{"paper_id":"2512.10606","title":"Local and Global Balance in Financial Correlation Networks: an Application to Investment Decisions","authors":["Paolo Bartesaghi","Rosanna Grassi","Pierpaolo Uberti"],"paper_date":"2025-12-11","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2512.10606v1","page_url":"https://thequant.space/flowcharts/local-and-global-balance-in-financial-correlation-networks/","scores":{"empirical_rigor":6,"math_complexity":7.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes an investment strategy that selects assets whose local balance deviates from a high global balance, concentrating allocations in those few assets.","quote":"The main idea proposed in this paper is to study deviations of local balance from global balance in a signed network to identify distinctive behaviors of specific nodes. This intuition is applied in a financial context by proposing an investment strategy that concentrates allocations in a few assets when two conditions","summary":"The global balance is a well-known indicator of the behavior of a signed network. Recent literature has introduced the concept of local balance as a measure of the contribution of a single node to the overall balance of the network. In the present research, we investigate the potential of using…","strategy":"Local‑balance based asset selection"},"assumptions":{"status":"available","items":[{"assumption":"Rolling window length ΔT is at least the number of assets N to ensure the correlation matrix is full rank.","type":"data","stated":"explicit","quote":"The experiments are performed using a standard rolling-window setting, where the window length ∆T satisfies the condition ∆T ≥N. This assumption ensures that C is a full-rank matrix."},{"assumption":"The rolling window is shifted by a step of Δt = 10 days.","type":"execution","stated":"explicit","quote":"In the applications, the rolling window is shifted by a step of ∆t = 10 days."},{"assumption":"Only stocks that remained in the index throughout the reference period are included.","type":"data","stated":"explicit","quote":"The number N of stocks in each dataset is smaller than the nominal number of index constituents, as the analysis is restricted to stocks that remained in the index throughout the reference period."},{"assumption":"When the selection conditions are met, the portfolio is equally weighted among the selected assets; otherwise it is equally weighted across all N assets.","type":"execution","stated":"explicit","quote":"Since the specific allocation rule is not the focus of the present work, we invest in an equally weighted portfolio of the selected assets when the conditions are met, and in an equally weighted portfolio (1/N) otherwise."},{"assumption":"High global balance corresponds to periods of financial crisis where most assets report losses.","type":"market","stated":"explicit","quote":"In contrast, when the global balance of a financial correlation signed network is high, all nodes (assets) tend to report losses."},{"assumption":"The analysis uses daily log returns as the input data.","type":"data","stated":"explicit","quote":"The datasets contain daily log returns from 05/01/2005 to 17/09/2020."}]},"datasets":{"status":"available","items":[{"name":"DAX","provider":null,"kind":"market-prices","access":"unknown","period":"05/01/2005 to 17/09/2020","universe":"18 stocks that remained in the DAX index throughout the period","quote":"The indices considered are DAX (N = 18), ESX (N = 42), FTSE (N = 80), and NIKKEI (N = 199)."},{"name":"ESX","provider":null,"kind":"market-prices","access":"unknown","period":"05/01/2005 to 17/09/2020","universe":"42 stocks that remained in the ESX index throughout the period","quote":"The indices considered are DAX (N = 18), ESX (N = 42), FTSE (N = 80), and NIKKEI (N = 199)."},{"name":"FTSE","provider":null,"kind":"market-prices","access":"unknown","period":"05/01/2005 to 17/09/2020","universe":"80 stocks that remained in the FTSE index throughout the period","quote":"The indices considered are DAX (N = 18), ESX (N = 42), FTSE (N = 80), and NIKKEI (N = 199)."},{"name":"NIKKEI","provider":null,"kind":"market-prices","access":"unknown","period":"05/01/2005 to 17/09/2020","universe":"199 stocks that remained in the NIKKEI index throughout the period","quote":"The indices considered are DAX (N = 18), ESX (N = 42), FTSE (N = 80), and NIKKEI (N = 199)."}],"known":[],"sample":{"start":"2005-05","end":"2020-09","market":"NIKKEI stocks","asset_class":"equities","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":false,"detail":null,"quote":null,"source":"front_matter","rule":"R1","candidates":{"front_matter":"no","evidence_card":"unclear","evidence_card_quoted":false},"meaning":"a trading or execution result is reported without costs"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"The proposed portfolio outperforms the equally weighted 1/N portfolio across all datasets.","kind":"performance","quote":"What clearly emerges is that the proposed portfolio outperforms the 1/N portfolio across all datasets."},{"finding":"Higher final portfolio value does not come with higher volatility.","kind":"performance","quote":"Moreover, the higher final value does not imply higher volatility for the improved strategy."},{"finding":"Global balance is negatively correlated with average market return and Sharpe ratio, indicating it measures systemic risk.","kind":"statistical","quote":"As shown in Table 2.1 for all four datasets, this relationship is negative. This indicates that the values of"},{"finding":"Selected stocks exhibit higher mean returns, comparable variance, thinner left tail, fatter right tail, and greater skewness than non‑selected stocks.","kind":"statistical","quote":"Both the distributions and the box plots show that the distribution of the selected stocks is characterized by a higher mean, comparable variance, a thinner left tail, a fatter right tail, and greater skewness."},{"finding":"Experiments confirm the effectiveness of the proposed strategy in outperforming the benchmark equally weighted portfolio in both descriptive and predictive contexts.","kind":"performance","quote":"Experiments conducted on four datasets of real financial data confirm both the intuition and the effectiveness of the proposed investment strategy in outperforming the benchmark equally weighted portfolio, which approximates the market."}],"reported_performance":{"strategy":"Local‑balance based asset selection","sharpe":null,"sharpe_benchmark":0.0453,"annual_return_pct":null,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"walk-forward","headline_out_of_sample":true}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":47,"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":"2026-11-14T17:19:40Z"},"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/"}