Paper: arXiv 2307.06339
Authors: Kosuke Tatsumura, Ryo Hidaka, Jun Nakayama, Tomoya Kashimata, Masaya Yamasaki
Abstract
Financial portfolio construction problems are often formulated as quadratic and discrete (combinatorial) optimization that belong to the nondeterministic polynomial time (NP)-hard class in computational complexity theory. Ising machines are hardware devices that work in quantum-mechanical/quantum-inspired principles for quickly solving NP-hard optimization problems, which potentially enable making trading decisions based on NP-hard optimization in the time constraints for high-speed trading strategies. Here we report a real-time stock trading system that determines long(buying)/short(selling) positions through NP-hard portfolio optimization for improving the Sharpe ratio using an embedded Ising machine based on a quantum-inspired algorithm called simulated bifurcation. The Ising machine selects a balanced (delta-neutral) group of stocks from an \(N\)-stock universe according to an objective function involving maximizing instantaneous expected returns defined as deviations from volume-weighted average prices and minimizing the summation of statistical correlation factors (for diversification). It has been demonstrated in the Tokyo Stock Exchange that the trading strategy based on NP-hard portfolio optimization for \(N\)=128 is executable with the FPGA (field-programmable gate array)-based trading system with a response latency of 164 \(μ\)s.
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 6.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced mathematical concepts such as NP-hard combinatorial optimization, quadratic unconstrained binary optimization (QUBO), and Ising models, demonstrating high mathematical density. Empirically, it is anchored by a real-time trading system tested on the Tokyo Stock Exchange with specific latency measurements (164 µs), but the excerpt lacks detailed backtesting metrics or comprehensive data analysis, suggesting moderate implementation depth.
Research Flowchart
flowchart TD
A["Research Goal: High-speed stock trading<br>via NP-hard optimization"] --> B["Key Methodology: Ising Machine<br>using Simulated Bifurcation"]
B --> C["Data Inputs: 128 Tokyo Stock Exchange stocks"]
C --> D["Computational Process:<br>NP-hard Portfolio Optimization<br>Max returns & Min correlation"]
D --> E{"Outcome: FPGA Trading System<br>Response Latency: 164 μs"}
E --> F["Result: Real-time Long/Short<br>Trading Execution"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a real-time trading strategy that selects potentially profitable, uncorrelated, and balanced stocks via NP‑hard combinatorial optimization using an FPGA‑based embedded Ising machine.
Source sentence
Here we propose a trading strategy based on selections of potentially profitable, uncorrelated, and balanced stocks by NP-hard combinatorial optimization and show through real-time trading that the strategy is executable with an automated real-time system using an FPGA-based embedded Ising machine for the discrete sele
Assumes 6
- Deviations of stock prices from the VWAP are treated as expected returns. market
Source sentence
Considering the deviations of stock prices from the VWAP as expected returns, the strategy takes long positions of the underperforming stocks and short positions of the outperforming stocks...
- The group of positions is intended to be market‑neutral (delta‑neutral) and correlation‑diversified. model
Source sentence
it simultaneously holds multiple positions selected through a discrete portfolio optimization problem making the group of positions being market‑neutral
- The latency of the system (164 µs) is sufficiently low relative to the lifetime of trading opportunities. execution
Source sentence
the trading strategy based on NP-hard portfolio optimization for N=128 is executable with the
- Orders issued are not necessarily filled at the ask/bid prices used for decision‑making. execution
Source sentence
Because of the latency of a system that executes the strategy and the activities of other trading entities, the orders issued are not necessarily filled at the ask/bid prices used for the decision‑making.
Show 2 more
- The market feed provides a full price list of ask and bid for the N‑stock universe at each tick. data
Source sentence
The proposed strategy determines the opening of positions based on an instantaneous market situation (a price list of ask and bid for the N‑stock universe).
- The selected 128 stocks are high‑liquidity constituents of the Nikkei 225 or TOPIX 100. data
Source sentence
The 128 stocks were selected from the Nikkei 225 or TOPIX 100 constituents in terms of high liquidity.
Data
- TSE historical market feeds order-book proprietary
Source sentence
that ... processes the historical market feeds provided by the TSE and emulates the internal state of the trading system.
- Real‑time TSE market feed order-book proprietary
Source sentence
The real trade data is from Feb. 1, 2023, to Mar. 31, 2023, being adjusted with the simulation at the first day.
Costs
Modeled
Finds 5
- The strategy achieved a Sharpe ratio of 1.23 over ~3 years, with an annualized return of 3.6 % and risk (SD) of 2.9 % for an investment of 16 million JPY. performance
Source sentence
The Sharpe ratio of the strategy over the simulation period (approximately 3 years) is 1.23, where the annualized return and risk (the standard deviation of the return) are 3.6 % and 2.9 %, respectively, for an investment of 16 million JPY...
- Cumulative transaction amount in real‑time trading (118,956,828 JPY) matched the simulation (118,948,300 JPY) within 0.01 %, confirming executability. performance
Source sentence
... is coincident well (+0.01 %) with the simulation value (118,948,300 JPY), indicating that the strategy proposed is executable with the trading system with a latency of 164 µs.
- System‑wide latency from market feed arrival to order packet issuance was 164 µs for a 128‑stock universe. performance
Source sentence
The system‑wide latency from the market feed arrival to the order packet issuance is 164 µs for a 128‑stock universe.
- Data transfer to the SBM core was reduced from O(N²) to O(N) by separating tick‑by‑tick and day‑by‑day components, improving latency. theoretical
Source sentence
we reduced the data size transferred to the SBM core from O(N 2) to O(N) when the market situation changes and improved the system latency.
- Real‑time trading over 42 business days (252 hours) showed good agreement with back‑cast simulation, confirming the strategy’s practical feasibility. performance
Source sentence
Based on the good agreement in the cumulative transaction amounts and detailed comparison analysis of transactions between the experiment and simulation, we have concluded that the response latency of the system with the SB‑based Ising machine is sufficiently low to execute the trading strategy...
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 25 related papers). Coverage is partial, so this is neither replication nor refutation.
2 papers in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.