{"paper_id":"2511.21850","title":"Black-Litterman and ESG Portfolio Optimization","authors":["Aviv Alpern","Svetlozar Rachev"],"paper_date":"2025-11-26","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2511.21850v1","page_url":"https://thequant.space/flowcharts/black-litterman-and-esg-portfolio-optimization/","scores":{"empirical_rigor":6,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a simple portfolio optimization strategy that incorporates ESG scores as a bias within the Black‑Litterman framework, using Stein shrinkage of equilibrium risk premiums.","quote":"We introduce a simple portfolio optimization strategy using ESG data with the Black-Litterman allocation framework. ESG scores are used as a bias for Stein shrinkage estimation of equilibrium risk premiums used in assigning Black-Litterman asset weights.","summary":"We introduce a simple portfolio optimization strategy using ESG data with the Black-Litterman allocation framework. ESG scores are used as a bias for Stein shrinkage estimation of equilibrium risk premiums used in assigning Black-Litterman asset weights. Assets are modeled as multivariate affine…","strategy":"Black-Litterman ESG portfolio with CVaR"},"assumptions":{"status":"available","items":[{"assumption":"Asset returns follow a multivariate affine normal‑inverse Gaussian distribution.","type":"model","stated":"explicit","quote":"Assets are modeled as multivariate affine normal-inverse Gaussian variables using CVaR as a risk measure."},{"assumption":"The prior for the mean vector is Gaussian with mean Π and covariance τΣ, derived from CAPM equilibrium.","type":"statistical","stated":"explicit","quote":"µ ∼ N(Π, τΣ). Here, the extra parameters Π and τ come from the Capital Asset Pricing Model (CAPM)."},{"assumption":"Equilibrium portfolio weights ω_eq are the relative market capitalizations of the components.","type":"market","stated":"explicit","quote":"the equilibrium portfolio weights ωeq are the relative market capitalization of the portfolio components."},{"assumption":"A soft turnover constraint is imposed via a penalty term ρ in the objective.","type":"execution","stated":"explicit","quote":"The last term on the right hand side is a soft turnover constraint, ρ, denoting the turnover penalty and ||·||1 being the L1 vector norm."},{"assumption":"ESG scores are yearly and used as a bias; missing ESG data leads to exclusion of assets for those years.","type":"data","stated":"explicit","quote":"The ESG data is yearly quotes from Robeco, released on the last trading day in December each year over the period from 2016 to 2020. The data set lacks ESG scores for Walt Disney (DIS) before 2019. Consequently DIS is not included in the portfolio for years without ESG data."},{"assumption":"The risk‑aversion parameter α ∈ [0,1] balances expected return against CVaR.","type":"model","stated":"explicit","quote":"The risk aversion parameter, α ∈[0, 1], is used to adjust the weighting of CVaR relative the the expected return."}]},"datasets":{"status":"available","items":[{"name":"Daily Bloomberg data of the Dow Jones 30","provider":"Bloomberg","kind":"market-prices","access":"commercial","period":"Jan 3, 2013 to Sep 3, 2021","universe":"Dow Jones 30 constituents (29 used)","quote":"For our analysis we use daily Bloomberg data of the Dow Jones 30 between Jan 3, 2013 and Sep 3, 2021. However, of those 30, we only had data for 29.","known_id":"bloomberg"},{"name":"Robeco yearly ESG scores","provider":"Robeco","kind":"alternative","access":"commercial","period":"2016 to 2020","universe":"ESG scores for Dow Jones constituents","quote":"The ESG data is yearly quotes from Robeco, released on the last trading day in December each year over the period from 2016 to 2020."}],"known":[{"id":"bloomberg","name":"Bloomberg","provider":"Bloomberg L.P.","kind":"market-prices","access":"commercial","quote":"For our analysis we use daily Bloomberg data of the Dow Jones 30 between Jan 3, 2013 and Sep 3, 2021.","mentions":1,"used_in":1}],"sample":{"start":"2017-01","end":null,"market":"Dow Jones 30","asset_class":"equities","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"unclear","headline_net_of_costs":false,"detail":null,"quote":null,"source":"conflict","rule":"R4b","candidates":{"front_matter":"no","evidence_card":"partial","evidence_card_quoted":true},"conflict_kind":"graded","card_detail":"A soft turnover penalty ρ (e.g., 5×10⁻⁴ to 40×10⁻⁴) is added to the objective to discourage excessive trading.","card_quote":"The last term on the right hand side is a soft turnover constraint, ρ, denoting the turnover penalty and ||·||1 being the L1 vector norm.","meaning":"the platform classifier and the paper-text extraction disagree, so no value is asserted"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"The best Black‑Litterman ESG strategies achieved annual returns of approximately 40‑45% with risk‑adjusted returns equal to or greater than the benchmark.","kind":"performance","quote":"Of the BL strategies that had acceptable turnover, the largest portfolios returns were between 40-45% annually. These portfolios had high RRRs equal to or greater than the benchmark."},{"finding":"Standard mean‑CVaR portfolios with high ESG weights and low turnover delivered lower annual returns (15‑20%) and lower reward‑to‑risk ratios.","kind":"performance","quote":"Portfolios with {λ = 0, α ≤0.5} and {λ = 0.25, α ≤0.4} have the lowest drawdowns of all employed strategies. However, the comparatively meager annual returns of 15-20% result in low RRR."},{"finding":"BL portfolios with the same turnover constraint often incurred excessive turnover, making them infeasible, whereas increasing the turnover penalty reduced turnover but also lowered returns.","kind":"performance","quote":"a majority of CVaR95 BL portfolios with the same constraint had excessive turnovers that made the strategies infeasible."},{"finding":"The benchmark (adjusted DJ index) returned over 31% annually, outperforming most ESG‑adjusted strategies except the top BL configurations.","kind":"performance","quote":"the benchmark returned over 31% annually and had higher RRR."},{"finding":"CVaR99 optimization performed worse than CVaR95, with lower returns and similar risk metrics, and most acceptable‑turnover strategies still lagged the benchmark.","kind":"performance","quote":"The performance compared with CVar95 was subpar. Strategies that had acceptable turnover rarely approach the benchmark."}],"reported_performance":{"strategy":"Black-Litterman ESG portfolio with CVaR","sharpe":null,"sharpe_benchmark":null,"annual_return_pct":42.5,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"available","supports":[{"paper":{"paper_id":"2501.12841","title":"Optimal vs. Naive Diversification in the Cryptocurrencies Market: The Role of Time-Varying Moments and Transaction Costs","paper_date":"2025-01-22","authors":["Heming Chen","Xiaojing Cai"],"empirical_rigor":7.5,"page_url":"https://thequant.space/flowcharts/optimal-vs.-naive-diversification-in-the-cryptocurrencies-ma/"},"question":"How does applying a turnover penalty impact portfolio turnover, transaction costs, and performance?","this_paper_finding":"BL portfolios with the same turnover constraint often incurred excessive turnover, making them infeasible, whereas increasing the turnover penalty reduced turnover but also lowered returns.","other_paper_finding":"The turnover penalty reduces transaction costs across all strategies but does not consistently improve overall performance.","other_paper_quote":"our results show that it effectively reduces transaction costs across all optimal strategies but does not necessarily enhance the final performance of every strategy.","note":"Both findings agree the turnover penalty reduces turnover/costs, but A finds it does not consistently boost performance while B shows it also lowers returns.","confidence":0.509}],"contradicts":[],"qualifies":[],"compared_with":49,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":1,"extracted_at":"2026-11-14T17:21:03Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}