{"paper_id":"2504.20116","title":"Compounding Effects in Leveraged ETFs: Beyond the Volatility Drag Paradigm","authors":["Chung-Han Hsieh","Jow-Ran Chang","Hui Hsiang Chen"],"paper_date":"2025-04-28","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2504.20116v1","page_url":"https://thequant.space/flowcharts/compounding-effects-in-leveraged-etfs-beyond-the-volatility/","scores":{"empirical_rigor":6,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a unified theoretical framework that incorporates discrete‑time and continuous‑time regime‑switching dynamics to characterize LETF behavior under various return dynamics and rebalancing frequencies.","quote":"We develop a uniﬁed framework incorporating discrete-time and continuous-time regime switching dynamics, allowing precise characterization of LETF behavior across diverse market conditions.","summary":"A common belief is that leveraged ETFs (LETFs) suffer long-term performance decay due to \\emph{“volatility drag”}. We show that this view is incomplete: LETF performance depends fundamentally on return autocorrelation and return dynamics. In markets with independent returns, LETFs exhibit positive…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Underlying daily returns are strictly stationary with mean µ and autocorrelation γk.","type":"model","stated":"explicit","quote":"Assume that the underlying daily returns {Xt}t≥1 is strictly stationary with mean µ = E[Xt] and autocorrelation γk = E[XtXt+k]."},{"assumption":"Fees and tracking errors are small (|Xt|, |f|, |et| ≤ M with M ≪ 1) and have zero mean.","type":"execution","stated":"explicit","quote":"Let f denote the daily fee, and {et} be the tracking error and |Xt|, |f|, |et| ≤M, for some suﬃciently small M ≪1 with E[et] = E[etXs] = E[etes] = 0 for t̸ = s."},{"assumption":"Daily returns are i.i.d. when γk = 0 for k ≥ 1.","type":"statistical","stated":"explicit","quote":"Moreover, under the i.i.d. model γk = 0 for k ≥1, hence E[CEn] = −nf, which is strictly negative only because of costs."},{"assumption":"Daily rebalancing is feasible and costs are captured by a constant daily fee f.","type":"execution","stated":"explicit","quote":"f is the daily cost (incorporating expense ratio and transaction costs, e.g., f ≈ 1/252(fexpense + ftransaction)."},{"assumption":"The AR(1) coefficient φ determines market regime (φ>0 momentum, φ<0 mean‑reversion).","type":"model","stated":"implied","quote":"Under AR(1) dynamics, arbitrage‑like constructions, as seen in Section 3.1.2, analogous to the independent case may persist when the AR coeﬃcient φ > 0 (momentum regime)."}]},"datasets":{"status":"available","items":[{"name":"SPY daily closing price data","provider":null,"kind":"market-prices","access":"public","period":"January 1993 to December 2023","universe":"S&P 500 ETF","quote":"closing price data for SPY, an ETF that tracks the S&P 500 index, spanning from January 1993 to December 2023 with about a total of 7,800 trading days."},{"name":"QQQ daily closing price data","provider":null,"kind":"market-prices","access":"public","period":"approximately 20 years","universe":"Nasdaq‑100 ETF","quote":"Using approximately 20 years of historical data from S&P 500 ETF (Ticker: SPY) and Nasdaq-100 ETF (Ticker: QQQ)..."}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"partial","headline_net_of_costs":null,"detail":"Fees and transaction costs amount to roughly 0.8–1.0 percentage points per annum; daily fee f approximated as 1/252 of expense plus transaction cost.","quote":"Note that transaction costs and fees account for approximately 0.8 to 1.0 percentage points per annum of the observed underperformance, while bid-ask spreads, slippage, and tracking error might contribute additional deviations.","source":"evidence_card","rule":"R2","candidates":{"front_matter":"unclear","evidence_card":"partial","evidence_card_quoted":true},"card_detail":"Fees and transaction costs amount to roughly 0.8–1.0 percentage points per annum; daily fee f approximated as 1/252 of expense plus transaction cost.","card_quote":"Note that transaction costs and fees account for approximately 0.8 to 1.0 percentage points per annum of the observed underperformance, while bid-ask spreads, slippage, and tracking error might contribute additional deviations.","meaning":"costs appear only as a robustness check, a flat fee without spread or impact, or outside the headline numbers"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"In markets with independent returns, LETFs exhibit a positive expected compounding effect, i.e., they tend to outperform their target multiples on average.","kind":"theoretical","quote":"In markets with independent returns, we show that LETFs exhibit positive expected compounding eﬀects to their target multiples, contrary to conventional wisdom."},{"finding":"Positive autocorrelation (momentum) enhances LETF returns, while negative autocorrelation (mean‑reversion) leads to underperformance.","kind":"theoretical","quote":"In serially correlated markets, trends enhance returns while mean reversion induces underperformance."},{"finding":"Daily rebalancing improves performance in momentum‑driven regimes, whereas less frequent rebalancing mitigates losses in mean‑reverting regimes.","kind":"performance","quote":"Daily rebalancing enhances returns in trending markets, whereas weekly or monthly rebalancing mitigates losses in oscillating or mean‑reverting regimes."},{"finding":"Empirically, positive‑β synthetic LETFs show consistently positive compounding effects during upward‑trending periods (2009–13, 2020–21) and negative effects during sideways markets.","kind":"performance","quote":"... positive‑β synthetic LETFs exhibit consistently positive compounding eﬀects."},{"finding":"The magnitude of compounding deviations increases with leverage; highest‑beta LETFs (β = ±3) experience the largest positive or negative effects, especially in high‑volatility periods.","kind":"performance","quote":"Across all market regimes, compounding eﬀects are consistently largest in magnitude for the highest leverage ratios (β = ±3), with extreme deviations occurring during high‑volatility periods."}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":35,"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":"2026-11-14T17:39:38Z"},"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/"}