Summary
Predicting stock returns is tough because models are often big and slow to update. The authors used a method that evolves small neural networks each day to quickly adjust to new data. They ran this method on a computer that sends updated neural nets to a small device, like a Raspberry Pi, which makes fast daily predictions for stocks. This approach performed much better than common prediction models on real stock data from 2022 to 2024.
What this means in practice
- •For quantitative trading teams: Deploy evolving small neural networks to improve daily stock return predictions on low-cost edge hardware.
- •For embedded systems developers: Implement real-time, evolving neural networks on resource-limited devices for fast prediction tasks with frequent updates.
Abstract
Accurate forecasting models are usually large, expensive to update online, and fixed in architecture once trained. We apply ONE-NAS, an online neuroevolutionary architecture search that evolves a population of small recurrent networks as each window of data arrives, to daily cross-sectional stock return prediction, and pilot it on a host and endpoint pipeline: the host runs the search and ships each generation's champion genomes over TCP/IP to a Raspberry Pi 4B, which predicts online. On the Pi a single champion predicts a 50-stock window in 24.6~ms and the ensemble of 40 island champions in 556~ms, far inside the daily decision cycle. On four panels of US mid-cap equities over 2022--2024, reading the population as a rank-mean ensemble of island champions returns $+27.5\%$ net of realised transaction costs, against $+11.3$ to $+14.8\%$ for online LSTM, online GRU and monthly-retrained LSTM baselines and $+4.5\%$ for the single best genome used in prior ONE-NAS work.