HedgeSPA Team

Home
Singapore
Singapore
HedgeSPA Team

Call Directions Website

HedgeSPA We are not just about predicting market trends or econometrics consulting.

Established in Singapore since 2009 and fast gaining global recognition, HedgeSPA (Hedge Funds and Sophisticated Products Advisors) is a provider of state-of-the-art investment analytics on a “front-to-back” platform to leading banks, investment managers and hedge funds worldwide. We are recognised as a commercial R&D entity, and are co-funded by the National Research Foundation and SPRING with upfront cash and in-kind commitments as well as potential equity financing in the future. We are a team comprised of seasoned industry professionals with rich experience from leading investment banks and asset managers, as well as one of the finest teams of quantitative PhD students and post-docs, who are passionate about using our award-winning analytics to help our client better understand how various financial market and economic scenarios can impact their portfolios. Our comments on financial markets and the global economy are regularly covered by Straits Times, Business Times, Channel News Asia, Asian Banker as well as other global and trade media outlets. We combine the best of both the open source and hosted application paradigms, at affordable prices with clients being charged based on actual usage only. With full connectivity support, we aim to eliminate the need for our clients to test their interfaces with fragmented exchanges across the region (especially in Asia). Our technical implementation is truly state-of-the-art, with realistic speed-up for on-line delivery to any users in orders of magnitude better than similar implementations at even the largest institutional investors.

25/10/2026

Thrilled to have been invited to submit the companion paper to the AGI discussion — with a full fee waiver.

The AGI transition can create real economic disruption. The big question: can a solution be used to better manage the transition?

16/10/2026

New invited submission! It is a genuine honour to receive the following from a top-quartile journal in mathematical physics: "Our Guest Editor reviewed your manuscript and thought it is suitable for our Special Issue."

A brief remark in passing: we do not suggest that a laboratory-scale version of AGI could not emerge sooner. We suggest only that, on the basis of detailed scientific estimates regarding computational, energetic, and fidelity constraints, the practically deployable versions are likely to be a decade or more away.

06/10/2026

Always a pleasure to meet our KPI. We've been seeing a lot of each other lately. Then there's the original DCRNN paper from Caltech with 7,000+ citations. We're still catching up. Respect where it's due!

27/09/2026

Excited to share that we've submitted our manuscript, "Quantum-Assisted Recommender Algorithms for Financial Asset Recommendation: Theoretical Bounds and Empirical Validation," to a Springer AI journal with a CiteScore of 11.5 — and we're honored to have been invited to submit with a full APC waiver. Equally grateful to be working alongside such distinguished colleagues from Cambridge University Computing Lab and Imperial College AI & Data Analytics Lab. More importantly, this work answers a question that matters: How does an AI model work? When does such an AI model expected to fail — and why? In finance, where fiduciary duty and regulatory suitability govern every decision, it's not enough to report high accuracy. Our unified framework makes the entire pipeline explainable, auditable, and predictable — connecting QAOA approximation ratios, spectral compression error, and rounding error directly to classification accuracy, and providing concrete sample-size guidance for compliance and deployment. The results speak for themselves: 97.33% in-sample and 94.67% out-of-sample accuracy, with theoretical predictions that closely match empirical outcomes. Proud to help bridge algorithmic theory and responsible deployment of quantum-assisted AI in high-stakes environments. Preprint coming soon on Research Square — stay tuned!

21/09/2026

Excited to share our latest research published in Electronics (CiteScore 7.0; Impact Factor 2.9)! The joint paper with Prof. Constantinides, "From Bellman to Real-Time: Extensions to Complex Weather Regimes, Physics-Informed Optimization, and Full-Scale Validation," is now officially available online as part of the Special Issue on Trustworthy and Data-Driven Intelligent Information Systems.
What this research achieves:
Too often, advanced AI models operate as "black boxes." In this work, we bridge the gap between high-level theory and real-world deployment through a methodical, multi-stage pipeline:
1. Rigorous Foundations: We establish a theoretical proof demonstrating why our AI engine achieves an empirical accuracy exceeding 98%, grounding the model in Lipschitz dynamics, Graph diffusion mixing, Bounded noise, Dynamical contraction, and the GRU echo state property.
2. Empirical Stress-Testing: We evaluate these foundational assumptions against empirical data across 24 distinct weather locations using EnergyPlus Weather datasets.
3. Failure Mode Analysis: To ensure real-world reliability, we conduct granular anomaly mapping to determine how the AI handles edge cases, simulating single sensor failure, single sensor degradation, and all-sensor degradation.
4. The Deployment Pipeline: This research details our structured migration pathway, moving from standard simulations to computational fluid dynamics-informed simulations, then to hardware-in-the-loop (HIL) testing, and finally to Front-End Engineering Design (FEED).
Looking Forward:
The success of this physics-informed framework provides a repeatable architecture that can be scaled well beyond stochastic control under weather regimes, offering reliable blueprints for application in other complex fields such as quantitative finance and medical diagnostics.
On a final note, our team is keeping the momentum going with two more manuscripts underway, including an invited paper slated for a Springer journal (CiteScore 11.5). More updates on that soon!
Access the open-source publication here:
Article Website: https://www.mdpi.com/2079-9292/15/18/4341
Full PDF: https://www.mdpi.com/2079-9292/15/18/4341/pdf

18/09/2026

Our SECOND paper on top-quartile journal Electronics has been ACCEPTED! Let me explain why this one hits different: 1) It includes a highly mathematical 4-page proof on why the AI engine can effectively learn from complex decisions. This isn't just theory — it's the FOUNDATION of so much of what we do. 2) The editors? Straight from the Fraunhofer Institute — Germany's TOP applied R&D institution. TBH... the review process was TOUGH. But that's exactly what makes this a major seal of approval from the best in Germany!

23/08/2026

Big milestone: Submitting to top Springer-Nature journal in mathematical finance. Built upon decades of work with three GOATs:
• Sandy Grossman — Clark Medalist, co-wrote the information-efficient markets bible with Stiglitz, founded QFS (RenTec/DE Shaw tier). I coded up his 1990 Econometrica simulations.
• Nicos Christofides — my PhD advisor and literal father of graph theory. His TSP algorithm has remained undefeated for over 40 years.
• Tony Constantinides — company science advisor, invented LPNN, co-created JPEG, and was the most-cited IEEE author back-to-back.
The outcome? A unified pipeline that does it all: 1. Info-theoretic profitability bounds 2. TSP combinatorial optimization for bond selection 3. Sub-second LPNN inference 4. Real-time regime detection. All wrapped in 40 pages of bound proofs and empirical validation. Alpha signals? IP-protected. Preprints available to institutional fixed-income investors.
Why this won't become LTCM 2.0:
1. We partner with investors holding existing bond inventories — no liquidity squeezes.
2. Information theory tells you exactly what's needed to hit performance targets, not a game of pure leverage.
3. It runs on fully automated AI infrastructure that does not hallucinate; far more likely to pass compliance.
The technique will also be extended to other FICC products. Full suite live demo at FTAHK, September. DMs open for preprints.

15/07/2026

Excited to announce the first in a two‑paper series now published in Electronics (CiteScore 7.0, firmly top quartile). We present a framework that renders a traditionally intractable stochastic control problem computationally feasible in real time—achieving sub‑second inference latencies (predominantly below 1 s) while requiring only a sparse network‑graph description (adjacency matrix) of the underlying process as input. The framework learns autonomously, attaining 98.4% in‑sample accuracy with practically no out‑of‑sample degradation. Beyond renewable generation, this advance enables safety‑critical and "soft‑touch" applications where operational integrity and safety certification are essential.
Paper: https://www.mdpi.com/2079-9292/15/14/3119
PDF: https://www.mdpi.com/2079-9292/15/14/3119/pdf

07/07/2026

Short Research Note: Japan vs. Korea – A Cyclical vs. Structural Bet

01/07/2026

New SG address @ Raffles Place

Address

12 Woodlands Square, #05-70, Woods Square, Queenstown Estate
Singapore
737715

Telephone

+6591831492

Website

https://www.hedgespa.com/

Alerts

Be the first to know and let us send you an email when HedgeSPA Team posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Want your business to be the top-listed finance company in Singapore?