← All problems

Shao’s 2010 conjecture on self-normalized maxima

Let X1,X2,…X_1,X_2,\ldots be i.i.d. random variables with EX1=0\mathbb{E}X_1=0, EX12=σ2∈(0,∞)\mathbb{E}X_1^2=\sigma^2\in(0,\infty), and E∣X1∣p<∞\mathbb{E}|X_1|^p<\infty for some 2<p≤32<p\le 3. Define Mn=max⁡1≤k≤n∑j=1kXjM_n=\max_{1\le k\le n}\sum_{j=1}^kX_j, Vn=(∑j=1nXj2)1/2V_n=\left(\sum_{j=1}^nX_j^2\right)^{1/2}, and βn,p=E∣X1∣p/(n(p−2)/2σp)\beta_{n,p}=\mathbb{E}|X_1|^p/(n^{(p-2)/2}\sigma^p). Is it true that, uniformly for 0≤x≤βn,p−1/p0\le x\le\beta_{n,p}^{-1/p}, P(Mn≥xVn)=2{1−Φ(x)}{1+O(1)βn,p(1+x)p}\mathbb{P}(M_n\ge xV_n)=2\{1-\Phi(x)\}\left\{1+O(1)\beta_{n,p}(1+x)^p\right\}, where Φ\Phi is the standard normal distribution function?

References

Source: arXiv

Progress summary

Refreshed
Open · solution reported

A 2026 preprint claims to prove Shao’s conjecture, but the result has not been independently checked.

Shao’s conjecture concerns sharp moderate-deviation estimates for self-normalized maxima under finite moment assumptions. Earlier work established the estimate under a finite fourth moment, while a later paper claimed the finite third moment suffices.

Known results

  • Hu, Shao, and Wang: under a finite fourth moment, the normalized maximum estimate holds uniformly for x∈[0,o(n1/6))x\in[0,o(n^{1/6})).
  • Liu, Shao, Shi, and Xu: an earlier result gives the same range under a finite third moment in the identically distributed case, and describes that moment condition as optimal.

October 2026 claimed resolution

An October 2026 preprint by Song-Hao Liu, Qi-Man Shao, Hao Shi, and Jing-Yu Xu claims a relative-error estimate reaching the truncated-moment scale throughout its stated moderate-deviation range, which would settle the conjecture. The claim is reported only in an unrefereed preprint and has no independent verification in the retrieved sources.

Current status (as of October 2026): The conjectured estimate is claimed proved by an unrefereed preprint, but the general resolution remains unverified.

Sources

Solutions 0

No solutions have been posted yet.