Forecasting High-Frequency Prices: Microstructure, Direction, and Simple Models
Summary
The document considers forecasting actual high-frequency prices, including bid and ask quotes, rather than forecasting volatility. It highlights practical challenges: model validity may be brief, validity is difficult to assess, and computationally expensive models may not be recalibrated in time. The questioner reports that only a few methods outperform naive forecasts in some short subperiods, naming Kalman filters and ARMA or ARIMA among the better approaches in their own research, while several smoothing and regression methods did worse. These are personal observations, not a controlled comparison.
Responses recommend first understanding market microstructure: exchange order handling, order-book behavior, trades, and price formation, then forming and testing hypotheses. Another response suggests predicting direction rather than exact prices because high-frequency series are noisy and computation is constrained; it mentions classifiers and signal-extraction methods such as Kalman filters, ARMA, PCA, and Fourier analysis. The document offers guidance rather than a validated forecasting recipe, and it gives no datasets, metrics, or reproducible results for the proposed approaches.
Key ideas
- High-frequency price models can lose validity quickly, making monitoring and timely recalibration practical concerns.
- The questioner reports that several classical forecasting methods failed to beat naive forecasts in their own experiments.
- Understanding exchange mechanics, order books, and trades can inform hypotheses about short-term price formation.
- One response favors predicting price direction when exact price forecasts are noisy and computationally demanding.
- The suggested methods are not supported by shared datasets or reproducible performance comparisons.
Tags
Full text
# How to forecast high-frequency data? # How to forecast high-frequency data? Introduction: I have seen a plenty of articles/books regarding volatility forecasting applied to high frequency data, but none of them were dedicated to forecasting the actual prices (for example bid/ask of currency pairs, stock prices). Example: Forecasting Volatility using High Frequency Data (P. R. HANSEN, A. LUNDE) I realize that forecasting hft data must be hard because of several problems: - How long model will be valid? - How to check such validity? - Will a new model be calculated on time? (NN and most of subtypes of NN need a lot of time to be recalculated) Questions: Given the above mentioned problems/restrictions I would like to ask you: - What's the best way of forecasting high-frequency data? - What kind of scientific papers/books should I follow in order to be up-to-date in this topic? Personal feedback: From my own research based on hft data I can write that generally only few models are better than naive forecasting (but only in some sub time series - few minutes etc.) - Kalman Filtering (calibrated accordingly) - ARIMA, ARMA models Generally worse than naive forecasting are: - Holts linear model (ARIMA (0,2,2) - Linear Regression - Exponential Smoothing - Moving averages (different types) ## Answer by wildbunny (score 1) https://quant.stackexchange.com/a/44132 A good place to start would be to understand the environment in which HFT takes place. What causes prices to form on the lowest possible level? What mechanisms do exchanges use to process orders, how would they affect the price discovery process? This in generally known as market-microstructure, and there is a lot of literature available on the subject, but you should also augment that with just physical observation of the orderbook, trades and the resulting price movements to get a better understanding of the whole process. Once you have an understanding of what you are dealing with, you can start to build and test hypothesis. ## Answer by Vitomir (score 0) https://quant.stackexchange.com/a/46563 at HF it is probably more important to guess the direction (up/down) than actual prices. For that you will have a lot of noise and financial time series have low signal to noise ratio. that is why classifiers (up/down) are the best practice in HF, not punctual predictors of prices. This has also to do with the computational complexity required to forecast a classifier vs a punctual value (simply put, in HF you need to cut down running time, thus computational complexity, in my experience many strategies fail due to implementation time and computational complexity). In any case, Kalman filters, ARMA, PCA, Fourier analysis are all good methods to extract signal. Once you have extracted the signal (i.e. removed most of the noise) then you should move on with other kind of models.
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.