{"id":"ec5e2393-881f-4d68-b622-86fe4c54ade9","arxiv_id":"2603.06851","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For heavy-tailed valuations with finite p-th moment (1<p<2), the minimax regret of contextual bilateral trade is T^{1-2β(p-1)/(βp+d(p-1))} up to logs, with matching upper and lower bounds.","lead":"This paper derives minimax regret rates for online bilateral trade when buyer and seller valuations have infinite variance but bounded density. It shows heavy tails turn logarithmic regret into a polynomial rate that depends on the p-th moment of the noise.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lower bound's moment-matching construction is under-specified: KL preservation requires P0 and P1 to share atomic support, which the text never states; if supports are disjoint, KL is infinite and Proposition 6.1 fails.","rationale":"The upper-bound half (Theorems 3.2–3.3 and Lemma 3.1) appears internally consistent: the generalized self-bounding property, the truncated-mean concentration, the epoch geometric-sum argument, and the bias–variance tradeoff all check out, including the p=2 and p→1+ sanity limits. The central risk is the lower bound, and the reader correctly identifies Section 6, Step 3 as the soft spot. However, the reader's formulation ('with disjoint compact supports the KL would be infinite') slightly misstates the intended construction: the two distributions should share each bump, and non-overlap refers to bumps centered at different atoms. The real problem is that the manuscript never says the atom locations coincide, so the KL-preservation claim is ambiguous and unsupported as written. This is fixable in a revision. I also flag the abstract's parameter-free algorithm claims—median-of-means pricing and cell-width tournament—which have no corresponding proof in the body; this is a separate completeness issue but does not change the conditional verdict. A concrete re-derivation of the smoothed KL would settle whether the lower bound is merely under-written or actually broken.","tokens_in":9885,"tokens_out":39727,"duration_ms":352618,"concrete_test":"Write out the smoothed construction explicitly: let P0=(1-γ)δ0+γδa and P1=(1-γ')δ0+γ'δa with γ=c(ε/σ)^{p/(p-1)}, γ'=γ-c^{1/p}(ε/σ)^{p/(p-1)}, a=σγ^{-1/p}; replace atoms with uniform bumps of width 1/L. Verify (i) densities ≤L, (ii) mean gap ε, (iii) p-th moments ≤const·σ^p, (iv) KL(P0∥P1)=KL(Bern(γ)∥Bern(γ'))=O(ε^{p/(p-1)}). If this computation succeeds, the proof sketch is confirmed and only needs rewriting; if the KL comes out infinite because the original two-point supports are disjoint, Proposition 6.1 is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 6, Step 3 smooths the two-point lower-bound distributions by replacing atoms with uniform bumps of width 1/L and claims 'the KL divergence is exactly preserved' because 'within each bump region, both distributions share the same support.' This is only true if the original discrete P0 and P1 place their atoms at the same locations (e.g., P0=(1-γ)δ0+γδa, P1=(1-γ')δ0+γ'δa). The manuscript never states this shared-support condition. The parenthetical '(Non-overlap of bumps requires 1/L<1/γ...)' refers to non-overlap of the bump intervals around distinct atoms within each distribution; a reader can easily misread it as requiring the two distributions' supports to be disjoint, in which case KL is +∞ and the Le Cam bound \\bar p_e≥1/4 is invalid. Since Proposition 6.1 is the only lower bound establishing exact minimax optimality, the argument is not fully rigorous as written. A standard same-support construction would fix it (KL then equals the Bernoulli KL and is O(ε^{p/(p-1)})), so the concern is about specification, not impossibility. Additionally, the abstract's 'fully parameter-free' algorithms (median-of-means pricing, cell-width tournament) do not appear anywhere in the body; this is a separate overclaim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies repeated contextual bilateral trade under full feedback when trader valuation noise has bounded density but possibly infinite variance. The main technical contributions are: (i) an extension of the self-bounding property to real-valued valuations (Lemma 3.1), (ii) epoch-based algorithms using coordinate-wise and cell-wise truncated-mean estimation that achieve O~(T^{(2-p)/p}) regret in the parametric case and O~(T^{1-2β(p-1)/(βp+d(p-1))}) in the nonparametric β-Hölder case under a finite p-th moment (1<p<2), and (iii) a claimed minimax lower bound via Assouad's method with a smoothed moment-matching construction. The rates interpolate between the p=2 nonparametric rate and a linear rate as p→1+.","tokens_in":10201,"tokens_out":19204,"duration_ms":170772,"significance":"If the lower bound is made rigorous, the paper would establish the first minimax characterization for bilateral trade in the infinite-variance regime and gives a clean illustration of how robust mean estimation interacts with the quadratic self-bounding structure of bilateral trade. The upper-bound analysis is careful, the rate formula is natural, and the extension of the self-bounding property to R-valued valuations is a useful standalone contribution. The main caveats are that the lower bound's KL-preservation step is under-specified as written and the abstract advertises parameter-free algorithms that do not appear in the body.","major_comments":[{"comment":"The KL-preservation claim after smoothing is not rigorous. If the discrete P0 and P1 have atoms at different locations (as in standard two-point constructions with +a and -a), replacing atoms by disjoint uniform bumps gives densities with disjoint supports and KL=∞, invalidating the subsequent Le Cam step. The text should explicitly specify a fixed-support construction (both distributions supported on a common set of atoms, e.g., {0,a}, with different weights) so that after smoothing the supports coincide on each bump; then KL is exactly preserved and is O(ε^{p/(p-1)}). This is load-bearing for Proposition 6.1 and for the claimed exact minimax rate.","section":"Section 6, Step 3 (Eq. (22))"},{"comment":"The abstract claims fully parameter-free algorithms (median-of-means pricing, cell-width tournament) that achieve the rates without knowledge of (p, σ_p) or β, and states that 'tail-adaptivity is free.' No such algorithms, definitions, or proofs appear in Sections 3–7; the Discussion lists only the truncated-mean results. This unsupported claim should either be substantiated with a dedicated section or removed.","section":"Abstract vs. body"},{"comment":"The truncation threshold is set to τ=(u n / log(dT))^{1/p}, but the cited Lemma 1 of [4] requires τ=(u n / δ)^{1/p} with δ=1/(dT), i.e., (u n dT)^{1/p}. The proof of Eq. (5) therefore does not follow from the cited result as written. This gap affects the upper bound of Theorem 3.2. It is likely fixable by choosing τ=(u n dT)^{1/p} or by providing a Bernstein-based proof, but the current text is not rigorous as stated.","section":"Section 4, Eq. (4)"}],"minor_comments":[{"comment":"The claim σ_p^p + (2L)^{-p} ≤ σ_p^p(1+o(1)) is not correct for fixed L, since the added bump contribution is a constant depending on L. The construction should either build in a margin in the discrete moment bound or state the bound as O(σ_p^p + L^{-p}).","section":"Section 6, Step 3"},{"comment":"The proposition adds the condition f_ξ(0)+f_ζ(0)>0, which is not part of Assumptions 2.1–2.3. This is acceptable for a minimax lower bound, but the statement should be phrased as 'there exist noise distributions satisfying the assumptions and this condition such that...' to avoid implying the lower bound holds for every instance in the class.","section":"Proposition 6.1"},{"comment":"The summation notation in the Assouad reduction is ambiguous ('subcube' singular). Please specify that the sum is over rounds whose context falls in the subcube where the two hypotheses differ, and spell out how the factor T h^d arises.","section":"Section 6, Eq. (20)"},{"comment":"The abstract says 'exact minimax rate' while the body consistently says 'up to logarithmic factors.' Unify the wording to avoid overclaiming.","section":"Abstract and Introduction"}],"recommendation":"major_revision","confidential_remarks":"The paper is a good candidate after revision. The lower-bound gap is likely fixable with a same-support moment-matching construction, and the threshold mismatch in Section 4 is a straightforward correction. The parameter-free claim in the abstract must either be implemented or removed, as it is currently unsupported. I do not see grounds for rejection if these issues are addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real contribution here is the upper-bound half: the generalized self-bounding property (Lemma 3.1) for real-valued valuations under bounded density and finite first moment is a legitimate extension of Bachoc et al., and the epoch-based truncated-mean analysis in Sections 4 and 5 is coherent and checkable. I verified the algebra in the balancing step; it recovers Stone's rate at p=2 and degrades to the trivial linear rate as p->1+. That part of the paper is in good shape.\n\nThe lower bound is another story. Proposition 6.1's smoothed moment-matching construction is under-specified. As written, Step 3 claims KL is \"exactly preserved\" while also requiring non-overlapping bumps. If the bumps for P0 and P1 have disjoint supports, KL is infinite and the Le Cam bound collapses. The fix is standard — smooth two-point distributions that share the same atomic locations, so the uniform bumps overlap and KL becomes the Bernoulli KL, which is O(epsilon^{p/(p-1)}). But that shared-support condition is never stated, and the parenthetical about non-overlap actively invites the wrong reading. This is a specification bug rather than an impossibility result, but it is load-bearing: without a valid lower bound, the exact minimax claim is unsupported. The parametric lower bound in Remark 6.2 inherits the same issue.\n\nThere's also a separate mismatch: the abstract promises fully parameter-free algorithms — median-of-means pricing and a cell-width tournament — and those appear nowhere in the body. That's an overclaim and should be either added or cut.\n\nNone of this is fatal. The core contribution, the heavy-tail upper bounds and the self-bounding extension, looks solid and novel. The lower bound is likely repairable with the standard same-support construction, and the parameter-free promise is either a missing section or an overstatement. Both are addressable in revision. I would send this to peer review, but I'd ask for those two fixes before accepting anything. The reader's assessment matches mine: conditional, not reject.","headline":"Solid heavy-tail upper bounds with a genuine lower-bound gap and an abstract that oversells; worth a careful revision, not a desk reject.","tokens_in":10648,"tokens_out":1112,"would_cite":true,"duration_ms":13669,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"For two-sided markets with heavy-tailed valuations, the minimax regret rate is set by the tail exponent p and the smoothness β; this paper proves it with matching upper and lower bounds.","keywords":["bilateral trade","heavy tails","infinite variance","minimax regret","bounded density","truncated mean","nonparametric regression","full feedback"],"falsifier":"Compute the KL divergence between the two smoothed noise distributions defined in Section 6, Step 3 (two-point distributions with atoms replaced by uniform bumps); if the bumps have disjoint supports, the KL divergence is infinite, contradicting the claimed O(ε^{p/(p-1)}) bound. A direct calculation would settle whether the lower-bound proof is valid, or an explicit overlapping-support pair with bounded density, finite p-th moment, mean gap ε, and KL O(ε^{p/(p-1)}) would repair it.","tokens_in":9748,"feed_emoji":"🤝","tokens_out":7923,"duration_ms":74902,"temperature":0.7,"pith_summary":"The paper asks what a price-setting broker can learn when trader valuations have heavy tails—so heavy that variance is infinite—but still have a bounded probability density and a finite p-th moment for some p between 1 and 2. The central claim is that the minimax regret is still sublinear: in the nonparametric case it scales as Õ(T^{1-2β(p-1)/(βp+d(p-1))}), interpolating between the classical finite-variance rate at p=2 and the linear rate as p approaches 1. The key structural discovery is that the self-bounding property—expected regret of mispricing by δ is at most Lδ²—extends to real-valued valuations under bounded density alone, which reduces regret control to robust mean estimation. The paper then shows epoch-based truncated-mean algorithms achieve this rate and a matching lower bound establishes minimax optimality. If correct, this closes the heavy-tailed full-feedback question for bilateral trade and shows tail-adaptivity is free under full feedback.","feed_headline":"Heavy-tailed markets: minimax price-learning rate found","feed_subtitle":"Matching upper and lower bounds hold for infinite-variance valuations, interpolating between known rates.","key_machinery":"The central object is the generalized self-bounding property, Lemma 3.1: with noise densities bounded by L, expected regret from deviating δ from the true market value is at most Lδ². This identity—derived from h'(δ) = -δ(f_ξ(δ)+f_ζ(δ))—squares the estimation error, so a robust estimator of the market value (coordinate-wise or cell-wise truncated means) directly controls regret. The other load-bearing piece is the lower-bound construction: a smoothed moment-matching pair of bounded-density, finite-p-th-moment noise distributions with mean gap ε and KL O(ε^{p/(p-1)}), combined with a standard reduction over many hypotheses. The epoch-based algorithm (doubling epochs, re-estimating from the pr","core_discovery":"The paper establishes that for contextual bilateral trade with full feedback, under bounded noise density and finite p-th moment (1<p<2), the minimax regret in T is Θ̃(T^{1-2β(p-1)/(βp+d(p-1))}) when the market value function is β-Hölder, and Θ̃(T^{(2-p)/p}) in the parametric linear case. The proof rests on a generalized self-bounding property: for any price π, E[g(m,V,W)-g(π,V,W)] ≤ L|m-π|², holding for real-valued valuations whenever the noise density is bounded and the first moment is finite. This turns pricing into mean estimation, and the paper uses coordinate-wise and cell-wise truncated means to handle infinite variance, together with an epoch-based schedule. A matching lower bound is","pith_inferences":["Because the self-bounding property holds under bounded density alone, the same quadratic regret–error coupling should apply to any market design with a similar interval structure, such as posted-price mechanisms with two-sided reservation values.","The lower-bound construction, once its KL computation is rigorously instantiated with overlapping bump supports, would also imply a new minimax rate for robust mean estimation under bounded density with p-th moments, a result of independent interest.","The rate has the form T^{1 - 2β(p-1)/(βp+d(p-1))}; as the dimension d grows, the tail exponent p has less influence, suggesting that high-dimensional pricing suffers mainly from the curse of dimensionality, with heavy tails almost free."],"forward_implications":["Minimax optimality is established up to log factors, so no full-feedback algorithm for heavy-tailed bilateral trade can improve on these rates in T.","At p=2 the nonparametric rate reduces to the classical nonparametric regression rate T^{d/(2β+d)}, recovering known finite-variance results.","As p→1+, the rate approaches T, matching the trivial worst-case and showing that the tail moment p fundamentally controls learnability.","The parameter-free variants (median-of-means pricing and cell-width tournament) attain the same rates without knowing p, σ_p, β, or the norm bound.","The generalized self-bounding property suggests that any future robust estimation method for the market value automatically transfers to regret bounds, decoupling estimation and pricing."],"fun_headline_variants":["Minimax regret rate found for heavy-tailed trade","Infinite variance? Pricing still hits minimax rate","Heavy tails: exact minimax pricing rate discovered","Tail-adaptive pricing is free under full feedback"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The lower bound's moment-matching step assumes two bounded-density noise distributions can be made with a mean gap ε while remaining statistically close, but the paper's construction (separate uniform bumps) would make them statistically infinitely far apart, so this step is the point most likely to fail.","fun_headline_variants_meta":{"raw":{"variants":["Minimax regret rate found for heavy-tailed trade","Infinite variance? Pricing still hits minimax rate","Heavy tails: exact minimax pricing rate discovered","Tail-adaptive pricing is free under full feedback"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000621,"raw_usage":{"total_tokens":2769,"prompt_tokens":852,"completion_tokens":1917,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":596,"completion_tokens_details":{"reasoning_tokens":1855}},"tokens_in":596,"tokens_out":1917,"duration_ms":13378,"temperature":1.0,"reasoning_tokens":1855,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T18:39:12.259544+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the KL divergence between the two smoothed noise distributions defined in Section 6, Step 3 (two-point distributions with atoms replaced by uniform bumps); if the bumps have disjoint supports, the KL divergence is infinite, contradicting the claimed O(ε^{p/(p-1)}) bound. A direct calculation would settle whether the lower-bound proof is valid, or an explicit overlapping-support pair with bounded density, finite p-th moment, mean gap ε, and KL O(ε^{p/(p-1)}) would repair it.","supporting_citations":[],"review_version":2}