arXiv subjects

Didier Sornette

Publications and source records attributed to Didier Sornette.

At least 109 records · Page 6Linked to original sources

arXiv2017

Lagrange regularisation approach to compare nested data sets and determine objectively financial bubbles' inceptions

Inspired by the question of identifying the start time $τ$ of financial bubbles, we address the calibration of time series in which the inception of the latest regime of interest is unknown. By taking into account the tendency of a given model to overfit data, we introduce the Lagrange regularisation of the normalised sum of the squared residuals, $χ^{2}_{np}(Φ)$, to endogenously detect the optimal fitting window size := $w^* \in [τ:\bar{t}_2]$ that should be used for calibration purposes for a fixed pseudo present time $\bar{t}_2$. The performance of the Lagrange regularisation of $χ^{2}_{np}(Φ)$ defined as $χ^{2}_{λ(Φ)}$ is exemplified on a simple Linear Regression problem with a change point and compared against the Residual Sum of Squares (RSS) := $χ^{2}(Φ)$ and RSS/(N-p):= $χ^{2}_{np}(Φ)$, where $N$ is the sample size and p is the number of degrees of freedom. Applied to synthetic models of financial bubbles with a well-defined transition regime and to a number of financial time series (US S\&P500, Brazil IBovespa and China SSEC Indices), the Lagrange regularisation of $χ^{2}_λ(Φ)$ is found to provide well-defined reasonable determinations of the starting times for major bubbles such as the bubbles ending with the 1987 Black-Monday, the 2008 Sub-prime crisis and minor speculative bubbles on other Indexes, without any further exogenous information. It thus allows one to endogenise the determination of the beginning time of bubbles, a problem that had not received previously a systematic objective solution.

q-fin.ST↗

arXiv2017

Learning like humans with Deep Symbolic Networks

We introduce the Deep Symbolic Network (DSN) model, which aims at becoming the white-box version of Deep Neural Networks (DNN). The DSN model provides a simple, universal yet powerful structure, similar to DNN, to represent any knowledge of the world, which is transparent to humans. The conjecture behind the DSN model is that any type of real world objects sharing enough common features are mapped into human brains as a symbol. Those symbols are connected by links, representing the composition, correlation, causality, or other relationships between them, forming a deep, hierarchical symbolic network structure. Powered by such a structure, the DSN model is expected to learn like humans, because of its unique characteristics. First, it is universal, using the same structure to store any knowledge. Second, it can learn symbols from the world and construct the deep symbolic networks automatically, by utilizing the fact that real world objects have been naturally separated by singularities. Third, it is symbolic, with the capacity of performing causal deduction and generalization. Fourth, the symbols and the links between them are transparent to us, and thus we will know what it has learned or not - which is the key for the security of an AI system. Fifth, its transparency enables it to learn with relatively small data. Sixth, its knowledge can be accumulated. Last but not least, it is more friendly to unsupervised learning than DNN. We present the details of the model, the algorithm powering its automatic learning ability, and describe its usefulness in different use cases. The purpose of this paper is to generate broad interest to develop it within an open source project centered on the Deep Symbolic Network (DSN) model towards the development of general AI.

cs.AI↗

arXiv2017

Objective Estimation of Spatially Variable Parameters of Epidemic Type Aftershock Sequence Model: Application to California

The ETAS model is widely employed to model the spatio-temporal distribution of earthquakes, generally using spatially invariant parameters. We propose an efficient method for the estimation of spatially varying parameters, using the Expectation-Maximization (EM) algorithm and spatial Voronoi tessellation ensembles. We use the Bayesian Information Criterion (BIC) to rank inverted models given their likelihood and complexity and select the best models to finally compute an ensemble model at any location. Using a synthetic catalog, we also check that the proposed method correctly inverts the known parameters. We apply the proposed method to earthquakes included in the ANSS catalog that occurred within the time period 1981-2015 in a spatial polygon around California. The results indicate a significant spatial variation of the ETAS parameters. We find that the efficiency of earthquakes to trigger future ones (quantified by the branching ratio) positively correlates with surface heat flow. In contrast, the rate of earthquakes triggered by far-field tectonic loading or background seismicity rate shows no such correlation, suggesting the relevance of triggering possibly through fluid-induced activation. Furthermore, the branching ratio and background seismicity rate are found to be uncorrelated with hypocentral depths, indicating that the seismic coupling remains invariant of hypocentral depths in the study region. Additionally, triggering seems to be mostly dominated by small earthquakes. Consequently, the static stress change studies should not only focus on the Coulomb stress changes caused by specific moderate to large earthquakes but also account for the secondary static stress changes caused by smaller earthquakes.

physics.geo-ph↗

arXiv2017

Decision trees unearth return sign correlation in the S&P 500

Technical trading rules and linear regressive models are often used by practitioners to find trends in financial data. However, these models are unsuited to find non-linearly separable patterns. We propose a decision tree forecasting model that has the flexibility to capture arbitrary patterns. To illustrate, we construct a binary Markov process with a deterministic component that cannot be predicted with an autoregressive process. A simulation study confirms the robustness of the trees and limitation of the autoregressive model. Finally, adjusting for multiple testing, we show that some tree based strategies achieve trading performance significant at the 99% confidence level on the S&P 500 over the past 20 years. The best strategy breaks even with the buy-and-hold strategy at 21 bps in transaction costs per round trip. A four-factor regression analysis shows significant intercept and correlation with the market. The return anomalies are strongest during the bursts of the dotcom bubble, financial crisis, and European debt crisis. The correlation of the return signs during these periods confirms the theoretical model.

stat.AP↗

arXiv2016

Theory of earthquakes interevent times applied to financial markets

We analyze the probability density function (PDF) of waiting times between financial loss exceedances. The empirical PDFs are fitted with the self-excited Hawkes conditional Poisson process with a long power law memory kernel. The Hawkes process is the simplest extension of the Poisson process that takes into account how past events influence the occurrence of future events. By analyzing the empirical data for 15 different financial assets, we show that the formalism of the Hawkes process used for earthquakes can successfully model the PDF of interevent times between successive market losses.

q-fin.ST↗

arXiv2016

Discrete hierarchy of sizes and performances in the exchange-traded fund universe

Using detailed statistical analyses of the size distribution of a universe of equity exchange-traded funds (ETFs), we discover a discrete hierarchy of sizes, which imprints a log-periodic structure on the probability distribution of ETF sizes that dominates the details of the asymptotic tail. This allows us to propose a classification of the studied universe of ETFs into seven size layers approximately organized according to a multiplicative ratio of 3.5 in their total market capitalization. Introducing a similarity metric generalising the Herfindhal index, we find that the largest ETFs exhibit a significantly stronger intra-layer and inter-layer similarity compared with the smaller ETFs. Comparing the performance across the seven discerned ETF size layers, we find an inverse size effect, namely large ETFs perform significantly better than the small ones both in 2014 and 2015.

q-fin.GN↗

arXiv2016

The gradual evolution of buyer--seller networks and their role in aggregate fluctuations

Buyer--seller relationships among firms can be regarded as a longitudinal network in which the connectivity pattern evolves as each firm receives productivity shocks. Based on a data set describing the evolution of buyer--seller links among 55,608 firms over a decade and structural equation modeling, we find some evidence that interfirm networks evolve reflecting a firm's local decisions to mitigate adverse effects from neighbor firms through interfirm linkage, while enjoying positive effects from them. As a result, link renewal tends to have a positive impact on the growth rates of firms. We also investigate the role of networks in aggregate fluctuations.

q-fin.GN↗

arXiv2016

Forms of social relationships in distinct cultural settings

We contribute to the understanding of social relationships within cultural contexts by proposing a connection between a social theory, relational models theory (RMT: Fiske 1991, 1992) and a social and political one, cultural or plural rationality theory (PRT: Douglas, 1982, Thompson et al., 1990). Drawing examples from the literature of both theories, we argue that each relational model of RMT may be implemented in ways compatible with each cultural bias of PRT. A cultural bias restrains the range of congruent implementations of relational models, but does not preclude any relational model altogether. This stands in contrast to earlier reconciliation attempts between PRT and RMT. Based on hypothetical one-to-one mappings, these attempts expect each cultural setting to be significantly associated with some, but not all, relational models. The framework we develop helps explain the findings of these previous attempts and provides insights into empirical research by clarifying which associations to expect between relationships and cultural contexts. We discuss the theoretical basis of our framework, including the idea that RMT and PRT apply to different levels of analysis: RMT's relational models are tied to relationships between two actors and PRT's cultural biases to structures of social networks.

physics.soc-ph↗

arXiv2016

Systematic Assessment of the Static Stress-Triggering Hypothesis using Inter-earthquake Time Statistics

A likely source of earthquake clustering is static stress transfer between individual events. Previous attempts to quantify the role of static stress for earthquake triggering generally considered only the stress changes caused by large events, and often discarded data uncertainties. We conducted a robust two-fold empirical test of the static stress change hypothesis by accounting for all events of magnitude M>=2.5 and their location and focal mechanism uncertainties provided by catalogs for Southern California between 1981 and 2010, first after resolving the focal plane ambiguity and second after randomly choosing one of the two nodal planes. For both cases, we find compelling evidence supporting the static triggering with stronger evidence after resolving the focal plane ambiguity above significantly small (about 10 Pa) but consistently observed stress thresholds. The evidence for the static triggering hypothesis is robust with respect to the choice of the friction coefficient, Skempton's coefficient and magnitude threshold. Weak correlations between the Coulomb Index (fraction of earthquakes that received positive Coulomb stress change) and the coefficient of friction indicate that the role of normal stress in triggering is rather limited. Last but not the least, we determined that the characteristic time for the loss of the stress change memory of a single event is nearly independent of the amplitude of the Coulomb stress change and varies between ~95 and ~180 days implying that forecasts based on static stress changes will have poor predictive skills beyond times that are larger than a few hundred days on average.

physics.geo-ph↗

arXiv2016

Modified Profile Likelihood Inference and Interval Forecast of the Burst of Financial Bubbles

We present a detailed methodological study of the application of the modified profile likelihood method for the calibration of nonlinear financial models characterised by a large number of parameters. We apply the general approach to the Log-Periodic Power Law Singularity (LPPLS) model of financial bubbles. This model is particularly relevant because one of its parameters, the critical time $t_c$ signalling the burst of the bubble, is arguably the target of choice for dynamical risk management. However, previous calibrations of the LPPLS model have shown that the estimation of $t_c$ is in general quite unstable. Here, we provide a rigorous likelihood inference approach to determine $t_c$, which takes into account the impact of the other nonlinear (so-called "nuisance") parameters for the correct adjustment of the uncertainty on $t_c$. This provides a rigorous interval estimation for the critical time, rather than a point estimation in previous approaches. As a bonus, the interval estimations can also be obtained for the nuisance parameters ($m,ω$, damping), which can be used to improve filtering of the calibration results. We show that the use of the modified profile likelihood method dramatically reduces the number of local extrema by constructing much simpler smoother log-likelihood landscapes. The remaining distinct solutions can be interpreted as genuine scenarios that unfold as the time of the analysis flows, which can be compared directly via their likelihood ratio. Finally, we develop a multi-scale profile likelihood analysis to visualize the structure of the financial data at different scales (typically from 100 to 750 days). We test the methodology successfully on synthetic price time series and on three well-known historical financial bubbles.

q-fin.ST↗

arXiv2016

The Extreme Risk of Personal Data Breaches & The Erosion of Privacy

Personal data breaches from organisations, enabling mass identity fraud, constitute an \emph{extreme risk}. This risk worsens daily as an ever-growing amount of personal data are stored by organisations and on-line, and the attack surface surrounding this data becomes larger and harder to secure. Further, breached information is distributed and accumulates in the hands of cyber criminals, thus driving a cumulative erosion of privacy. Statistical modeling of breach data from 2000 through 2015 provides insights into this risk: A current maximum breach size of about 200 million is detected, and is expected to grow by fifty percent over the next five years. The breach sizes are found to be well modeled by an \emph{extremely heavy tailed} truncated Pareto distribution, with tail exponent parameter decreasing linearly from 0.57 in 2007 to 0.37 in 2015. With this current model, given a breach contains above fifty thousand items, there is a ten percent probability of exceeding ten million. A size effect is unearthed where both the frequency and severity of breaches scale with organisation size like $s^{0.6}$. Projections indicate that the total amount of breached information is expected to double from two to four billion items within the next five years, eclipsing the population of users of the Internet. This massive and uncontrolled dissemination of personal identities raises fundamental concerns about privacy.

stat.AP↗

arXiv2016

Micro-foundation using percolation theory of the finite-time singular behavior of the crash hazard rate in a class of rational expectation bubbles

We present a plausible micro-founded model for the previously postulated power law finite time singular form of the crash hazard rate in the Johansen-Ledoit-Sornette model of rational expectation bubbles. The model is based on a percolation picture of the network of traders and the concept that clusters of connected traders share the same opinion. The key ingredient is the notion that a shift of position from buyer to seller of a sufficiently large group of traders can trigger a crash. This provides a formula to estimate the crash hazard rate by summation over percolation clusters above a minimum size of a power sa (with a > 1) of the cluster sizes s, similarly to a generalized percolation susceptibility. The power sa of cluster sizes emerges from the super-linear dependence of group activity as a function of group size, previously documented in the literature. The crash hazard rate exhibits explosive finite-time singular behaviors when the control parameter (fraction of occupied sites, or density of traders in the network) approaches the percolation threshold pc. Realistic dynamics are generated by modelling the density of traders on the percolation network by an Ornstein-Uhlenbeck process, whose memory controls the spontaneous excursion of the control parameter close to the critical region of bubble formation. Our numerical simulations recover the main stylized properties of the JLS model with intermittent explosive super-exponential bubbles interrupted by crashes.

q-fin.TR↗

arXiv2015

Secular bipolar growth rate of the real US GDP per capita: implications for understanding past and future economic growth

We present a quantitative characterisation of the fluctuations of the annualized growth rate of the real US GDP per capita growth at many scales, using a wavelet transform analysis of two data sets, quarterly data from 1947 to 2015 and annual data from 1800 to 2010. Our main finding is that the distribution of GDP growth rates can be well approximated by a bimodal function associated to a series of switches between regimes of strong growth rate $ρ_\text{high}$ and regimes of low growth rate $ρ_\text{low}$. The succession of such two regimes compounds to produce a remarkably stable long term average real annualized growth rate of 1.6\% from 1800 to 2010 and $\approx 2.0\%$ since 1950, which is the result of a subtle compensation between the high and low growth regimes that alternate continuously. Thus, the overall growth dynamics of the US economy is punctuated, with phases of strong growth that are intrinsically unsustainable, followed by corrections or consolidation until the next boom starts. We interpret these findings within the theory of "social bubbles" and argue as a consequence that estimations of the cost of the 2008 crisis may be misleading. We also interpret the absence of strong recovery since 2008 as a protracted low growth regime $ρ_\text{low}$ associated with the exceptional nature of the preceding large growth regime.

q-fin.GN↗

arXiv2015

Macroeconomic Dynamics of Assets, Leverage and Trust

A macroeconomic model based on the economic variables (i) assets, (ii) leverage (defined as debt over asset) and (iii) trust (defined as the maximum sustainable leverage) is proposed to investigate the role of credit in the dynamics of economic growth, and how credit may be associated with both economic performance and confidence. Our first notable finding is the mechanism of reward/penalty associated with patience, as quantified by the return on assets. In regular economies where the EBITA/Assets ratio is larger than the cost of debt, starting with a trust higher than leverage results in the highest long-term return on assets (which can be seen as a proxy for economic growth). Our second main finding concerns a recommendation for the reaction of a central bank to an external shock that affects negatively the economic growth. We find that late policy intervention in the model economy results in the highest long-term return on assets and largest asset value. But this comes at the cost of suffering longer from the crisis until the intervention occurs. The phenomenon can be ascribed to the fact that postponing intervention allows trust to increase first, and it is most effective to intervene when trust is high. These results derive from two fundamental assumptions underlying our model: (a) trust tends to increase when it is above leverage; (b) economic agents learn optimally to adjust debt for a given level of trust and amount of assets. Using a Markov Switching Model for the EBITA/Assets ratio, we have successfully calibrated our model to the empirical data of the return on equity of the EURO STOXX 50 for the time period 2000-2013. We find that dynamics of leverage and trust can be highly non-monotonous with curved trajectories, as a result of the nonlinear coupling between the variables.

econ.GN↗

arXiv2015

Two-state Markov-chain Poisson nature of individual cellphone call statistics

Humans are heterogenous and the behaviors of individuals could be different from that at the population level. We conduct an in-depth study of the temporal patterns of cellphone conversation activities of 73'339 anonymous cellphone users with the same truncated Weibull distribution of inter-call durations. We find that the individual call events exhibit a pattern of bursts, in which high activity periods are alternated with low activity periods. Surprisingly, the number of events in high activity periods are found to conform to a power-law distribution at the population level, but follow an exponential distribution at the individual level, which is a hallmark of absence of memory in individual call activities. Such exponential distribution is also observed for the number of events in low activity periods. Together with the exponential distributions of inter-call durations within bursts and of the intervals between consecutive bursts, we demonstrate that the individual call activities are driven by two independent Poisson processes, which can be combined within a minimal model in terms of a two-state first-order Markov chain giving very good agreement with the empirical distributions using the parameters estimated from real data for about half of the individuals in our sample. By measuring directly the distributions of call rates across the population, which exhibit power-law tails, we explain the difference with previous population level studies, purporting the existence of power-law distributions, via the "Superposition of Distributions" mechanism: The superposition of many exponential distributions of activities with a power-law distribution of their characteristic scales leads to a power-law distribution of the activities at the population level.

physics.soc-ph↗

arXiv2015

"Speculative Influence Network" during financial bubbles: application to Chinese Stock Markets

We introduce the Speculative Influence Network (SIN) to decipher the causal relationships between sectors (and/or firms) during financial bubbles. The SIN is constructed in two steps. First, we develop a Hidden Markov Model (HMM) of regime-switching between a normal market phase represented by a geometric Brownian motion (GBM) and a bubble regime represented by the stochastic super-exponential Sornette-Andersen (2002) bubble model. The calibration of the HMM provides the probability at each time for a given security to be in the bubble regime. Conditional on two assets being qualified in the bubble regime, we then use the transfer entropy to quantify the influence of the returns of one asset $i$ onto another asset $j$, from which we introduce the adjacency matrix of the SIN among securities. We apply our technology to the Chinese stock market during the period 2005-2008, during which a normal phase was followed by a spectacular bubble ending in a massive correction. We introduce the Net Speculative Influence Intensity (NSII) variable as the difference between the transfer entropies from $i$ to $j$ and from $j$ to $i$, which is used in a series of rank ordered regressions to predict the maximum loss (\%{MaxLoss}) endured during the crash. The sectors that influenced other sectors the most are found to have the largest losses. There is a clear prediction skill obtained by using the transfer entropy involving industrial sectors to explain the \%{MaxLoss} of financial institutions but not vice versa. We also show that the bubble state variable calibrated on the Chinese market data corresponds well to the regimes when the market exhibits a strong price acceleration followed by clear change of price regimes. Our results suggest that SIN may contribute significant skill to the development of general linkage-based systemic risks measures and early warning metrics.

q-fin.ST↗

arXiv2015

Currency target zone modeling: An interplay between physics and economics

We study the performance of the euro/Swiss franc exchange rate in the extraordinary period from September 6, 2011 and January 15, 2015 when the Swiss National Bank enforced a minimum exchange rate of 1.20 Swiss francs per euro. Based on the analogy between Brownian motion in finance and physics, the first-order effect of such a steric constraint would enter a priori in the form of a repulsive entropic force associated with the paths crossing the barrier that are forbidden. Non-parametric empirical estimates of drift and volatility show that the predicted first-order analogy between economics and physics are incorrect. The clue is to realise that the random walk nature of financial prices results from the continuous anticipations of traders about future opportunities, whose aggregate actions translate into an approximate efficient market with almost no arbitrage opportunities. With the Swiss National Bank stated commitment to enforce the barrier, traders's anticipation of this action leads to a vanishing drift together with a volatility of the exchange rate that depends on the distance to the barrier. We give direct quantitative empirical evidence that this effect is well described by Krugman's target zone model [P.R. Krugman. The Quarterly Journal of Economics, 106(3):669-682, 1991]. Motivated by the insights from this economical model, we revise the initial economics-physics analogy and show that, within the context of hindered diffusion, the two systems can be described with the same mathematics after all. Using a recently proposed extended analogy in terms of a colloidal Brownian particle embedded in a fluid of molecules associated with the underlying order book, we derive that, close to the restricting boundary, the dynamics of both systems is described by a stochastic differential equation with a very small constant drift and a linear diffusion coefficient.

q-fin.ST↗

arXiv2015

Financial Knudsen number: breakdown of continuous price dynamics and asymmetric buy and sell structures confirmed by high precision order book information

We generalise the description of the dynamics of the order book of financial markets in terms of a Brownian particle embedded in a fluid of incoming, exiting and annihilating particles by presenting a model of the velocity on each side (buy and sell) independently. The improved model builds on the time-averaged number of particles in the inner layer and its change per unit time, where the inner layer is revealed by the correlations between price velocity and change in the number of particles (limit orders). This allows us to introduce the Knudsen number of the financial Brownian particle motion and its asymmetric version (on the buy and sell sides). Not being considered previously, the asymmetric Knudsen numbers are crucial in finance in order to detect asymmetric price changes. The Knudsen numbers allows us to characterise the conditions for the market dynamics to be correctly described by a continuous stochastic process. Not questioned until now for large liquid markets such as the USD/JPY and EUR/USD exchange rates, we show that there are regimes when the Knudsen numbers are so high that discrete particle effects dominate, such as during market stresses and crashes. We document the presence of imbalances of particles depletion rates on the buy and sell sides that are associated with high Knudsen numbers and violent directional price changes. This indicator can detect the direction of the price motion at the early stage while the usual volatility risk measure is blind to the price direction.

q-fin.TR↗