research
Publications, talks, and the questions I keep coming back to.
ORCID 0009-0001-6146-8606
# interests
My background is in physics, and most of what I work on sits where mathematics meets implementation. Three threads run through it:
Econophysics — phase transitions in leveraged markets
This is the active program, not an abstract interest. Crypto perpetual futures are an unusually clean laboratory: the destabilising mechanism is written down in the exchange rulebook rather than inferred, and the leverage and order-flow panel is published alongside price. The line asks, in order: does a cascade announce itself before it fires, and in which state variable? If nothing univariate does, what order of transition is this, and how hard does the amplification actually run? And then, what would a model have to contain to reproduce it? The papers are listed below.
Complex systems
Emergent structure in networks and large interacting systems — how local rules produce global behaviour. This includes applying network analysis to domains outside physics, such as the structure of legal codes.
Cryptography — implementation & study
I study cryptography by implementing it. Recent labs cover the short integer solution problem (SIS), the number-theoretic transform, and an IOP-style proof system (a Baby-Ligero prover). The goal is to understand the primitives before reaching for libraries — the writing collects those notes.
# publications
Measuring the engine of a liquidation cascade: subcritical branching inside a first-order transition
Ramon Marc Garcia Seuma
arXiv preprint — physics.soc-ph, cross-listed q-fin.ST · 2026
PreprintPart I found no early-warning signal that survives across events. This paper asks the next question: if these cascades are not critical transitions, what order of transition are they, and how hard does the amplification actually run? Across seven BTC liquidation cascades we measure the correlation fabric with a mean pairwise coupling as order parameter and a susceptibility proxy, and find a discontinuous jump with no divergence and no finite-size drift — the signature of a first-order transition rather than a critical one. We then measure the engine itself on the 10 October 2025 event, reconstructing forced flow from a transaction-level fill log: the branching ratio runs at roughly 0.1–0.2 through nucleation and peak, i.e. firmly subcritical, with 87.8% of post-onset forced sells landing in the first thirty minutes and the backstop vault absorbing the climax. The record cascade was an exogenous front-loaded sweep, not a chain reaction. Severity is set by the shock, the map into the path, and the liquidity withdrawn — none of which is gradeable in advance.
Where does the criticality live? Early-warning signals are event-heterogeneous across seven crypto-perpetual liquidation cascades
Ramon Marc Garcia Seuma
arXiv preprint — q-fin.ST (Statistical Finance), cross-listed physics.soc-ph · under review at Physica A · 2026
PreprintDo crypto perpetual-futures crashes carry a reproducible early-warning fingerprint of a critical transition, and in which state variable? We study seven major BTC liquidation cascades (2022–2025, including the record $19B event of 10 October 2025) using minute-level price and 5-minute leverage/order-flow data. On detrended residuals we compute rolling variance and lag-1 autocorrelation and test their pre-cascade trend with the Kendall-τ statistic, sweeping 39 analysis configurations per variable per event. No variable is event-invariant. Price carries the critical-slowing-down signature in five of seven events but is silent in exactly the two sudden-news (tariff) shocks, suggesting a two-type structure: endogenous-buildup versus exogenous-shock cascades. The October 2025 event — whose in-sample analysis suggests the signature lives in leverage rather than price — turns out to be the outlier, not the rule. The one regularity surviving all events with data is a compression of taker order-flow variance, which passes a 300-onset placebo test (Fisher-combined p ≈ 5×10⁻⁶) but is a population-level precursor, not a per-event alarm. Single-event critical-slowing-down claims in crypto derivatives are therefore fragile by construction. We argue the pattern of failures is itself diagnostic: slowing down is absent exactly where the destabilising mechanism is most abrupt, as one would expect if these cascades are discontinuous, shock-driven transitions rather than critical ones.
From Interview to Compromise: A Case Study of a Targeted Web3 Supply-Chain Attack
Ramon Garcia Seuma
Procedia Computer Science — iSCSi 2026, International Conference on Industry Sciences and Computer Sciences Innovation, Azores, Portugal (May 20–22, 2026) · Elsevier (ScienceDirect) · 2026
Accepted — forthcomingA detailed case study of a targeted supply-chain attack delivered through a Web3 recruitment workflow. The attack leveraged social engineering via professional networking platforms to distribute a malicious development environment, ultimately enabling command-and-control (C2) communication and potential remote code execution. The analysis is based on a full forensic dataset — packet captures, extracted C2 streams, and a snapshot of the malicious project environment — to reconstruct the complete attack chain from initial contact to payload execution. The paper compares the observed techniques with previously documented campaigns targeting job seekers and proposes practical detection heuristics for developers and organizations operating in the Web3 ecosystem.
A market model for exploitation and cooperation using the Minority Game
Ramon Marc García Seuma
MSc Thesis — IFISC (UIB-CSIC), supervised by Pere Colet · 2019
PublishedThe Minority Game, the mathematical formulation of Brian Arthur's "El Farol Bar" problem, embodies basic market mechanisms while keeping mathematical complexity to a minimum. Agents seek to differentiate from competitors and end up in the minority group. This thesis analyses a model for market dynamics based on the Minority Game and extends it to explore market features, cooperation and competition.
Information cascades in complex social networks
Ramon Marc García Seuma
BSc Thesis (TFG) — Universitat de Barcelona, supervised by Albert Díaz-Guilera · 2017
PublishedAn undergraduate study of how information cascades propagate on complex social networks: cascade-size distributions on Barabási–Albert and modular (FARZ) graphs, the role of node position and clique structure in whether a cascade spreads or dies, and an integrate-and-fire model of activation on random graphs. It is a small piece of work, but it is where the thread starts: the same question about whether a system telegraphs a cascade before it fires, asked about social networks a decade before I asked it about leveraged markets.
# talks
From Interview to Compromise: A Case Study of a Targeted Web3 Supply-Chain Attack
iSCSi 2026 — International Conference on Industry Sciences and Computer Sciences Innovation · Vila Galé Collection São Miguel, Azores, Portugal · 5/21/2026