Exploring asset managers services · Return to your project overview
A systematic macro trader engaged Marsbridge to build a cross-asset portfolio optimizer that allocates across global futures and adjusted equities while accounting for real-world frictions (contract granularity, partial data overlaps, transaction costs). We stabilized the optimization via covariance construction fixes, regularization penalties, and integer-aware deployment paths.
Mandate: deliver a robust, auditable optimizer that remains stable under small data perturbations, supports integer lot sizes for futures, and reflects realistic transaction costs—built on a dependable data foundation.
Standard MVO can be unstable with small input changes and integer constraints yielding materially different allocations. Early covariance construction did not ensure positive-semi-definite properties, affecting solver reliability.
A compact Risk & Research squad—Quant Lead, Optimization/Risk Engineer, Data Engineer, Execution Analyst—(1) hardened data, (2) stabilized optimization, and (3) delivered practical integer-aware workflows with clear runbooks.
Built internal continuous contract series aligned to liquid activity and roll schedules. Implemented corporate-action adjustments with cross-checks from multiple public data sources. Standardized to a common bar frequency and introduced correlation alignment for non-overlapping trading hours; removed persistently illiquid instruments after QA.
Rebuilt covariance inputs to ensure PSD properties; aligned the risk term to the intended volatility definition. Added concentration and transaction-cost penalties to reduce knife-edge portfolios and promote allocation stability. Evaluated mixed-integer routes and, where appropriate, used continuous solve + controlled rounding with trade-gap thresholds for realism.
Supported auto-select most-active versus fixed-expiry configurations; exposure checks aggregate across expiries. Introduced a simple config + database layer for adding/removing instruments and backfilling corporate actions.
Python, NumPy, Pandas
Mixed-integer solver interfaces, Covariance regularization
Broker/data APIs, In-house continuous-series builder, Corporate-action fields with web cross-checks
Server environment for research and testing
Kickoff & data plan—continuous series + adjusted equities; handle missing overlaps; define common resampling. Covariance & solver hygiene—ensure PSD; align objective to volatility; add stabilizing penalties. Integer path & fallbacks—trial mixed-integer routes; add thresholded rounding if required. Config & roll controls—config/database layer; auto most-active vs specific expiry; expiry aggregation. Review & sign-off—parity checks; validate stability under small input perturbations.
Stabilized allocations via PSD covariance, transaction-cost and concentration penalties, plus an integer-aware deployment path. Data realism through in-house continuous series and adjusted equities with common-horizon panels and cross-market correlation alignment. Operational simplicity via config-driven onboarding and exposure checks aligned with real-world trading practice.
Fighting unstable MVO outputs? We harden your data, repair covariance, and regularize the objective—so allocations survive the real world.
Drop us a line! We are here to answer your questions within 1 business day.
Once we’ve received and processed your request, we’ll get back to you to detail your project needs and generally sign an NDA to ensure confidentiality.
After examining your project requirements, our team will devise a proposal with the scope of work, team size, time, and cost estimates.
We’ll arrange a meeting with you to discuss the offer and nail down the details.
Finally, we’ll sign a contract and start working on your project with agreed timeline
