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 describes where C++ is used in high-frequency trading and why firms rely on it. The main use is the latency-sensitive path that receives market data, updates forecasts or risk information, makes a trading decision, and sends, changes, or cancels…
The discussion compares C++, C#, Java, and Python for quantitative finance and argues that language choice depends on the work being done. C++ is described as common for application backbones and computationally demanding tasks such as derivative pricing,…
The document explains why electronic order behavior cannot be learned from one universal description. Exchanges differ in the order types they support and in their matching rules, which determine matters such as queue priority and the effects of order…
The document compares ways to represent an order book while supporting price ordering, fast updates, and access to the best bid and ask. For order-based equity books, one detailed proposal combines an order-ID lookup structure with metadata linking each…
The document explains that a useful forecast target depends on the market-taking task. For a large order, a short-horizon mid-price forecast can help time execution. For a simple single-exchange strategy, the relevant question is whether prices will cross…
The document considers whether Bulk Volume Classification (BVC), which estimates the buy-trade share within a bar, should be applied to short or long bars. One response argues that larger bars provide more trades and therefore tend to yield more accurate…
This exchange addresses a market data session that stopped receiving snapshots after a broker reported missing client heartbeats. The key diagnostic is that client heartbeats and server initiated test requests are distinct messages with different expected…
The discussion examines whether top-of-book queue sizes help predict the direction and timing of the next price move. The questioner implements a Markovian limit order book model and finds that empirical probabilities conditioned on bid and ask size are…
The document explains the Roll model, which estimates the bid-ask spread from the negative autocovariance of consecutive high-frequency price changes. The negative relationship arises when trades alternate between the bid and ask: transaction prices move…
The document explores whether some constraints in high-frequency trading apply broadly, even to firms with substantial spending on hardware, software, and co-location. The response discusses data-center scale, venue fees, processor hardware, co-location, and…
The document discusses coordinating two high-frequency agents that trade opposite sides of a market while managing inventory and exposure limits. It frames market making as a control problem in which quotes must balance expected execution quality against the…
The document asks for practical accounts of how trading, high-frequency trading, or asset-pricing companies are established, including choices about data sources and hiring. The answers point to a general high-frequency trading book and to firsthand material…
The document discusses why high-frequency traders submit and quickly cancel limit orders. One explanation is routine event-driven market making: algorithms update quotes as order-book events change their estimate of fair value, producing frequent…
The discussion explains how a strategy with a slower decision schedule can use high-frequency trades, quotes, or order-book updates without treating every update as a trading signal. The proposed approach is to calculate features from the updates, such as…
The accepted response argues that an internet-connected retail brokerage API is unsuitable for high-frequency market making. It cites limits on streaming market data and API messaging, restrictions on active orders, and cancellation fees as barriers to…
The document asks why the tick test is used to infer whether trades are buyer or seller initiated. The rule labels a trade by comparing its price with the previous different trade price: an uptick suggests a buy, while a downtick suggests a sell. The…
The document asks how to connect models written in R to an Interactive Brokers account for live trading. It considers whether the trader must write a C or C++ interface, whether an R-only route is available, and whether the IBrokers package remains usable.…
The document compares field-programmable gate arrays with CPUs and other accelerators in trading systems. Its central distinction is that FPGAs can run many simple, repeated operations in parallel, making them suitable for low-latency feed parsing, data…
The document asks how traders might infer very short-term price direction from a cryptocurrency limit order book. It describes a trader who takes long or short positions over horizons ranging from seconds to a few minutes, apparently using visible order book…
The discussion compares ways for a Europe-based high-frequency trader to obtain real-time S&P 500 E-mini futures prices. Responses distinguish retail data services from more expensive institutional arrangements that offer market data and access to points of…
The document distinguishes quantitative research from production execution when choosing programming languages. It presents R and Python as suitable environments for research and alpha-model development, while noting that a complete trading system must also…
The document compares storage approaches for real-time and historical market data, focusing on column-oriented databases, row stores, in-memory arrays, flat files, and time-series systems. Column stores can speed analytical reads by scanning selected columns…
The document discusses ways to reduce the time needed for repeated strategy backtests. It suggests identifying the data the strategy actually uses and maintaining a buffer of relevant observations as new events arrive. For computationally intensive logic, it…
The document considers whether exchanges could use proof-of-work challenges to discourage rapid order submission and cancellation, potentially limiting information probing, manipulation, or advantages from proximity. It explains that proof of work is more…