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
This guide explains two ways to schedule asynchronous work in an event-driven trading application. A loop-run task registers an asynchronous callback at a specified interval, measured in seconds, and returns an identifier that can later be used to unregister…
The document explains how a trading application can use a remote procedure call (RPC) service to share events and handle requests across separate processes. It frames RPC as a way to work around Python’s global interpreter lock limiting CPU-bound work in a…
The document outlines several mechanisms that can produce negative autocorrelation in short-horizon returns. The classic explanation is bid-ask bounce: trades initiated by buyers and sellers alternate between ask and bid transaction prices, creating…
The document outlines ways high-frequency traders can structure equity trading to manage commissions and exchange fees. One route is direct exchange access through a broker identifier, allowing a firm to pay or receive the exchange’s active or passive fees…
The discussion asks whether high-frequency foreign-exchange ask quotes are considered less reliable than bids. The answers reject that general rule: they describe FX quotes as actionable and argue that both sides are needed to construct a mid-price and then…
The document concerns imbalance bars and run bars, sampling methods discussed in Advances in Financial Machine Learning. The author is implementing them and asks for further literature, explaining that the material consulted so far emphasizes volume…
The document distinguishes a speed advantage in reacting to market data from arbitrage in the stricter sense of exploiting price differences. It describes how direct exchange feeds can reveal quote changes before a slower consolidated feed, allowing a…
The document questions whether standard GARCH models adequately describe one-minute Bitcoin returns. The sample is characterized by frequent zero returns, low return magnitudes, and apparent microstructure noise; the author also reports that squared…
The document describes the inventory challenge faced when quoting many option strikes and levels in a fast market. Frequent fills change the portfolio’s delta, but immediately hedging every fill with taker orders can incur substantial spread costs. Waiting…
The document explores why historical replay alone may not faithfully test market-making or other microstructure strategies. A replay uses recorded order-book events, but cannot fully capture how other participants would react to the strategy's orders. The…
Realized variance is an estimate of intraday variation, not by itself a model for forecasting volatility across days. To forecast future volatility, a model must specify how information from realized measures relates over time. The document describes…
The document discusses whether stochastic-control models built around the Hamilton–Jacobi–Bellman equation have practical value in market making and algorithmic trading. It describes their role as research and teaching tools for reasoning about quote widths,…
The discussion distinguishes computational limits by workload. For real-time trading, it argues that network throughput is often the constraint, while historical analysis is more commonly limited by memory capacity or bandwidth; ordinary CPU use may not be…
The discussion argues that the right graduate education for quantitative finance depends on the work a person wants to do. Financial engineering programs may be especially relevant for valuation and pricing model development, while quantitative trading has…
The document explains why high-frequency trading can remain relevant when a market allocates fills pro rata, reducing the importance of queue position. It distinguishes liquidity provision from liquidity taking: speed may matter less for winning a place in…
The document compares Excel DDE and ActiveX connections for streaming market quotes and explains why an RTD server may be a better option when faster spreadsheet updates are needed. RTD supports a configurable refresh interval, including millisecond…
The exchange explains what trade, quote, and order book feeds reveal about order types. Public feeds generally do not label market orders or iceberg orders. Matching trades with identical timestamps may help approximate market order activity, but it can…
The document discusses whether QuickFIX is fast enough for an order management system and argues that the answer depends on measured requirements and the full processing architecture. A slow component elsewhere, including network communication, can dominate…
The document compares ways to calculate a bond’s yield to maturity from price when repeated updates make speed important. One approach computes yields for three selected prices, fits a quadratic price–yield relationship, and uses it to estimate yields for…
The document asks whether hourly observations can be used to estimate dynamics that matter over a horizon of several days, particularly mean reversion or cointegrating relationships. It refers to a proposed analogy between estimating monthly relationships…
The discussion models exchange round-trip latency as the sum of network transit time and matching-engine message-processing time. For venues in separate data centers, network distance and routing usually dominate: fiber routes are indirect, switches and…
The document considers why a stock price can move repeatedly even when company news does not arrive at the same pace. It presents price as a noisy proxy for underlying value: after news, market participants may take time to interpret its implications and…
The note explains why drift is often neglected when modeling asset prices at very short time intervals. For geometric Brownian motion with constant drift and volatility, it compares how each component scales as the interval shrinks: drift scales linearly…
The document asks how to select the optimal bandwidth for a realized-kernel volatility estimate when the desired output uses one-minute bins. It gives the bandwidth formula from the cited reference and notes that its noise-to-signal input uses integrated…