Deep Learning for High-Frequency Stock Price Direction Forecasting
Summary
The document summarizes machine learning applications in quantitative finance, focusing on a study that uses recurrent neural networks with LSTM units to predict up or down price moves from high-frequency stock data. It reports that a model trained across a pooled set of stocks improved forecast accuracy by up to 10% relative to high-frequency VAR models, and that the pooled model outperformed stock-specific models while generalizing to stocks outside its training sample.
The summary describes evidence that order flow and price history contain patterns shared across stocks, with forecasting performance reported as stable over an out-of-sample period. It also lists other suggested reading, including work on machine learning in asset pricing and a book on financial machine learning. The main caveat is that predictive accuracy alone does not show that a strategy is profitable: transaction costs and latency were not assessed. The document offers a selective reading lead, not a comprehensive survey of recent developments.
Key ideas
- An LSTM model is described as forecasting high-frequency stock price direction from order-flow and price data.
- The cited study reports improved accuracy over VAR models and gains from pooling data across stocks.
- The pooled model is reported to generalize to stocks absent from its training sample.
- The document says that history contributes predictive information and performance was stable out of sample.
- Trading profitability remains untested because costs and latency were outside the study's scope.
Tags
Full text
# What are some interesting recent machine learning related developments in the QF domain?
# What are some interesting recent machine learning related developments in the QF domain?
In 2020 I wrote a MSc thesis on the hedging of exotic options using recurrent neural networks (loosely based on the paper Deep Hedging (2018)by Buehler et al.).
Since then I have been interested in all types of ML applications in the financial domain, but I haven't been reading any of the recent publications/following any of the developments.
Hence my question is as follows: In the past few years, have there been any particularly interesting publications/developments in terms of machine learning being used in the QF domain?
## Answer by Pleb (score 7)
https://quant.stackexchange.com/a/71425
### Sirignano, J., & Cont, R. (2019) ( High-frequency stock forecasting ):
The authors apply a large-scale deep learning model (recurrent neural network with Long Short-term Memory units) to high-frequency financial stock data in order to forecast the next price direction (up or down).
#### Results:
- Not only do they uncover that the prediction accuracy for the deep learning model substantially increases by up to 10% contrary to high-frequency VAR-based models, but they also observe how a "universal" deep learning model trained on a pooled dataset of 500 stocks is able to capture relations between order flow and price variations which are common to all stocks in the dataset.
- The universal deep learning model is able to outperform the stock-specific counterpart and most importantly is able generalize to stocks not included in the training set: if the model was trained on data for $\{1,\ldots,N\}$ stocks, then it can maintain the out-of-sample forecast accuracy for stock $N+1$. In conclusion, the model might viably predict the price direction on illiquid stocks where econometric models might falter due to lack of data.
The authors neatly summarizes the main findings of their deep learning model:
> Nonlinearity: Data-driven models trained using deep learning substantially outperform linear models in terms of forecasting accuracy (Section 3.1). Universality: The model uncovers universal features that are common across all stocks (Section 3.2). These features generalize well: they are also observed to hold for stocks which are not part of the training sample. Stationarity: The model performance in terms of price forecasting accuracy is remarkably stable across time, even a year out of sample. This shows evidence for the existence of a stationary relationship between order flow and price changes (Section 3.3), which is stable over long time periods. Path-dependence and long-range dependence: Inclusion of price and order flow history is shown to substantially increase the forecast accuracy. This provides evidence that price dynamics depend not only on the current or recent state of the limit order book but on its history, possibly over long time scales (Section 3.4).
It should be noted that the paper only focuses on the predictive accuracy of the deep learning model and do not determine whether this predictability leads to profitable trading strategies. Implementing transaction costs and accounting for latency in the high-frequency data would be out of the scope of the paper.
Additional Info: There is a Quant Stack Exchange question asking about appropriate machine learning models for financial time-series forecasting. Here, the highest voted answer has a good list of recent papers, including the paper provided in @MiLuk's answer.
## Answer by TwoII (score 6)
https://quant.stackexchange.com/a/71424
Empirical Asset Pricing via Machine Learning (2020) by Gu, Kelly and Xiu
## Answer by IDontKnowCode (score 0)
https://quant.stackexchange.com/a/71447
Advances in Financial Machine Learning(2018) by Lopez De PradoShown 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.