Currently, the bond market remains heavily reliant on exchanges of
information between counterparties. Information on prices is decentralized,
and consequently, traders operate with different limitations in data access.
Some bonds may also be illiquid and/or trading infrequently, so accurate prices
are not observed at sufficient frequency, which creates an informational deficit.
“At a time when the entire market is struggling to determine a fair
value price for bonds in real-time, Overbond’s R&D team has delivered
an AI model that can predict prices in the future,” says Vuk Magdelinic, CEO of Overbond.
“Overbond achieved this by developing an altered mathematical formula which introduced a
net new factor to achieve credit curve convexity necessary for longer tenor bond optimizations.”
R&D project snapshot
The objective of the research project was to study the theory,
performance, and practical implementation of AI models for predicting
bond prices and yield curves in the future. The team focused on three
main areas of research that determine the dynamics of the yield curve
for government and corporate bond issuers. These were:
- Yield curve dynamics and price momentum
- Yield curve shape and nonlinear methods and
- Liquidity and information effects
Within each of the themes, a number of sub-projects were identified
which broadly fell into another three categories. These were:
- Feature engineering:
A critical task prior to the construction of
a sound predictive model was the design of sound features, such as covariates, encompassing material
predictive power to determine the shape of yield curves. This entailed determining the effects needed
to be captured by the various explanatory variables. These included momentum, market volatility, liquidity,
equity market signals, correlations between various segments of the curve, and macroeconomic situation.
- Artificial intelligence learning methods:
This category used a set of covariates to construct
a performance predictive machine learning algorithm to forecast yield curves.
Various statistical and AI learning methods such as linear models, random forests,
neural networks, support vector machines were then tested.
- Time series dynamics assessment:
This important last step involved modeling the time evolution of the covariates. Understanding this
evolution was crucial since the goal was to produce one to five-day forecasts
of yield curves. Time series models were applied to make the forecasts about
the value of covariates in the near future. In turn, these were used to perform
bond price forecasting using predictive analytics algorithms.
The significance of Overbond R&D Labs
Overbond R&D Labs is a member of the Institute for Data Valorization (IVADO)
in the Montreal MILA cluster. IVADO is a Quebec-wide collaborative institute
specializing in digital intelligence, dedicated to transforming new scientific
discoveries into concrete applications that benefit society. Located in the heart
of Quebec's AI ecosystem, Mila is a community of 450 researchers specializing
in machine learning and dedicated to scientific excellence and innovation.
With the support of IVADO and in partnership with HEC Montreal, Concordia University,
and Queen’s University, Overbond R&D Labs sources one of the deepest vertical data stacks
in the world for fixed income capital markets AI modeling, including traditional and
alternative data sets. Overbond R&D Labs have a capacity to onboard more than 40 masters,
doctoral and post-doctoral fellows from the three participating Canadian universities.
These labs focus on advanced AI applications for bond pricing, price momentum, liquidity scoring,
fundamental analysis, and modeling the shape of yield curves.
Overbond provides advanced models and AI capabilities to help reduce
development and research costs for its clients. All advanced algorithmic
outputs such as bond price momentum recommendations, pre-trade ideas,
pricing tension signals and buyer preference monitoring are made available
through Overbond’s cloud-accessible platform. Its R&D Labs also offer a rapid
prototyping environment as well as data cleansing and a normalization layer so
clients can drive meaningful AI modeling results without lengthy data sourcing
and structuring IT integration projects.
With offices in Toronto, Montreal, and New York, Overbond is an artificial intelligence (AI)
quantitative analytics provider for institutional fixed income capital markets. We provide data
aggregation solutions and a comprehensive suite of AI algorithms (known as COBI) for bond pricing,
liquidity risk, auto-execution and hedging, pre-trade signals, and market surveillance. Founded in 2015,
Overbond is transforming how global investment banks, institutional investors, corporations, and governments
connect and access bond markets’ AI data analytics.
Our fully-digital platform and suite of AI models utilizes the world’s most advanced algorithms
in finance. We produce market analytics with AI pricing and trade-tiering optimizations that
enable full trade automation. Our tools and processes facilitate faster, more precise trade
execution, eliminate information flow inefficiencies, reduce execution costs, and minimize
compliance risks. In addition, the Overbond platform assists market participants with
systematic price discovery and liquidity risk management, enabling sell-side traders
to execute more profitable trades in larger volumes and buy-side traders to achieve pre-trade
best execution and counterparty selection.
Our global client base comprises buy-side institutions with over $2 trillion of
combined AUM and corporate and government issuers with more than $20 billion
in bonds outstanding. These clients include the European Central Bank, the
European Stability Mechanism (ESM), Wells Fargo, Fidelity, ATB Financial, Hydro One,
and EPCOR. In addition to its three offices, Overbond has R&D labs with three leading
Canadian universities and the Institute for Data Valorization (IVADO).