Paper: arXiv 2405.01598
Authors: Emily Tallman, Mike West
Abstract
We discuss and develop Bayesian dynamic modelling and predictive decision synthesis for portfolio analysis. The context involves model uncertainty with a set of candidate models for financial time series with main foci in sequential learning, forecasting, and recursive decisions for portfolio reinvestments. The foundational perspective of Bayesian predictive decision synthesis (BPDS) defines novel, operational analysis and resulting predictive and decision outcomes. A detailed case study of BPDS in financial forecasting of international exchange rate time series and portfolio rebalancing, with resulting BPDS-based decision outcomes compared to traditional Bayesian analysis, exemplifies and highlights the practical advances achievable under the expanded, subjective Bayesian approach that BPDS defines.
Complexity vs Empirical Score
- Math Complexity: 8.0/10
- Empirical Rigor: 4.0/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper is highly theoretical, featuring advanced Bayesian statistics, entropy tilting, and multi-dimensional utility scoring functions, which significantly increases mathematical complexity. While it includes a case study on FX portfolio rebalancing, the text lacks explicit code, datasets, or detailed backtesting results, focusing instead on methodological framework and conceptual advances, resulting in lower empirical rigor.
Research Flowchart
flowchart TD A["Research Goal<br>Address model uncertainty in<br>portfolio decision synthesis"] --> B["Methodology: BPDS Framework<br>Bayesian Predictive Decision Synthesis"] B --> C["Data Inputs<br>International Exchange Rate Time Series"] C --> D["Computational Process<br>Sequential Learning &<br>Recursive Portfolio Rebalancing"] D --> E["Key Outcome 1<br>Operational analysis for<br>Bayesian model uncertainty"] D --> F["Key Outcome 2<br>Enhanced predictive<br>decision outcomes"] D --> G["Key Outcome 3<br>Superior portfolio performance<br>vs traditional Bayesian methods"] E --> H["Conclusion<br>BPDS provides practical advances<br>in financial forecasting"] F --> H G --> H
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 Bayesian Predictive Decision Synthesis (BPDS), a fully Bayesian framework that combines model forecasts and decision outcomes for portfolio analysis.
Source sentence
Bayesian Predictive Decision Synthesis (BPDS), introduced in Tallman and West (2023), closes these gaps. BPDS is a fully Bayesian, theoretically founded framework to evaluate, compare, and combine sets of candidate models based on both their anticipated and historical decision outcomes as well as pure predictive perfor
Assumes 5
- Returns are approximated as normally distributed for score uncertainty analysis. statistical
Source sentence
With daily returns on FX and stock indices, f and q1/2 are typically of the order of 0.01 or less; this is relevant in exploring score uncertainty via the variance matrix Vp = Vp[s(y, x)]. Were p(r) to be normal, then it is easy to show that Vp(−e2/2) = (f 2 + q/2)Vp(e).
- Initial model probabilities are derived from traditional Bayesian model averaging (BMA). model
Source sentence
Initial model probabilities πtj = Pr(Mj|Dt−1), such as from traditional Bayesian model averaging (BMA) based on prior data and information, and/or from historical performance in goal-focused forecasting...
- The tilting vector is constrained so that the risk‑reward ratio dt = τ1/τ2 is less than the target return r*. execution
Source sentence
We additionally restrict the tilting vector such that dt < r∗ t so that the portfolio target can be at most doubled; this is in order to obviate large values earlier seen in Figure 3, and is enforced in the computations to evaluate τ t at each time point.
- The data are split into training, model‑selection, and hold‑out periods. data
Source sentence
The dataset is divided into 3 time periods. The period to the beginning of 2015 is used to fit a large set of models, the next 4 years (2015-2018 inclusive) is used to select a subset of “better-performing models”, and the final 3 years (2019-2021 inclusive) defines a hold-out data test period.
Show 1 more
- Portfolio decisions are based on one‑step ahead forecasts unless otherwise specified. execution
Source sentence
BMA focuses wholly on 1-step ahead predictive accuracy, so the initial probabilities used here in BPDS generally favour models based on predictive performance at this short-term horizon.
Data
- Daily FX returns for 9 currencies market-prices public
Source sentence
The data set includes daily returns for q = 9 currencies beginning in January 2001 and ending in December 2021.
Costs
Not modeled
Finds 3
- BPDS achieves superior cumulative returns and Sharpe ratios compared to BMA and improved BMA across all settings of the score‑improvement parameter. performance
Source sentence
BPDS is able to achieve superior cumulative returns and Sharpe ratios compared to BMA and “improved BMA”, across all settings of mt.
- Five‑day horizon portfolios show weaker improvements than one‑day portfolios, likely due to using initial BMA‑based weights that favor short‑term predictive accuracy. performance
Source sentence
The 5-day portfolios improve over BMA comparisons but generally see relatively lack‑luster results compared to the 1-day portfolios. This is mainly due to the use of initial BMA-based weights.
- BPDS results are close to those of a “BMA 2” benchmark and do not consistently outperform it in the first set of comparisons. performance
Source sentence
BPDS outcomes are generally close to those of BMA 2, but do not empirically improve upon the latter in this first set of comparisons.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 24 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper 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.
Cited References (103)
- De F inetti scoops M arkowitz (2006)
- Bruno de F inetti and mean-variance portfolio selection (2006)
- The problem of \lq full-risk insurances\rq (1940)
- Forecasting with B ayesian Vector Autoregression (2013)
- Forecast combination and model averaging using predictive measures (2007)
- Finding good predictors for inflation: A B ayesian model averaging approach (2004)
- Bayesian forecast combination for VAR models (2008)
- Model averaging and variable selection in VAR models (2012)
and 95 more
- Bayesian P redictive D ecision S ynthesis: M ethodology and A pplications (2024)
- Bayesian predictive decision synthesis (2023)
- Perspectives on constrained forecasting (2023) — link
- Perspectives on constrained forecasting: F urther details and examples (2023) — link
- Decision synthesis in monetary policy (2024)
- On entropic tilting and predictive conditioning (2022)
- Bayesian forecasting of multivariate time series: Scalability, structure uncertainty and decisions (with discussion) (2020)
- The evolution of forecast density combinations in economics (2019) — [link](https://doi.org/doi: 10.1093/acrefore/9780190625979.013.381)
- Bayesian analysis of latent threshold dynamic models (2013)
- Generalized density forecast combinations (2015)
- Bayesian Theory (1994)
- Model uncertainty (2004)
- Bayesian model averaging in the $\mathcal M -$open framework (2013)
- A general framework for updating belief distributions (2016)
- Time-varying combinations of predictive densities using nonlinear filtering (2013)
- Prediction using several macroeconomic models (2017)
- Gibbs posterior for variable selection in high-dimensional classification and data mining (2008)
- From $\epsilon$-entropy to KL- entropy: A nalysis of minimum information complexity density estimation (2006)
- Information theoretic upper and lower bounds for statistical estimation (2006)
- Economic predictions with big data: T he illusion of sparsity (2021)
- Time Series: Modeling, Computation and Inference (2021)
- Bayesian Forecasting and Dynamic Models (1989)
- Bayesian Forecasting and Dynamic Models (1997)
- Modelling of probabilistic agent opinion (1992)
- Modelling agent forecast distributions (1992)
- Is our view of B ayesian statistics too narrow? (1992)
- Forecasting using relative entropy (2005) — link
- Using entropic tilting to combine BVAR forecasts with external nowcasts (2017) — link
- Mixed-frequency B ayesian predictive synthesis for economic nowcasting (2021)
- Dynamic B ayesian predictive synthesis in time series forecasting (2019)
- Bayesian predictive synthesis– Discussion of: U sing stacking to average B ayesian predictive distributions, by Y .~ Y ao \em et al (2018)
- Multivariate B ayesian predictive synthesis in macroeconomic forecasting (2020)
- Bayesian predictive synthesis with outcome-dependent pools (2023) — link
- Bayesian emulation for multi-step optimization in decision problems (2019)
- Focused B ayesian prediction (2021)
- Adaptive variable selection for sequential prediction in multivariate dynamic models (2021) — link
- Bayesian forecasting and scalable multivariate volatility analysis using simultaneous graphical dynamic linear models (2017)
- GPU -accelerated B ayesian learning and forecasting in simultaneous graphical dynamic linear models (2016)
- B ayesian model averaging: A tutorial (with discussion) (1999)
- An analysis of international exchange rates using multivariate DLMs (1987)
- Building a Framework for Predictive Science (2012) — link
- Bayesian dynamic factor models and portfolio allocation (2000)
- Dynamic dependence networks: F inancial time series forecasting & portfolio decisions (with discussion) (2016)
- Bayesian model monitoring (1986)
- Bayesian nonparametric calibration and combination of predictive distributions (2018)
- Using B ayesian model averaging to calibrate forecast ensembles (2005)
- Strictly proper scoring rules, prediction, and estimation (2007)
- Quantifying time-varying forecast uncertainty and risk for the real price of oil (2023)
- Statistical postprocessing for weather forecasts: R eview, challenges, and avenues in a big data world (2021)
- Optimal prediction pools (2011)
- Prediction with misspecified models (2012)
- Modelling expert judgements for B ayesian updating (1985)
- On the reconciliation of probability assessments (1979)
- Improving predictive distributions (1980)
- Approximate models and robust decisions (2016)
- A loss discounting framework for model averaging and selection in time series models (2024) — link
- Information and Exponential Families: In Statistical Theory (1978)
- Ensemble based systems in decision making (2006)
- Eliciting prior information to enhance the predictive performance of Bayesian graphical models (1995)
- Bagging predictors (1996)
- The strength of weak learnability (1990)
- Stacked generalization (1992)
- Using stacking to average Bayesian predictive distributions (with discussion) (2018)
- A Bayesian boosting model (2012) — link
- Why does bagging work? A Bayesian account and its implications (1997)
- Bagging and the Bayesian bootstrap (2001)
- Random forests (2001)
- BART: Bayesian additive regression trees (2010)
- Bayesian perspectives on combining models (2012)
- Combining density forecasts (2007)
- Combining large numbers of density predictions with Bayesian predictive synthesis (2023)
- Predictive density combination using a tree-based synthesis function (2023)
- Combining density forecasts using focused scoring rules (2017)
- A flexible predictive density combination for large financial data sets in regular and crisis periods (2023)
- Optimal portfolio choice under decision-based model combinations (2016)
- Forecast density combinations of dynamic models and data driven portfolio strategies (2019)
- Model uncertainty and forecast combination in high-dimensional multivariate volatility prediction (2015)
- Forecast accuracy and economic gains from Bayesian model averaging using time-varying weights (2010)
- Dynamic prediction pools: An investigation of financial frictions and forecasting performance (2016)
- Turning Bayesian model averaging into Bayesian model combination (2011)
- Clustering and predicting multiple time series count data via mixture of Bayesian predictive syntheses: Analysis of COVID-19 hospitalisation in Japan and Korea
- Structural scenario analysis with SVAR s (2021)
- Identification of financial factors in economic fluctuations (2019)
- Measuring the macroeconomic impact of monetary policy at the zero lower bound (2016)
- Asymmetric conjugate priors for large B ayesian VAR s (2022)
- Bayesian forecasting (1976)
- An averaging framework for minimum-variance portfolios: O ptimal rules for combining portfolio weights (2024)
- Optimal Portfolio Choice with Parameter Uncertainty (2007)
- Optimal Versus Naive Diversification: How Inefficient is the $1/N$ Portfolio Strategy? (2009)
- Markowitz meets T almud: A combination of sophisticated and naive diversification strategies (2011)
- A Comparison of Ensemble Methods in Financial Market Prediction (2012)
- Forecasting realized volatility in a changing world: A dynamic model averaging approach (2016)
- Forecasting the price of gold using dynamic model averaging (2015)
- Forecast combinations: An over 50-year review (2023)
- Portfolio Selection (1952)
0 of 103 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.