Navigazione – Mappa del sito

HomeNumeri84ArticoliWhat is a financial crash?


What is a financial crash?

Some remarks on the ontological and epistemic preconditions
Emiliano Ippoliti
p. 7-24


What is a financial crash, and why does it happen? The answers to these fundamental questions require an investigation of the ontological and epistemic state of the financial markets which will identify the causes of a financial crash, the entities involved, and the relations between them.
To this end, I examine several theories on financial systems which have conceptualized financial crashes. I analyze how these theories: a) identify different causes of a crash; b) deal with the basic entities and units of analysis; c) define what we can know about those causes and entities, and what they can reveal about an impending crash. On this basis, I outline some ontological and epistemic preconditions that are necessary for a financial crash to occur. I also examine an example of a price drop on the stock market and the role played by the epistemic and ontic states of the market in that event.

Torna su

Testo integrale

1. Main issues in financial crashes

1What is a financial crash? Why does it happen? Different theories on financial systems have tried to answer these fundamental questions. In this paper, I examine these theories philosophically, and I show that, to adequately answer these two questions, they would have to:

  1. characterize the ontological and epistemic states of the financial markets and their interrelation;

  2. identify the causes of a financial crash, the basic entities involved, and their relations.

2On this basis, I outline some ontological and epistemic preconditions that are necessary for a financial crash to occur; I also examine an example of price drop which, although not a proper flash crash, can help to better understand my analysis.

  • 1 See De Bruin et al. 2020, Ippoliti 2020a for a philosophical analysis of the notion of “intrinsic” (...)
  • 2 Algo-trading is the use of automated trading to execute orders when market conditions match the com (...)

3In first approximation, the standard answer to the ontological question “what is a financial crash?” is that it is a steep and quick decline in the value of asset prices. On the other hand, the standard answer to the explanatory question “why does a financial crash happen?” is that it comes about when institutions or assets are overvalued. The standard answer requires the controversial notion of intrinsic (“true”) value of an asset, which is openly questioned by financial theories like the Reflexive Market Hypothesis.1 The standard answer also says that a crisis can be exacerbated by herd-like behavior of investors, that is, by a set of specific beliefs and expectations about other’s belief on the future price trend. A typical example of such a behavior is the reaction to a rapid series of sell orders resulting in a lower assets price, which in turn pushes individuals to dump those assets. According to this view, common causes of financial crashes are factors like incentives to take too much risk, and unanticipated or uncontrollable human behavior. Also, algo-trading2 can cause financial crashes. In this case, fast sequences of large volume of automated selling orders can produce imbalances in offer and demand, pushing down the price of indexes or stocks very quickly (an example is the gold flash crash, on 6 January 2014).

4In the following sections, I argue that for financial crashes to happen, the two fundamental states of the market, the “ontic” and the “epistemic” state must meet certain preconditions. The “ontic” state is the specific and provisional setup of basic entities and their relations, such as liquidity, demand and offer. The “epistemic” state is what the agent believes about future states of the market. Each agent enters the market with a private valuation of an asset and different time horizons, and they both condition her order submission. The consequent variety of valuations can produce several outcomes and generate the prevalent “epistemic state” of the market. I argue that, to happen, a crash requires the ontic state to include some imbalance such as uneven levels of liquidity, or an imbalance in the relationship between supply and demand, while the epistemic state must rapidly convergence towards a single, monistic, belief.

5More specifically, in section 1, I examine the three main reasons why it is so difficult to understand and predict the performance of the financial systems and their crashes. These reasons are the role of the reflexive effect, data incompleteness and variability, and the interplay of quantitative and qualitative investigation.

6In section 2, I examine the philosophical setup of five main theories accounting for financial crashes: Efficient Market Hypothesis (EFH) (Fama 1970); Financial Instability Hypothesis (FIH) (Minsky 1980), Reflexive Markets Hypothesis (RMH) (Popper 1957; Soros 1987), Econophysics & Social bubbles (EPH) (Sornette 2003; Gisler , Sornette 2010), and Behavioral Finance (BFH) (Tversky, Kahneman 1986; Gigerenzer, Goldstein 1996; Gigerenzer, Todd 1999). I analyze how these theories employ different combinations of ontological, epistemological, and methodological hypotheses to characterize financial crashes.

7In section 3, I argue that even if all these theories differ in many respects, they all must admit basic preconditions for financial crashes to occur. I identify these preconditions, I present three different kinds of crashes, and then I consider the example of a rapid price drop (but not a proper crash that would require too much space) to show the interplay of epistemic and ontic states in starting the severe price drop required by a crash.

1.1. Understanding and predicting financial crashes

8There are at least three main reasons why it is so difficult to understand and predict financial systems and their crashes: the role of the reflexive effect; data incompleteness; and the interplay of quantitative and qualitative investigation.

a) The reflexive effect

9The first reason why it is so difficult to understand and predict financial crashes is the well-known reflexive (feedback) effect of prediction in social system. This effect states that predictions or expectations alter the future states of the social systems, and this is much more evident in financial systems, especially in stock markets. In this case, the agents continually try to foreshadow the future state of the system by means not only of public predictions but also of the investors’ private beliefs and predictions that are revealed by their sales and purchases. This can trigger a regression to infinity, which makes a prediction very difficult, if not logically impossible, as stated by the Morgenstern’s paradox (Morgenstern 1928: 93-100; 1935). Indeed, the prediction can alter the system’s dynamics in two ways. It can push it to diverge from the behavior that the system would have followed if the prediction had not been revealed (the prediction becomes self-refuting). But the opposite scenario is also possible: the prediction can become self-fulfilling. In this case, it becomes a fixed point towards which the system converges. For example, assume that a crash prediction is issued stating that a crash between 19% and 31% will occur in the next five weeks. Now, if most of the investor believes the warning, panic spreads and the market crashes as consequence. The prediction thus looks self-fulfilling, but its success is ascribed more to the panic effect than to its predictive power. If sufficiently many investors believe that the prediction may be correct, they adjust their investments, and the prices decline slowly. Here the prediction is self-refuting.

  • 3 Several solutions to this problem have been proposed. See, for example, Grunber, Modigliani 1954, 1 (...)

10Thus, the eventual convergence between a given prediction and the future state of the market is the crucial problem in predicting financial systems.3 Data variability further complicates the difficulty of predicting. Even if we admit that a prediction is possible, it is extremely complex since the salient data and variables for making predictions can change radically from crash to crash: a set of variables that are reliable for one crash may turn out to be useless for the next one (see Silver 2012).

b) Data incompleteness

11The second reason why it is so hard to understand and predict financial crashes is the fact that we cannot observe the financial phenomenon at a fundamental level, since we do not know who trades what. Bank accounts identifiers of the transactions are hidden, so we cannot track and reconstruct the financial dynamics and its machinery at the individual (atomic) level since data are incomplete. It should be noted here that there are theories arguing that the effect of this limitation could be mitigated – for example by employing an emergentist approach (see also Ippoliti 2020). According to this approach, it is not necessary to predict the behavior of a system at the microscopic level, that is, in terms of its individual constituents, but it is enough to predict some critical collective events like boom-and-bust (Sornette 2003). These critical events are the fundamental junctions that drive the long-term behavior of the system and are essential for understanding and predicting it. Thus, although a detailed, microscopic understanding of a complex system seems beyond our reach, this does not rule out the possibility of predicting the events that really matter like crashes or bubbles.

c) Qualitative and quantitative investigation

12The third reason why it is so hard to understand and predict financial crashes is that two kinds of investigations are needed: quantitative and qualitative. The first aims at accounting for the quantitative expression of the market’s dynamics by fitting data (prices) of financial systems with mathematical functions (and predicting the next data). The difficulty here is the well-known underdetermination of curve-fitting: in principle, there are infinite possible functions that can fit the data and choosing one of them is highly controversial. The second aims at identifying how the individual and collective behavior of the investors determines the quantitative expression of the system, for instance how the financial contracts are originated, what rules govern their trades, or how financial agents make decisions and interact with each other through these contracts. The difficulty here is that this description could not be detailed enough (or too narrow) to produce a general theory, leaving us only partial accounts of the local behavior of some markets or investors’ communities.

13These two investigations are not necessarily connected – they can be pursued independently, and one can make progress without the other. Indeed, different financial theories have diverse tenets on the relation between qualitative and quantitative investigations and their roles in understanding and predicting financial markets. We can identify at least three tenets:

  1. the quantitative expression of the financial market is enough to foreshadow a bubble or a crash (see, for example, Econophysics & Social bubbles hypothesis, EPH); a qualitative understanding can be dispensed with, although it remains possible.

  2. a better understanding of the quantitative expression of the financial market does not lead to a qualitative understanding of it or of its predictability. For example, the fractal theory (see, for example, Mandelbrot 2006; Peters 1994) provide us with a new statistical way to interpret the financial data (prices) that captures and formalizes their non-normal distribution and self-similar “pattern” (indifference to scale). Nevertheless, these mathematical models do not improve our understanding of qualitative dynamics, that is, the behavior and decisions of the investors. Moreover, fractal models do not offer a way of predicting the market due to their sensitivity to initial conditions. In fact, a rounding error of a stock price by just 1/10th of a percent would make the model completely useless as a means of predicting the prices.

  3. the bottom-up study of the qualitative nature of the financial market can offer a quantitative understanding of them and, thus, of their predictability (see for example Ippoliti 2020).

2. Theories on financial crashes: an outline

14In what follows, I consider five different theories that attempt to account for financial crashes. Each of them tries to:

  1. Structure the relation between “ontic” and “epistemic” states. The ontic state of the financial market is the specific and provisional setup of basic entities and their relations, whereas the epistemic state is what the agents believe about future states of the market;

  2. Differentiate between causes of financial crashes, such as “endogenous” vs “exogenous”, and “remote” vs “proximate”. The endogenous causes are those internal to the financial investors’ community, for instance imitating behavior. The exogenous ones are those coming from outside the financial system, typically news and reports on economic conditions. The remote causes are those acting over long time scale and on structural forces, for instance an increasing level of debts of many financial actors. The proximate ones are those arising shortly before the crash and which are accidental, typically a series of selling orders of very few investors triggering a price drop.

  3. Identify basic entities and units of analysis (for example, “individuals” vs “aggregates”).

  4. Define what we can know about those causes and entities, and what they can reveal about an impending financial crash.

2.1. The efficient market hypothesis: the ontic state primacy

15The EMH argues that the ontic state and the epistemic state are separate and in an asymmetric relation: the ontic state precedes and determines the epistemic state or, better, the two states line up since the epistemic state simply reflects the ontic state. To maintain this thesis, the EMH employs a reductionist ontology, and it states that (i) the basic entities and units of analysis are individuals (not aggregates) acting under a shortage of resources, and (ii) everything, including systemic behavior, can be explained by means of the properties of individuals. The independence and asymmetry of the ontic and epistemic states also imply the independence of supply and demand, that is, the fact that the former does not affect the latter.

16On this basis, the EMH argues that the causes of financial crashes are proximate in time and exogenous: as new, high-impact events external to the financial system occur (for example, wars or pandemics), information about them is released and the financial markets react and adjust their own evaluations. Thus, in this respect, a crash is still a correct evaluation by the financial system of the market, and not a failure of it.

17The EMH includes also an epistemological hypothesis stating that the current prices do not allow us to know the movements of future prices and crashes since they are caused by a random flux of events and information about those events. This implies the well-known methodological hypothesis that it is useless to study crashes quantitatively and we cannot predict them.

2.2 Reflexive markets hypothesis: the circularity of epistemic and ontic states

18Drawing on the theories provided by Merton (1948), Popper (1957), and Flanagan (1981), the RMH maintains that the ontic and epistemic states are in a circular relation that affects prices continually. RHM argues that the epistemic state can deeply alter the ontic state by means of the reflexive effect of investors’ expectations and beliefs. RHM shows that reflexivity alters not only market prices but also economic fundamentals such as profitability, revenue, assets, liabilities, and the growth potential of a company.

  • 4 RMH argues that supply and demand cannot be considered independent because they incorporate partici (...)

19The ontology of crashes offered by the RMH is built upon two theoretical pillars: i) the market is always biased, in one way or another; ii) the market can influence the events that it predicts. These two tenets imply that notions such as balance, rational investors, independence between supply and demand4 are substantially inadequate to account for the financial market’s behavior. Also, these tenets imply that a core principle of the EMH is unsustainable. According to this principle, the (upwards) trends that we observe in financial markets are only temporary deviations from equilibrium that will be eliminated by the fundamental forces of supply and demand, restoring the “natural” random walk of prices. This position ignores the reflexive effects and the fact that there is no guarantee that only the ontic state defines the epistemic state. In this case the market’s fundamental forces won’t correct the trends and it is probable that the epistemic state (and speculation) will alter supply and demand, pushing the market toward a tipping point where a crash will occur.

20Thus, the main epistemic hypothesis of the RMH is that since prices are wrong or biased most of the time (Soros 1987), they generate distort trends that lead to frequent big price corrections when they reach prices that are too far from real economic conditions. Therefore, unlike EMH, RMH characterizes the market basically as an epistemic and reflexive activity: the essence of the financial market is to figure out not simply which securities are good by themselves, but which securities are considered good by most of the other players (in analogy with the well-known Keynesian beauty contest). Thus, instead of simply elaborating investment prospects and global economic conditions in a rational way, financial investors consider each other’s epistemic states (for example, they try to figure out whether most of them appreciate certain companies, whether they are willing to sell or buy certain securities, when and at what price). Of course, the ontic state still plays a role in the formation of the epistemic state since the investors also take global economic conditions into account. Nevertheless, many investors analyze and try to front run the internal conduct of financial community, that is, what many other investors belief. Thus, they perform a purely epistemic exercise on other’s epistemic states. That would explain why crashes are so frequent in financial markets.

21Thus, the RMH puts forward a totally opposite epistemological hypothesis to the EMH: since real investors are in a state of incomplete understanding and information asymmetry, and their decisions cannot be rational in the neoclassical sense, current prices are intrinsically unstable and provide us with some information about upcoming crashes. So, the neoclassical idea that markets accurately predict future developments is misleading: as a matter of fact, in many cases the prediction does not describe future events, but future events are shaped by a current prediction (until big price corrections, like a crash or a bubble, occur).

22This implies the methodological hypothesis that we can look at price trends to detect when their endpoints, that is, crashes, are approaching. Unfortunately, this methodology is purely qualitative: the RMH does not provide us with quantitative tools to identify the trends’ tipping points.

2.3 Behavioral finance: the primacy of the epistemic state

23The BFH (Ackert, Deaves 2009; Ross 2005) and its two main approaches – Heuristics & Biases (Tversky, Kahneman1986), and Fast & Frugal (Gigerenzer, Goldstein 1996) – maintain that the epistemic state precedes and determines the ontic state. To do that, like the EMH, the BFH employs a reductionist ontology, and it states that the basic entities and units of analysis are individuals acting under a shortage of resources. On the other hand, however, the BFH employs a different epistemological apparatus than the EMH. In fact, according to the BFH, there is plenty of evidence that individuals follow lines of reasoning that are very different from the optimizing rationality attributed to financial agents by the EMH (for example, homo economicus). As a matter of fact, they are guided mostly by non-standard preferences, non-standard beliefs, and non-standard decision-making.

24In particular, the Heuristics & Biases theory argues that a crash, such as a market failure, can be produced endogenously because of the inefficiencies of financial agents in processing information. In this sense, completeness of information or data is not enough to prevent investors from making mistakes or to avoid the resulting big price corrections. As the H&B findings point out, even when human agents are provided with all the right information, they are not capable of using this information correctly to produce the right decision because emotional and cognitive distractors (for example, risk-aversion) generate inefficiencies and market failures. This implies a methodological hypothesis: to study crashes, we must understand the psychology of investors towards price trends and the heuristic reasoning they are following.

2.4 Financial instability hypothesis: the epistemic state destabilization of the ontic state

25Unlike the EMH, the FIH argues that the epistemic state determines the ontic state by progressively destabilizing it: the crash, the Minsky moment, is a cyclic, natural, outcome of an endogenous financial mechanism. This theory states that market crashes are not only exogenous, that is generated by a shock coming from outside the markets, as EMH states, but they can be endogenous, meaning produced internally by the financial community because of their beliefs.

26Minsky argues that financial markets are characterized by a peculiar phenomenology, a “circular dynamic”, whereby the seeds of the next crash are sown during the times of economic prosperity and stability. More specifically, in a period of economic stability, financial agents and institutions become complacent: they assume that the moment of prosperity will persist and, reassured by this epistemic state, they take greater risks in the search for greater profits. Thus, the epistemic state progressively takes over the ontic state triggering Minsky’s well-known four-staged cycle culminating in a big crash (1. cover financing ➛ 2. speculative financing ➛ 3. Ponzi or ultra-speculative financing ➛ 4. Minsk moment/crash).

27By loosening the criteria for granting credit, finance acts on the economic system as an internal destabilizing factor responsible for triggering the future crash. Using Minsky’s words, the finance is not simply pipes, but a pump: contrary to what is claimed by EMH, finance is not simply a neutral intermediary that moves money through the economic system from savers to borrowers, but it is a web of institutions aimed at profit with an incentive to increase loans and take the risks that incurs. From a certain moment on, the entire Minsky cycle is epistemically driven: the prevalent belief of the markets (what most of the financial agents believe) is that the moment of economic growth and prosperity will continue (and prices and the asset values will increase), making it possible for the vast majority to repay their debts in the mid-term. Then, this belief becomes a euphoria, which leads to a relaxation in the criteria for granting loans, and then to the final crash. A crash is thus the natural endpoint of financial mechanisms, if they are not counteracted.

28According to the FIH, the prices are altered by the epistemic state of the markets, from a certain point on they can signal an impeding crash; the FIH supports, however, the methodological hypotheses that a quantitative analysis of prices is not enough to predict and, more importantly, prevent crashes, as we need a qualitative understanding of the relations between the main financial actors in order to block the Minsky cycle.

2.5 Econophysics and social bubbles: the ontic state subtly foreshadows the epistemic state

  • 5 The analogy used to explain this is the unstable position of an object. For instance, a pencil held (...)

29An interesting version of EPH (Sornette 2003) maintains that the ontic state subtly foreshadows the epistemic state. The EPH employs an emergentist ontology: the basic entities and units of analysis are aggregates and their quantitative expressions (for example, prices), and the properties of the single investor can be neglected without losing the capacity to explain and predict critical events (crashes). According to this theory, most of the time the causes of financial crashes are endogenous, and exogenous shocks serve only as accidental triggers.5 Moreover the remote causes are the real ones, while the proximate causes are accidental. Therefore, a crash is the correction of a wrong evaluation of the market.

30This version of the EPH argues that the origin of a crash is subtle and profound: it is generated progressively by the whole market, and it is provoked by the slow build-up of long-range correlations leading to global cooperation in the market and then to a crash lasting a very short time. Therefore, a crash can be triggered by auto-reinforcing local imitation between traders: if the tendency of traders to imitate their peers increases up to a certain critical point, then many traders will place the same order at the same time. If this order is “sell”, a crash will take place.

31The epistemological hypothesis endorsed by the EPH is that current prices enable us to know impeding crashes: prices subtly release information about impending crashes, which agents are not yet aware of. This implies the methodological hypothesis that it is possible to study the prices quantitatively and foreshadow a crash, even if only in a probabilistic way.

32The EPH argues that a specific set of functions – the LPPL functions – are the key to predicting incoming crashes. Indeed, price dynamics tell us that a critical event c is approaching: the combination of an increasing price with an increase in the oscillations of that price foreshadows an incoming crash. A historical example that fits this account is the 1929 crash (Fig. 1).

Fig. 1. The 1929 crash approximated by a LPPL function (in red)

Fig. 1. The 1929 crash approximated by a LPPL function (in red)

33Thus, if data (prices) can be approximated by a LPPL function, then according to the EPH we are approaching a crash. On this basis, this theory maintains the thesis that mathematics (quantitative tools) is the key to predicting an incoming crash. It also implies a kind of dispensability argument about the qualitative investigation, since we do not need to understand the qualitative behavior of the system to foresee a critical event. Nonetheless, the theory offers a qualitative explanation of crashes (they are the outcome of increasing cooperation and imitation between investors, a “social bubble” – see Ginsler, Sornette 2010).

3. Some remarks on ontological and epistemic preconditions for financial crashes

  • 6 Boldyerv &Ushakov 2016; Brisset 2018; Callon 2007; Latour 2005; McKenzie 2006; Ippoliti 2019.
  • 7 See also Svetlova 2012 on this point.

34Even if these theories differ under many respects, they all must admit some basic preconditions for financial crashes to be triggered. Now, a condition for the existence of a market is that agents attribute varying prices to an asset (see, for example, Merton, Bodie 1995; Chami et al. 2010). This is also an ontological precondition for equilibrium, as required by the EMH. Each agent enters the market with a private valuation of an asset, which influences her order submission. This variety of valuations can produce several outcomes (equilibrium is just one) and determines the overall epistemic state of the market. When this difference in valuations is flattened, the market is dominated by a prevailing belief, and the epistemic state of the market affects its future dynamics by pushing it to converge rapidly towards the same, single order submission in the same time interval. In this case, critical events become more likely. This means that a monistic epistemic state of the market is a precondition for a crash. Different hypotheses have been proposed to explain how this state is determined. According to one of them, models can act as coordinating epistemic tools that flatten the epistemic state of the market – in this sense performativity6 can be seen as a way of generating crashes7 and predicting markets.

35Another ontological precondition for crashes is the presence of potential or actual instability in the setup of basic entities, such as unbalanced levels of debt or liquidity, and imbalance in supply and demand. In fact, once a specific series of orders is submitted, this instability emerges and can worsen for merely epistemic reasons, since there are no enough ontic market forces, such as orders of the opposite sign, counter-acting the epistemic state expressed by a strongly prevailing belief pushing towards the same, concomitant order submissions.

36This does not mean that crashes are all the same kind. Not only do they have different triggering factors and effects, but they also have different dynamics – such as, what precedes or follows a crash. It is possible to distinguish at least three different kinds of crashes based on their dynamics:

  1. “Big crash” (for example, the well-known 1929 crash – or “Black Tuesday”). In this case, a severe drop in prices occurs, pushing the level of price very far away from that prior to the crash. It takes a (very) long time for the prices to return to their level prior to the crash.

  2. “Boom and bust” (for example, the well-known 1998 bubble). In this case, a severe drop ends a boom (a rapid and significant increase in prices), pushing the prices back to about the same level as prior to the boom.

  3. “Drop and recovery” (for example, the 2010 flash crash). In this case, a severe drop is followed by a quick recovery that pushes the prices back to about their level prior to the drop.

Fig. 2. Big crash (for example, 1929 Black Tuesday)

Fig. 2. Big crash (for example, 1929 Black Tuesday)

Fig. 3. Boom and bust (for example, 1998 bubble)

Fig. 3. Boom and bust (for example, 1998 bubble)

Fig. 4. Drop and recovery (for example, 2010 flash crash)

Fig. 4. Drop and recovery (for example, 2010 flash crash)

3.1. How a crash starts and the role of the epistemic and ontic states: a (micro) example

37In order to illustrate the beginning of a rapid price drop, and the role played by the epistemic and ontic states of the market, I will examine a micro example (see, for example O’Hara 2010). It is a very simplified example involving two algorithmic agents and, even if it is not strictly speaking a crash, it illustrates the very first steps that may lead to a crash without requiring to take into consideration many factors like other well-known examples of crashes do.

38In this micro example, we start from an initial setting where the best ask (the selling price or demand) for a share m is 10.00 € and the best bid (the buying price or offer) is 9.85 €. Thus, the ask/bid spread is 15 cents. At this point a financial algo, Alien, enters the market. Alien wants to buy m at the best possible price (it is now 10 €). To trigger a process to get a better price, it places a selling order (ask) for 9.99 and, right after, cancels it: on the order book, an ask order of 9.99 appears (Tab. 1) and disappears in terms of milliseconds (t1-t3). Of course, a fundamental issue here is why Alien cancels the order immediately after its submission. The goal is to stop the order from being executed, meaning to prevent someone from buying it, since at that point, Alien would have to sell something that it wants to buy. It is worth noting that no one is openly violating laws here – it is not forbidden to place and cancel orders, which can be done for legitimate reasons. Nonetheless, Alien’s peculiar mode of operation differentiates this example from other classical crashes.

39Now, Alien is so quick that no human eye can see its orders. But they can be detected by another algo, say Predator.

Tab 1: The submission and immediate removal of an ask order by the algo agent Alien


order book

Ask (selling price/Demand)

Bid (buying price/offer)





(Alien) 9,99



40Predator wants to sell m, therefore it reacts to this ask order by placing a new ask at 9.98 to provide a better ask. This order is real, meaning that it is not cancelled. Alien, in turn, submits a new ask order at 9.97 and, again, cancels it right after. Anyway, Predator replies by lowering the order at 9.96, then Alien submits a new ask at 9.95 and cancels it right after. And so on, for a while (Tab. 2). Of course, another algo, say T-Rex, could enter in the process and lowering the price with other selling orders. This would make the price drop more rapid and drastic.

Tab. 2. The rapid series of ask orders from Alien and Predator

Tab. 2. The rapid series of ask orders from Alien and Predator

41This series of order submissions ends when Alien enters a bid order of 9.88, that is, it buys m at a much lower price, as desired. Therefore, in a few seconds, the price for m drops from 10.00 to 9.88 – that is 1.2% (Fig. 5).

Fig. 5. The rapid price drop of the share m in consequence of Alien and Predator order submissions (but not execution)

Fig. 5. The rapid price drop of the share m in consequence of Alien and Predator order submissions (but not execution)

42A salient feature of this example is that the price drops even if no one is selling or buying shares until the very last bid order submitted by Alien. There are no transactions occurring here. This means that the price for m is not given by the balance between the real supply and demand, but by purely epistemic factors – namely, what the two agents believe about each other’s intentions, and how they interpret intentions of others.

  • 8 If the number of buyers is more than the number of sellers, the supply will be low(er). Conversely, (...)

43The most interesting philosophical aspect of this example is the interplay of the ontic and epistemic states. To work, Alien’s strategy requires a very specific ontic state: scarce liquidity (see O’Hara 2010), that is, an ontic state of the market whereby buying and selling orders are present in adequate quantities8 and the activity is minimal (very few agents active). The orders must be submitted and cancelled during this specific time frame and in those market conditions: in this case, in the presence of buying and selling orders, they can create a potential instability and imbalance (for example, too many selling orders, which cannot be absorb by buying orders). This scenario, especially in the high-speed trading landscape, produces a rapid convergence towards the same epistemic state (ask orders submissions). Indeed, if there were many agents operating during the submission of orders, the probability of anyone executing Alien’s order would be higher, and Alien would have to sell m: its strategy would thus become counterproductive. Also, the fact that algo trading takes place at light speed reduces the reaction times drastically, making it more difficult to form private valuations of the asset that contrast the monistic view.

44This example shows how a price can go down rapidly for epistemic reasons alone: it drops in a purely endogenous way as there is no news or event occurring (not even any transactions). It begins with a minimal ontic state, that is, scarce liquidity, and then, because of the minimal reaction times imposed by algo trading, it converges rapidly towards an epistemic monistic view (the same orders submission).

Torna su


Ackert, L., Deaves, R. 2009, Behavioral Finance: Psychology, Decision-Making, and Markets, Mason, Cengage Learning.

Boldyerv, I., Ushakov, A. 2016, Adjusting the model to adjust the world: constructive mechanisms in postwar general equilibrium theory, “Journal of Economic Methodology”, 23, 1: 38-56.

Brisset, N. 2018, Models as speech acts: the telling case of financial models, “Journal of Economic Methodology”, 25, 3: 1-21.

Callon, M. 2007, What does it mean to say that economics is performative?, in D. MacKenzie, F. Muniesa, L. Siu (eds), Do Economists Make Markets? On the Performativity of Economics, Princeton, Princeton University Press: 310-357.

Chami, R., Fullenkamp, C., Sharma, S. 2010, A framework for financial market development, “Journal of Economic Policy Reform”, 13, 2: 107-135.

De Bruin, B., Herzog, L., O’Neill, M., Sandberg, J. 2020, Philosophy of money and finance, in E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2020 Edition),

Devletoglou, E.A. 1961, Correct public prediction and the stability of equilibrium, “Journal of Political Economy”, 69: 142-146.

Fama, E.F. 1970, Efficient capital markets: a review of theory and empirical work, “Journal of Finance”, 25, 2: 383-417.

Flanagan, O.J. 1981, Psychology, progress, and the problem of reflexivity: a study in the epistemological foundations of psychology, “Journal of the History of the Behavioral Sciences”, 17: 375-386.

Friedman, T.L. 1999, The Lexus and the Olive Tree: Understanding Globalization, New York, Farrar Straus & Giroux.

Galatin, M. 1976, Optimal forecasting in models with uncertainty when the outcome is influenced by the forecast, “The Economic Journal”, 86: 278-295.

Gigerenzer, G., Goldstein, D.G. 1996, Reasoning the fast and frugal way: models of bounded rationality, “Psychol. Rev.”, 104: 650-669.

Gigerenzer, G., Todd, P.M. 1999, Fast and frugal heuristics: the adaptive toolbox, in G. Gigerenzer, P.M. Todd, The ABC Research Group (eds), Simple Heuristics that Make us Smart, Oxford, Oxford University Press: 3-34.

Gisler, M., Sornette, D. 2010, Bubbles everywhere in human affairs, in L.K. Bogataj, K.H. Mueller, I. Svetlik, N. Tos (eds), Modern RISC-​Societies. Towards a New Framework for Societal Evolution, Wien, Edition echoraum: 137-​153.

Grunberg, E.J., Modigliani, F. 1954, The predictability of social events, “Journal of Political Economy”, 62: 465-478.

Grunberg, E.J., Modigliani, F. 1963, Economic forecasting when the subject of the forecast is influenced by the forecast: comment, “American Economic Review”, 53: 734-737.

Henshel, R.L. 1978, Self-altering predictions, in J. Fowles (ed.), Handbook of Futures Research, Westport, Greenwood Press: 99-123.

Ippoliti, E. 2019, Models and data in finance: les liaisons dangereuses, in A. Nepomuceno Fernández, L. Magnani, F. J. Salguero-Lamillar, C. Barés-Gómez, M. Fontaine (eds), Model-Based Reasoning in Science and Technology, Berlin, Springer: 393-406.

Ippoliti, E. 2020, Mathematics and finance: some philosophical remarks, “Topoi”,

Ippoliti, E. 2020a, Un filosofo a Wall Street, Milano, Egea.

Kemp, M.C. 1962, Economic forecasting: when the subject of the forecast is influenced by the forecast, “American Economic Review”, 52: 492-496.

Kemp, M.C. 1963, Economic forecasting when the subject of the forecast is influenced by the forecast: reply, “American Economic Review”, 53: 738-740.

Keynes, J.M. 1936, The General Theory of Employment, Interest and Money, New York, Harcourt Brace and Co.

Latour, B. 2005, Reassembling the Social: An Introduction to Actor-Network-Theory, Oxford, Oxford University Press.

Lehmann-Waffenschmidt, M. 1990, Predictability of economic processes and the morgenstern paradox, “Swiss Journal of Economics and Statistics”, 126, 2: 147-161.

MacKenzie, D. 2006, An Engine, not a Camera. How Financial Models Shape Markets, Cambridge, The MIT Press.

Mandelbrot, B. 2006, The Misbehavior of Markets: A Fractal View of Financial Turbulence, New York: Basic Book.

Merton, R.K. 1948, The self-fulfilling prophecy, “The Antioch Review”, 8, 2: 193-210.

Merton, R.K., Bodie, Z. 1995, A conceptual framework for analyzing the financial environment, in D.B. Crane, K. A. Froot, S.P. Mason et al. (eds), The Global Financial System: A Functional Perspective, Boston, Harvard Business School Press: 3-31

Minsky, H. 1980, Capitalist financial processes and the instability of capitalism, “Journal of Economic Issues”, 14, 2: 505-523.

Morgenstern, O. 1928, Wirtschaftsprognose. Eine Untersuchung ihrer Voraussetzungen und Möglichkeiten, Vienna, Springer.

Morgenstern, O. 1935, Perfect foresight and economic equilibrium, in A. Schotter (ed.), Selected Economic Writings of Oskar Morgenstern, New York, New York University Press: 169-183.

O’Hara, M. 2010, What is a quote?, “The Journal of Trading”, 5, 2: 10-16.

Peters, E.E. 1994, Fractal Market Analysis, New York, Wiley.

Popper, K. 1957, The Poverty of Historicism, London, ARK Paperbacks.

Ross, D. 2005, Economic Theory and Cognitive Science: Microexplanation, Boston, The MIT Press.

Searle, J. 1969, Speech Acts: An Essay in the Philosophy of Language, Cambridge, Cambridge University Press.

Sornette, D. 2003, Why Stock Markets Crash, Princeton, Princeton University Press.

Soros, G. 1987, The Alchemy of Finance, New York, Wiley & Sons.

Svetlova, E. 2012, On the performative power of financial models, “Economy and Society”, 43, 2: 418-434.

Tversky, A., Kahneman, D. 1986, Rational choice and the framing of decisions, “The Journal of Business”, 59, 4: S251-S278.

Torna su


1 See De Bruin et al. 2020, Ippoliti 2020a for a philosophical analysis of the notion of “intrinsic” or “true” value.

2 Algo-trading is the use of automated trading to execute orders when market conditions match the computerized instructions.

3 Several solutions to this problem have been proposed. See, for example, Grunber, Modigliani 1954, 1963; Devletoglou 1961; Galatin 1976; Henshel 1978; Kemp 1962, 1963; Lehmann, Waffenschmidt 1990.

4 RMH argues that supply and demand cannot be considered independent because they incorporate participants’ expectations about events, and therefore they are shaped by those expectations. Especially in the financial market, buying and selling decisions are based on future price expectations, and future prices, in turn, are influenced by current buying and selling decisions. Thus, RMH concludes it is misleading to consider supply and demand as if they were determined by forces independent of the expectations of market participants.

5 The analogy used to explain this is the unstable position of an object. For instance, a pencil held vertically on the palm of your hand is in a very unstable position, which will cause the pencil to fall sooner or later: it is the position itself that creates the (endogenous) conditions for the fall, and not actions like a small accidental movement of your hand.

6 Boldyerv &Ushakov 2016; Brisset 2018; Callon 2007; Latour 2005; McKenzie 2006; Ippoliti 2019.

7 See also Svetlova 2012 on this point.

8 If the number of buyers is more than the number of sellers, the supply will be low(er). Conversely, if there are more sellers than buyers, the demand will be low(er). These two market scenarios result in low(er) liquidity.

Torna su

Indice delle illustrazioni

Titolo Fig. 1. The 1929 crash approximated by a LPPL function (in red)
File image/png, 163k
Titolo Fig. 2. Big crash (for example, 1929 Black Tuesday)
File image/png, 95k
Titolo Fig. 3. Boom and bust (for example, 1998 bubble)
File image/png, 112k
Titolo Fig. 4. Drop and recovery (for example, 2010 flash crash)
File image/png, 12k
Titolo Tab. 2. The rapid series of ask orders from Alien and Predator
File image/png, 113k
Titolo Fig. 5. The rapid price drop of the share m in consequence of Alien and Predator order submissions (but not execution)
File image/png, 26k
Torna su

Per citare questo articolo

Notizia bibliografica

Emiliano Ippoliti, «What is a financial crash?»Rivista di estetica, 84 | 2023, 7-24.

Notizia bibliografica digitale

Emiliano Ippoliti, «What is a financial crash?»Rivista di estetica [Online], 84 | 2023, online dal 01 février 2024, consultato il 15 juin 2024. URL:; DOI:

Torna su

Diritti d’autore


Solamente il testo è utilizzabile con licenza CC BY-NC-ND 4.0. Salvo diversa indicazione, per tutti agli altri elementi (illustrazioni, allegati importati) la copia non è autorizzata ("Tutti i diritti riservati").

Torna su
Cerca su OpenEdition Search

Sarai reindirizzato su OpenEdition Search