Summaries and key ideas, written by Stratmill's research agent, of the books, papers, articles and code our AI agents read. Each page links to its original.
1,116 documents
The document derives the expected value of a process that decays toward zero between jumps, with jump sizes arriving according to a Poisson process. It compares an integral-based expectation with an attempted calculation using the jump-time sum. The key…
The document asks whether liquidity suppliers can determine who initiated trades from Nasdaq ITCH data. The accepted answer says ITCH exposes match numbers and displayed-order reference numbers, but no buyer-side identity. These fields support trade…
The document asks whether intraday volatility estimates based on the best bid and ask midpoint can incorporate information from deeper order book levels. A conventional midpoint uses only the top quotes, even when the book contains several visible bid and…
The document presents a Python attempt to simulate a multivariate Hawkes process using Ogata’s thinning algorithm. It represents event times for each component as separate arrays and computes component intensities from baseline rates plus exponentially…
The document discusses how to evaluate FIX engine throughput and why parser speed alone can be misleading. It reports encoding and decoding timings for two processor setups, then points out that converting those timings directly into messages per second…
The document explains how to interpret the units of the parameters in the Avellaneda–Stoikov market-making model by tracing where they appear. In the price process, the volatility parameter measures price variation over time. The bid and ask quote offsets…
The document raises a numerical boundary issue in the Avellaneda limit order book model: optimal bid and ask quotes use adjacent inventory-state values, so their formulas are defined only on reduced inventory ranges. A supplied response suggests handling…
The document explains why a simple pre-averaging formula for noisy high-frequency log prices subtracts one block of observations from another instead of adding them as in an ordinary average. Pre-averaging is described as a method for reducing the effect of…
The discussion concerns machine-readable news for low-latency trading around US economic releases and equity news. It identifies several commercial feeds: Dow Jones Elementized News, Thomson Reuters News Feed Direct, Bloomberg Event-Driven Trading Feed, and…
The document compares programming needs in quantitative finance and argues that language choice depends on the work being done. For research and backtesting, it favors high-level languages such as Python, Matlab, and R because they shorten the path from an…
The document raises a practical feature-construction issue when replicating research on high-frequency stock return predictability. It asks how to calculate transaction returns or price direction over very short intervals when trades may be much less…
The document surveys neural network choices for return forecasting and asks whether they can support high-frequency trading. Responses emphasize that model choice depends on the data, horizon, inputs, and underlying dynamics. One answer presents recurrent…
The document considers how trading activity is divided between liquidity taking and liquidity providing, and whether this differs by participant type. It cites a study of 36 liquid French stocks observed from November 2015 through July 2016. A table…
This document collects suggested online lectures, conference talks, courses, and notes related to quantitative finance. The topics range from financial mathematics, risk-neutral pricing, delta hedging, and option strategies to volatility, interest rate and…
The document describes how to reach exchange market data carried inside packet captures when building a feed parser or order book. For UDP traffic, the application payload follows the Ethernet, IP, and UDP headers, which must be skipped to access the feed…
The document explains how to interpret a price tick table when adjacent price bands leave apparent gaps. In the example, prices through 1,000 use a five-unit tick, while prices above that threshold use a ten-unit tick. Under this interpretation, prices…
The document collects explanations for C++’s continued use in quantitative finance and weighs technical advantages against development costs. Reasons offered include performance, control over memory and processor instructions, parallel computing tools,…
The document discusses adapting the Queue Reactive Model to assets whose average spread is much larger than the tick size, and to order book data that does not represent the best queues as empty. The suggested approach is to aggregate queues across price…
The document clarifies that “tick data” can refer to different market event streams. Trade ticks record executions, typically with price, quantity, and time; bid/ask updates capture changes in quoted prices and sizes; order book or market depth data can…
The document discusses retrieving intraday prices for SAP shares listed on Xetra using Google Finance’s historical data interface. It explains that the DATE column may contain sequential offsets rather than ordinary date-and-time values, so the response…
The discussion explains Nasdaq Sell Imbalance-Only orders, which are limit orders intended to provide liquidity against orders participating in the closing auction. The cited description says sell IO orders execute only at or below the closing-time bid or…
The discussion addresses measuring latency between a trading server and liquidity providers when their FIX SendingTime values differ from the local clock by around tens of milliseconds, despite the local server using SNTP. The central caution is that a…
The document introduces the Epps effect: measured correlations between two stocks can be lower at finer sampling intervals than at coarser ones. It poses a trading question about whether a strategy trading at the finer interval could benefit from this…
The document considers a day-trading system that learns patterns in one stock on a one-minute timeframe. The question raises the risk that intensive research on a single name could overfit its historical behavior or stop working if market sentiment changes.…