Building Short-Horizon Price Signals from Limit Order Flow
Summary
The document asks how to predict near-term best bid or ask price direction from high-frequency limit order book data. One answer recommends reconstructing the actual book from a venue’s full order log when available, rather than relying on a simulated book. It also notes that simple top-of-book opportunities can disappear very quickly, emphasizing the role of data, analysis, and order-delivery latency.
Another answer suggests building order-book signals with information-driven sampling, such as tick- or volume-imbalance bars, and considering sequence models for forecasting incoming orders. Other candidate features include cancelled orders, market impact, volume-weighted prices, and implementation shortfall. These are starting points rather than a validated recipe: the question covers multiple horizons and instruments, and the replies provide no experimental results, feature definitions, or evaluation protocol. Any prediction should be assessed with time-appropriate validation and realistic execution assumptions, since a directional forecast alone does not establish a tradable edge.
Key ideas
- Full venue order logs can support reconstruction of the actual limit order book.
- Order-book features can be sampled using tick or volume imbalance bars.
- Candidate signals include cancellations, market impact, volume-weighted prices, and implementation shortfall.
- Sequence models are one possible way to forecast subsequent order flow.
- Latency and execution conditions matter because short-lived top-of-book signals may vanish quickly.
Tags
Full text
# Predicting price direction from order flow at high frequency # Predicting price direction from order flow at high frequency I have access to high frequency data for a few instruments using which I can simulate a limit order book.I would like to predict direction of price(best bid/ask) in the short term(1 sec, 5 sec and 10 sec) using that. What would be a good model/reference point to start with? Additionally, if there are any research papers/books on similar problems, please let me know. ## Answer by Sergei Rodionov (score 1) https://quant.stackexchange.com/a/63954 If you have access to full order log from a trading venue, you can build (not simulate) the actual limit order book, with tick by tick changes. Basic top-of-book mispricing is arbitraged away within 500 μs which includes getting the market data, updating the book, performing analysis, issuing an order and delivering it to the exchange. Check out IEX SEC filings and the rule book for CQI formula to get a sense of moving parts. ## Answer by Felix (score 0) https://quant.stackexchange.com/a/63936 Your question is quite broad, but I try my best to give you some hints to tackle this: To predict the price direction, you need to build a signal from your order book. I recommend Information-driven bars like Tick Imbalanced Bars or Volume Imbalanced Bars. Then you can run a LSTM or something to get a good prediction of the next incoming orders. However, there are many many ways to construct signals from the order book, so depends on your intend. If you have access to the cancelled orders as well, you can calculate: market impact = |execution price - bendchmark price| * shares executed at execution price You can also calculate some volume-weighted average price (VWAP) or the imlementation shortfalls to get some signals from the order book. Good approaches can be found here: - De Prado, M. L. (2018). Advances in financial machine learning. John Wiley & Sons. - Jansen, S. (2020). Machine Learning for Algorithmic Trading: Predictive Models to Extract Signals from Market and Alternative Data for Systematic Trading Strategies with Python. Packt Publishing Limited.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.