answering, but the basic idea is that it is simply because the wave function of a ping pong ball is a superposition of the wave functions of all its many many particles and…since the probability of the whole ping-pong ball crossing the wall is proportional to the product of all the probabilities of its particles crossing it (not really, these particles are bound together by molecular interaction, but you get the idea)…it’s very unlikely, because we are multiplying a huge number of very small figures, so we get something really infinitesimal.

It is like saying, if you want: what is the probability that all among a zillion monkeys pass an 100-questions English test? Since they will type answers randomly, there is a tiny chance for each one of them to pass it … but then what about all the others?

Still…the probability is strictly non zero as soon as the probability for just one of them is!

Monkey actually make for a very good metaphorical parallel with the subtler case of bond molecules in a ping-pong ball or in any solid macroscopic object, for they tend to highly influence each other: a monkey is, metaphorically, the quintessential copycat, so the distribution in answers will tend to herd towards a given pattern, much like bond molecules tend to move altogether. How high are the chances this will be the right pattern of answers?

So, once you understand such subtle phenomena, as I was mentioning, you feel at quite some height above the average guy: not only you understand the most counterintuitive form of randomness known to man in nature, but you also know why we do not experience it in real life !
But then Nassim Taleb comes along and he knows how to hit you hard:

“If there is one thing on this planet that is not so uncertain, it is the behavior of a collection of subatomic particles! Why? Because, as I have said earlier, when you look at an object, composed of a collection of particles, the fluctuations of the particles tend to balance out.
But political, social, and weather events do not have this handy property, and we patently cannot predict them, so when you hear “experts” presenting the problems of uncertainty in terms of subatomic particles, odds are that the expert is a phony. As a matter of fact, this may be the best way to spot a phony.”

Excerpt From Chapter 18 of The Black Swan

Quantum randomness may, indeed, be the most counterintuitive form of randomness in nature, but it is by far neither the most dangerous nor the most impactful, unless you are concerned with an academic career or are into something like hardware engineering.
If it comes to broadly meant real life, the golden medal goes to a different kind of randomness, associated with extreme events, famously named black swans by Taleb in his hit. Technically, the observer has a role here as well: a black swan is an event for which the statistical tools used by the observer of the system are inadequate to its inclusion in the range of possibile events, but let’s not bee too sublte here. Better take a look at what they look like instead:

So welcome to the world of complex systems. We can forget about quantum physics, for classical one is enough to make these kind of messes come true. It is not just rogue waves of course: think of hurricanes, tornados, pandemics or stock market crashes, what Raphael specializes in.
Take a look, for instance, at the 1929 crash of the Dow Jones Industrial Average index:

So how come such phenomena come about, when the movement of water masses or traders behaviour can be modelled with equations from classical physics?
Let’s let Raphael explain this using a simple behavioural example we can all relate to, instead of bringing fluid dynamics into play. The only underlying assumptions are that traders and investor

1. always try to maximize their personal financial gains.
2. tend to believe that others have more information about markets than themselves.

If it sounds like I am saying that humans tend to be greedy and suspicious that others will leave them behind, so that they are prone to herding instead of thinking on their own, that’s exactly what I am saying, especially when it comes to money. And I doubt anyone will marvel too much.
So let’s see what happens when, starting with a healthy economy, people push it too much, keeping it very basic. So traders and investors see their peers making profits and they want it for themselves too. Since there are stocks which perform very well, everybody wants to purchase them when things are good, so people flock to certain kinds of investments, which tend to make them, consequently, even more attractive to others, even those who lack the money to do this. So the latter, since the economy is initially doing well and interest rates are not so high, start to borrow money to purchase the same assets or to spend more than they can earn…debt and leverage slowly but surely grow, up to the point where all of this becomes unsustainable. So, once in a while, a strong correction ensues and what you see in the picture above may happen…and does! This is just the most catastrophic event that the American stock market has experienced so far, but there have been hundreds of similar occurrences all through the history of financial markets.
So, all you need to understand such phenomena is that individuals in a complex system such as the economy try to maximise their own utility in the short term…pretty simple and intuitive, no walls crossed by virtue of some quantum mechanical magic. So where does randomness come from in all this? The answer is: nobody can predict how big the financial bubble will grow before popping, when it will pop and how much damage it will inflict to the economy when it does. There is an entire branch of statistics dealing with such extreme phenomena, but this is not the place to discuss it further.
If you would like to delve deeper into this subject, it turns out that Ray Dalio, a legendary investor, has summarized this economical patterns in an excellent educational video which is certainly worth your time. And if you are mathematically minded and curious about Raphael’s contributions, you can perhaps take a look at some of his research papers or his book about a method for financial markets analysis called Polymodels Theory.

Torta Crêpes®, fractals and coffee: you can call it a day !

On Friday, October 14th, after visiting Grotta della Poesia, all participants were offered nice merchandise in the form pairs of espresso cups and related saucers, as souvenirs of this opening event.
The motifs printed on the cups were designed by Laina Loreau-Douady, a London-based, accomplished brand designer and Raphael’s older daughter. The merchandise was delivered during an aperitivo+Torta Crêpes® refreshment which we were delighted to offer to our fellow participants.
But the most interesting part is the story behind the motifs. These represent the self-similar, fractal structures discovered by the legendary mathematician Benoit Mandelbrot, together with Adrien Douady. Such recursive structures are found in the behaviour of many complex systems in nature, not least financial prices.
Adrien accidentally happens to be Raphael’s late father: as he smirkingly mentioned during the week, “Mandelbrot did the plots and my dad did the math”. 😂

For my next picture featuring my cat on my lap, these will definitely be the cups to go with! 😀

Grotta della Poesia, its archeological treasures and Vivarch Aps

The world-class fame of Grotta della Poesia started about 10 years ago, when a world-class magazine published a list of the 10 most beautiful natural pools in the world, in which it was included.
I personally remember when, before that article came out, it was a paradise known only to locals, where one could enjoy diving and swimming with very little company around all year round, except maybe for the very central days of August. But after that release, mass tourism suddenly started and the fragility of the soil, which is made of nothing but brittle limestone, was seriously put in jeopardy.

The cultural heritage brought to light near Grotta della Poesia owes its fame to the late Professor Cosimo Pagliara, a prominent archeologist from the University of Salento (back then University of Lecce) who bumped into protohistoric carvings on the stone walls of a smaller cave nearby, connected to the main one by an underground water canal. After spending some time trying to decipher them, these inscriptions were revealed as prayers to the God Tautor, pledging a sacrifice to Him in the event that the person praying were to make it back from the trip to present day Albania. Yes, because the channel of Otranto is the tightest sea corridor in the Eastern Mediterranean sea; therefore, it has always been paramount for commercial exchanges. In a day with a clear sky, one can easily see the Albanian mountains over the horizon.
So Cosimo Pagliara started to wonder where the remnants of such sacrifices were to be found and he was stubborn enough in his pursuit as to keep digging around, despite alleged discouragement coming from all of his colleagues. Turns out…he was right, for the amphoras that once contained the oil and wine offered to the divinity were found nearby in the village of Roca Vecchia, the part of the archeological complex characterized by a medieval layer superimposed to the Messapian site.
The Medieval village was already much more known to the general public because of the tragedy of the siege and defeat of the Roca defences by the hand for the Ottomans in 1480/1481, in their expansionist wave which started after the fall of Constantinople in 1453.

There would me much more to recount, but this would spoil the pleasure of personally visiting the site and being told the full story by an expert local guide. Much more interesting, at this point, is to understand the experience of Vivarch Aps, the non profit organization led by Mimmo Lorusso who -in cooperation with the previous administration- has taken over the site, imposed a fee on tourist entrances, prohibited bathing and pledged to use the proceeds to the cultural development and environmental preservation of the Melendugno municipality.
Well…that is essentially the business model in a nutshell. And it’s unique in Italy. To realize whether it actually works, it would be enough to observe how much value the archeological site of Roca has acquired in the last few years. As a local, I can testify that it was quite a leap forward!

As a representative of our company and as a citizen of Melendugno, my sincere wish is that these people are allowed to continue their effort and, as anticipated elsewhere, to bring their business model to Torre dell’Orso, keeping its core idea in place and tweaking it just as much as necessary to preserve the tourists’ possibility to enjoy our beach and bay, once the undisputed pearls of the Adriatic coast and now in danger of being wiped out by coastal erosion, normally a natural phenomenon, but in this case a second order effect of unthoughtful human intervention. So we stand with Mimmo and we will continue, as a company, to voice our support in all possible ways !

I must take the opportunity to conclude this post by thanking all the R+ meeting participants for bringing their precious experience to Italy and for sharing it in their talks, but especially in the informal discussions we had all through the week, which were surely the best part of this community-cementing effort.

It was a honor of mine to help this all happen and it was the pleasure for Dentoni to host all of you.

’till next time, which we all hope will be soon enough !

Mirko Serino