Eugen Soloviov
Trading-systems engineer
Trading-systems engineer building bots since 2017: cross-exchange arbitrage (connected up to 30 venues), cointegration-based pairs arbitrage across spot and futures, scalping, news and sentiment-driven strategies, trend algorithms, and portfolio management and balancing algorithms. Also builds sub-millisecond order execution, big-data warehouses, backtesting engines, AI agents, and trading interfaces (incl. open-source profitmaker.cc). Stack: JS/TS, Python, Rust/Zig/Go, DevOps, backend, frontend, architecture.
Articles
Model Pruning for Low-Latency Trading Inference
Magnitude and structured pruning, the Lottery Ticket Hypothesis, movement pruning, distillation and 2:4 sparsity — the methods behind shrinking a trading model, and what still has to be measured before any of it ships.
Updating the Volume Curve Intraday: Does Adaptive Forecasting Actually Help?
Our VWAP article shipped a static, weekly-refit volume curve and called the forecaster the weakest link. This is the follow-up: a Bayesian intraday updater run against the same 500-parent BTCUSDT harness, with the IS delta conditioned on realized curve error.
Koopman Operators and DMD: Do Market Modes Survive Out-of-Sample?
Dynamic Mode Decomposition fits a linear operator to nonlinear market dynamics. The only question that matters: do the fitted modes persist from one window to the next, and does the rolling spectral radius lead realised volatility? Here is the measurement protocol and the code to run it.
Knowledge Distillation: Compressing Trading Models for Low-Latency Deployment
The blog's standing answer to the accuracy-vs-latency tension is a two-stage fast/slow split. Distillation is a different answer: train one small model to mimic the ensemble. The KD loss, temperature, born-again nets, early exits for a variable latency budget, and the distill-to-FPGA pipeline — plus the measurements that would decide whether it beats the two-stage split.
The Other Way a Regression Lies: Endogeneity, 2SLS, and the Gamma Calibration Problem
Selection bias in the search is not the only way a regression fools you. When the regressor is correlated with the error, more data makes the estimate more confidently wrong. Instrumental variables applied to the one endogeneity problem this blog has already left open: permanent impact from net taker flow.
Hawkes Processes for Order Arrival and Market Event Modeling
Fitting a self-exciting point process to real crypto trade tape: where the three numbers (mu, alpha, beta) come from, how to estimate the branching ratio n, whether the exponential kernel survives a goodness-of-fit test, and how much n moves when you change the estimation window.
Hamiltonian Neural Networks: Does a Financial System Conserve Anything?
Hamiltonian Neural Networks are provably stable — but stability is worthless if the conserved quantity does not exist. Learning a scalar H via autograd, symplectic integration, and the falsification test that decides whether a financial (q, p) pair is canonical at all.
Toda-Yamamoto vs Differenced Granger: Does the BTC Lead-Lag Survive?
Granger causality on crypto prices done two ways — differenced returns and Toda-Yamamoto on levels — with a correct Wald implementation, an effective-N corrected causality matrix, and a rolling-stability test of whether the lag is tradeable at all.
Procesos gaussianos para modelado de precios no paramétrico
Diseño de kernel para series de tiempo financieras (rugosidad materna, composición periódica local, mezclas espectrales) y la probabilidad marginal como regularizador que no necesita conjunto de validación. Además de la lista honesta de lo que aún debe medirse antes de que algo sea comercializable.
Operador neuronal de Fourier para modelado financiero basado en PDE
El aprendizaje del operador mapea espacios funcionales completos, no puntos: cómo el operador neuronal de Fourier parametriza un operador de solución PDE en el espacio de frecuencia, qué compra para el precio de opciones y cuáles de sus afirmaciones aún necesitan medirse en hardware real.
¿Un contexto de día completo supera a uno de diez minutos? Atención flash y la pregunta sobre la longitud de la secuencia
Flash Attention hace que un contexto comercial de 23,400 pasos sea computacionalmente libre: mosaico, softmax en línea y un límite de IO de N^2 d^2 / M. Si ese contexto más largo mejora el modelo es una cuestión aparte, y debe medirse, no afirmarse.
Métodos conjuntos: combinación de alumnos débiles para una alfa sólida
Tres cosas sobre el comercio de conjuntos que este blog no ha cubierto: por qué el término de covarianza (no el recuento de modelos) domina el error del conjunto, cómo funciona el apilamiento como una capa de combinación de señales con metacaracterísticas desplegadas y si la compensación de rotación entre muchas señales en realidad reduce el costo de ejecución.