Pith. sign in

REVIEW 4 major objections 5 minor 65 references

Randomness, exchangeability, and conformal prediction

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read Every IID-valid confidence predictor can be turned into a conformal predictor with only a bounded loss in efficiency.

desk verdict A clean functional translation of the universality result, but the two load-bearing lemmas live in the companion paper, so review needs [50] in hand. read the letter →

arxiv 2501.11689 v3 pith:BXHZVD2E submitted 2025-01-20 cs.LG math.STstat.MLstat.TH

classification cs.LGmath.STstat.MLstat.TH MSC 60G0968Q3062F03
keywords conformalpredictionexchangeabilityIIDassumptionfunctionaltheoryofrandomnesse-valuesp-valuesalgorithmicclassification
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that the algorithmic theory of randomness, which links IID data to exchangeability but leaves unspecified additive constants, can be replaced by a functional theory of randomness in which the same statements become explicit inequalities between classes of confidence predictors. The main result, Corollary 3, says that in classification with a finite label space $\mathcal{Y}$, every confidence predictor valid under IID data can be transformed into a conformal predictor with only a bounded loss: for any $\delta\in(0,1)$ there is an IID e-variable $G$ such that $P'(z_1,\dots,z_n,x_{n+1},y)\le \mathrm{e}(|\mathcal{Y}|-1)^2 \delta^{-1} G(z_1,\dots,z_{n+1})^2 P(z_1,\dots,z_n,x_{n+1},y)^{1-\delta}$ for every false label $y$. A sympathetic reader cares because conformal prediction is distribution-free and widely implemented; the claim is that believing data are IID rather than merely exchangeable cannot buy more than this explicit factor in classification.

What carries the argument

The machinery is the cube of eight function classes obtained from three dichotomies: the data-generating assumption (IID versus exchangeability), the type of confidence value (p-values versus e-values), and train-invariance (optional). The two vertices that matter are $P_R$, the most general IID p-predictors, and $P_{tX}$, which coincides with conformal predictors. The argument travels the red path $P_R\to E_R\to E_X\to E_{tX}\to P_{tX}$, and the load-bearing identities are Theorem 3, $E_R=E_X E_{iR}$, which factors any IID e-variable into an exchangeability e-variable and an invariant IID e-variable; Theorem 4 and Theorem 6, which bound the loss in the false-label and train-invariance steps; and the two calibrators $f(p)=\delta p^{\delta-1}$ and $e\mapsto 1/e$ that enter and leave the e-value world.

What would settle it

Check the companion proof [50] for the precise conditions under which Theorem 4 and Theorem 6 are proved; if either theorem needs a restriction that the present statement does not state, such as $n$ large, $\mathcal{X}$ finite, or $|\mathcal{Y}|$ above a threshold, then Corollary 3 as formulated in this paper is false. Concretely, test the minimal case $|\mathcal{Y}|=2$, $n=1$: if the claimed factors $1/(\mathrm{e}(|\mathcal{Y}|-1))$ and $1/(|\mathcal{Y}|-1)$ cannot be attained by any e-variable $G$ satisfying $\mathbb{E}_Q[G]\le 1$ for all IID probability measures $Q$, the inequality (19) fails.

Watch

Extended reading notes

Core claim

The central claim is that conformal prediction is universal under the IID assumption in a quantitative, constant-free sense. Formally, Corollary 3 asserts that for every IID p-predictor $P\in P_R$ and every $\delta\in(0,1)$, there exists a conformal predictor $P'\in P_{tX}$ and an IID e-variable $G$ satisfying the inequality above for all training sequences, test objects, and false labels. The proof moves along the chain $P_R\to E_R\to E_X\to E_{tX}\to P_{tX}$: convert the p-value to an e-value by calibration, factor an IID e-variable into an exchangeability e-variable times an invariant IID e-variable, pass to false labels with the factor $1/(\mathrm{e}(|\mathcal{Y}|-1))$, make the predictor train-invariant with the factor $1/(|\mathcal{Y}|-1)$, and convert back to a p-value. If the paper is right, any method that reports valid p-values under the standard IID assumption can be mimicked by a conformal method, up to the stated factors, unless the whole augmented data sequence itself looks non-IID.

Load-bearing premise

The two steps that carry the whole argument, Theorem 4 and Theorem 6, are stated here without proof and deferred to the companion preprint, so the paper stands or falls on their validity as stated for all $n$ and all measurable object spaces.

Editorial extensions

If this is right

  • If Corollary 3 is correct, the universality result of Nouretdinov, V'yugin, and Gammerman no longer depends on unspecified constants: the translation from IID p-predictors to conformal predictors is governed by explicit factors involving $\mathrm{e}(|\mathcal{Y}|-1)^2/\delta$ and the IID e-variable $G$.
  • Because conformal predictors are exactly the train-invariant exchangeability p-predictors, the result identifies which IID methods can be replaced: any method whose p-values are valid under IID can be converted into a conformal one, so the weaker exchangeability assumption supports the same level of confidence up to these factors.
  • The fundamental limitation of conformal prediction, that its p-values cannot go below $1/(n+1)$, is shown to be a limitation of any IID-valid method in classification, giving $D_R \le \log(n+1)+O(\log\log(n+1))$ in the prediction-proper regime.
  • In the train-invariant case, Corollary 4 improves the bound to $e(|\mathcal{Y}|-1)/\delta \cdot G \cdot P^{1-\delta}$, making the loss especially small for the most natural class of IID predictors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The squared factor $(|\mathcal{Y}|-1)^2$ in (19) comes from composing two independent losses, Theorem 4 and Theorem 6; since Theorem 5 shows the single $|\mathcal{Y}|$ factor is asymptotically optimal, a direct p-to-p argument that bypasses e-values might reduce the squared factor, an improvement the paper's own route cannot give.
  • In practice this suggests that the IID-versus-exchangeability debate changes character: when data are IID, conformal methods should be competitive with bespoke IID methods up to these factors, so improving the nonconformity measure likely matters more than weakening the validity assumption.
  • For regression or infinite label spaces the constants have no literal meaning, since the comparison must depend on a metric between labels; the classification framing here is the favorable case, and the regression analysis in [50] may need separate constants.
  • The same calibrate-factor-average-calibrate template could be applied to other pairs of validity assumptions, such as covariate shift or conditional validity, suggesting that functional e-value decompositions are a reusable tool for comparing prediction settings.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper reviews the relationship between the IID and exchangeability assumptions, with special attention to conformal prediction, and presents a translation of Nouretdinov, V'yugin, and Gammerman's universality result into the 'functional theory of randomness.' In the functional setting, the central result is Corollary 3: for every IID p-predictor P there exist a conformal (train-invariant exchangeability) predictor P' and an IID e-variable G such that inequality (19) holds for every false label, with the explicit constant e(|Y|-1)^2/delta and the factor P^{1-delta}. The paper contains a full proof of Theorem 3, and the proof of Corollary 3 is a transparent chaining of calibration, Kolmogorov's step, the train-invariance step, and e-to-p calibration. However, the two load-bearing steps, Theorem 4 (the 1/(e(|Y|-1)) transfer from invariant IID e-values to false labels) and Theorem 6 (the 1/(|Y|-1) train-invariance step), are stated without proofs and deferred to the author's companion preprint [50].

Significance. If the deferred theorems are correct in the claimed generality, the paper is a valuable contribution: it removes unspecified additive constants from the algorithmic theory, provides explicit constants, gives a self-contained proof of the key factorization theorem, and clearly identifies which steps incur the finite-label-space penalties. The paper is also honest about what is deferred, which is a methodological strength. The included proof of Theorem 3 and the chaining argument show that the overall strategy is sound, but the manuscript as submitted does not fully establish the advertised universality result because its two key technical links are not proved in the text.

major comments (4)
  1. [6.2, Theorem 4 (Eq. (14))] Theorem 4 is essential for Corollaries 1, 2, 3, and 4, but no proof is given; the text only states 'A formal proof is given in [50].' Since [50] is the author's own companion preprint, the manuscript is not self-contained at the exact point where the constant 1/(e(|Y|-1)) is introduced. Please include the proof in this paper, or state precisely the hypotheses under which the result holds (including any restrictions on n, on the object space X, and on the e-variables) and verify that Corollary 3 remains valid under those hypotheses.
  2. [6.3, Theorem 6] The train-invariance step with factor 1/(|Y|-1) is likewise deferred, with the text saying 'For a simple proof, see [50].' This theorem is used in the proof of Corollary 3 at inequality (22), so without it the chained universality statement cannot be verified from this manuscript alone. Please provide the proof or make the result a clearly marked import with a precise statement of its conditions.
  3. [6.2, paragraph after Corollary 1] The claim that when E is train-invariant, 'the resulting predictor E' will also be train-invariant' is asserted without proof. Corollaries 2 and 4 rely on this preservation to obtain a conformal predictor, so this is another unproved link in the main chain. Please prove the preservation or give the exact argument in [50].
  4. [6, first paragraph] The paper states 'Very few proofs will be given, and most of them can be found in [50].' This is an explicit limitation, but it conflicts with the paper's status as a full research article, since the abstract's advertised translation result depends on these deferred proofs. The revision should either include the proofs or clearly mark the paper as a research announcement with the main results imported from [50].
minor comments (5)
  1. [6.1] In the sentence 'Our argument will also establish the closeness of the conformal e-predictors (i.e., X/p/t predictors) to the R/e predictors', the parenthetical should presumably read 'X/e/t predictors' rather than 'X/p/t predictors'.
  2. [6.2, Theorem 5] Theorem 5, the asymptotic optimality claim, is also deferred to [50]. It is not used in the proof of Corollary 3, so it is not blocking, but the paper should state explicitly that this is a citation result and, if possible, give the precise theorem number in [50].
  3. [Theorem 3 proof] The line 'It is obvious that E' is an element of E_X' is terse; adding a one-sentence justification using the permutation-average argument would make the included proof of Theorem 3 easier to check.
  4. [2] The approximation 'e ≈ 2.72' is slightly misleading; using 'e ≈ 2.718' would be more accurate.
  5. [Abstract and Corollary 3] The abstract's phrase 'without losing much in predictive efficiency' should be read in the sense of inequality (19), where the loss is controlled up to an e-variable G and a power P^{1-delta}; making this qualification explicit in the abstract would prevent over-interpretation.

Circularity Check

1 steps flagged · score 4.0 of 10

Corollary 3 rests on Theorems 4 and 6, both stated without proof and deferred to the author's own companion preprint [50]; this load-bearing self-citation raises the circularity score, though the result is not forced by definition.

  1. self citation load bearing [Section 6.2, Theorem 4 (inequality (14)); Section 6.3, Theorem 6; used in proof of Corollary 3, Section 6.4, Eq. (19)]
    ""A formal proof is given in [50]." (after Theorem 4, inequality (14)); "For a simple proof, see [50]." (after Theorem 6)."

    Corollary 3 derives the universality of conformal predictors by chaining calibration, Corollary 1, Theorem 6, and e-to-p calibration. The two substantive links are Theorem 4, which transfers an invariant IID e-value to a false label with constant 1/(e(|Y|-1)), and Theorem 6, the train-invariance step with constant 1/(|Y|-1). Both are stated in this manuscript without proof, and the only support offered is the author's own companion preprint [50]. The central claim thus inherits its non-definitional content from same-author material rather than from an argument contained in the paper; if [50] carries unstated conditions, the advertised reduction is not established in the stated generality. This is load-bearing self-citation, not a definitional reduction.

full rationale

The paper is not circular in the strong sense: Theorem 3, the ER = EX EiR factorization, is proved independently; the calibration and e-to-p steps are standard; and Corollary 3 is a genuine chaining of inequalities rather than a renamed input or a fitted parameter presented as a prediction. The circularity burden comes from the two central lemmas, Theorems 4 and 6, which are required for the key transitions in Corollary 3 and are deferred to [50], a companion preprint by the same author. Theorem 5 is also deferred to [50] but is not needed for Corollary 3. Because the only support for those load-bearing steps is a same-author citation, the derivation is not fully self-contained; however, the conclusion is not equivalent to its assumptions by construction, so the appropriate score is moderate rather than high.

Assumptions & free parameters 1 free parameters · 6 assumptions · 0 invented entities

One arbitrary constant delta appears in the calibrator and in the final bound, but no data are fitted and no physical, empirical, or invented constants are introduced. The formal results rely on standard measure-theoretic probability, de Finetti's theorem, calibration results, and the algorithmic-randomness framework. The unproved Theorems 4 through 6, delegated to the author's companion [50], are entered as ad hoc assumptions because the manuscript contains neither their statements as proven nor their derivations. No new particles, forces, dimensions, or similar entities are postulated.

free parameters (1)
  • delta = arbitrary value in (0,1)
    Chosen by hand in the calibrator f(p) = delta * p^(delta-1), Equation (18), and it appears in the final efficiency bound (19). It is not fitted to data and is a theorem parameter rather than an empirical constant.
assumptions (6)
  • domain assumption Classification setting: Z = X * Y, Y finite, |Y| >= 2, discrete sigma-algebra on Y; data are generated by IID or exchangeable probability measures on Z^(n+1).
    Section 6 fixes this as the scope of all function classes and corollaries. The theorem does not cover regression or infinite label spaces.
  • standard math de Finetti's representation theorem for exchangeable infinite sequences holds on standard Borel spaces.
    Section 2 uses it to argue that IID and exchangeability are equivalent in the infinite limit, motivating the finite-sequence analysis.
  • standard math The algorithmic theory of randomness framework, including universal p-tests and e-tests, is sound.
    Sections 3 through 5 rely on this background. Universal tests are attributed to Martin-Lof, Levin, and Gacs, and the p/e comparison inequalities are cited to [28].
  • standard math The calibrators from [60] are valid, including f(p) = delta * p^(delta-1) and the optimal e-to-p calibrator 1/e.
    Used in the proof of Corollary 3 to convert between p-values and e-values.
  • ad hoc to paper Theorems 4, 5, and 6, as stated in Sections 6.2 and 6.3, are true.
    They are the unproved core of the universality translation. Proofs are delegated to the author's companion [50], so this manuscript treats them as external results.
  • domain assumption The characterization of conformal predictors as P_tX and conformal e-predictors as E_tX holds, from [27, Proposition 1] and [49].
    This identification is imported from prior literature and is not reproved. It is needed for the interpretation of Corollaries 1 through 4.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Randomness, exchangeability, and conformal prediction." pith.science (2026). https://pith.science/paper/BXHZVD2E

@misc{pith2026250111689,
  author       = {Pith},
  title        = {Pith review of: Randomness, exchangeability, and conformal prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BXHZVD2E}},
  note         = {Machine review of arXiv:2501.11689}
}
read the original abstract

This paper argues for a wider use of the functional theory of randomness, a modification of the algorithmic theory of randomness getting rid of unspecified additive constants. Both theories are useful for understanding relationships between the assumptions of IID data and data exchangeability. While the assumption of IID data is standard in machine learning, conformal prediction relies on data exchangeability. Nouretdinov, V'yugin, and Gammerman showed, using the language of the algorithmic theory of randomness, that conformal prediction is a universal method under the assumption of IID data. In this paper (written for the Alex Gammerman Festschrift) I will selectively review connections between exchangeability and the property of being IID, early history of conformal prediction, my encounters and collaboration with Alex and other interesting people, and a translation of Nouretdinov et al.'s results into the language of the functional theory of randomness, which moves it closer to practice. Namely, the translation says that every confidence predictor that is valid for IID data can be transformed to a conformal predictor without losing much in predictive efficiency.

Figures

Figures reproduced from arXiv: 2501.11689 by the authors.

Figure 1
Figure 1. Connections between 4 deficiencies of typicalness: The connection [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Binary conformal prediction as special case of both conformal predic [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. A cube representing eight function classes. The polygonal chain [PITH_FULL_IMAGE:figures/full_fig_p019_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 62 canonical work pages

  1. [50]

    Universality of conformal prediction under the assumption of randomness

    Vladimir Vovk. Universality of conformal prediction under the assumption of randomness. Technical Report arXiv:2502.19254 [cs.LG], arXiv.org e- Print archive, June 2025

  2. [48]

    Non-algorithmic theory of randomness

    Vladimir Vovk. Non-algorithmic theory of randomness. In Andreas Blass, Patrick C´ egielski, Nachum Dershowitz, Manfred Droste, and Berndt Fink- beiner, editors, Fields of Logic and Computation III: Essays Dedicated to Yuri Gurevich on the Occasion of His 80th Birthday, volume 12180 ofLec- ture Notes in Computer Science, pages 323–340, Cham, 2020. Springer

  3. [1]

    Angelopoulos, Rina Foygel Barber, and Stephen Bates

    Anastasios N. Angelopoulos, Rina Foygel Barber, and Stephen Bates. The- oretical foundations of conformal prediction. Technical Report arXiv:2411. 11824 [math.ST], arXiv.org e-Print archive, March 2025. Pre-publication version of a book to be published by Cambridge University Press

  4. [2]

    Angelopoulos and Stephen Bates

    Anastasios N. Angelopoulos and Stephen Bates. Conformal prediction: A gentle introduction. Foundations and Trends in Machine Learning, 16(4):494–591, 2023

  5. [3]

    Angelopoulos, Emmanuel J

    Anastasios N. Angelopoulos, Emmanuel J. Cand` es, and Ryan J. Tibshirani. Conformal PID control for time series prediction. InAdvances in Neural Information Processing Systems 36 (NeurIPS 2023), 2023

  6. [4]

    Eugene A. Asarin. Some properties of Kolmogorov ∆-random finite se- quences. Theory of Probability and its Applications, 32:507–508, 1987. Russian original: О некоторых свойствах∆-случайных по Колмогорову конечных последовательностей

  7. [5]

    Eugene A. Asarin. On some properties of finite objects random in the algorithmic sense.Soviet Mathematics Doklady, 36:109–112, 1988. Russian original: О некоторых свойствах случайных в алгоритмическом смысле конечных объектов, published in 1987. 28

  8. [6]

    Cand` es, Aaditya Ramdas, and Ryan J

    Rina Foygel Barber, Emmanuel J. Cand` es, Aaditya Ramdas, and Ryan J. Tibshirani. Conformal prediction beyond exchangeability.Annals of Statis- tics, 51:816–845, 2023

Show all 65 references
  1. [7]

    Ars Conjectandi

    Jacob Bernoulli. Ars Conjectandi. Thurnisius, Basel, 1713

  2. [8]

    Conformal prediction in Python with crepes.Proceedings of Machine Learning Research, 230:236–249, 2024

    Henrik Bostr¨ om. Conformal prediction in Python with crepes.Proceedings of Machine Learning Research, 230:236–249, 2024. COPA 2024

  3. [9]

    BS ISO 80000-2:2009

    British Standards Institution.Quantities and units, Part 2: Mathematical signs and symbols to be used in the natural sciences and technology, 2010. BS ISO 80000-2:2009

  4. [10]

    Flexible and systematic uncertainty estimation with conformal prediction via the MAPIE library.Proceedings of Machine Learning Research, 204:549–581, 2023

    Thibault Cordier, Vincent Blot, Louis Lacombe, Thomas Morzadec, Ar- naud Capitaine, and Nicolas Brunel. Flexible and systematic uncertainty estimation with conformal prediction via the MAPIE library.Proceedings of Machine Learning Research, 204:549–581, 2023. COPA 2023

  5. [11]

    A. I. Dale. A study of some early investigations into exchangeability.His- toria Mathematica, 12:323–336, 1985

  6. [12]

    Freedman

    Persi Diaconis and David A. Freedman. Finite exchangeable sequences. Annals of Probability, 8:745–764, 1980

  7. [13]

    Exact expressions for some randomness tests.Zeitschrift f¨ ur Mathematische Logik und Grundlagen der Mathematik, 26:385–394, 1980

    Peter G´ acs. Exact expressions for some randomness tests.Zeitschrift f¨ ur Mathematische Logik und Grundlagen der Mathematik, 26:385–394, 1980

  8. [14]

    Learning by transduction

    Alex Gammerman, Vladimir Vovk, and Vladimir Vapnik. Learning by transduction. In Proceedings of the Fourteenth Conference on Uncertainty in Artificial Intelligence, pages 148–155, San Francisco, CA, 1998. Morgan Kaufmann

  9. [15]

    Machine Learning: Progress and Prospects

    Alexander Gammerman. Machine Learning: Progress and Prospects. An Inaugural Lecture by Alexander Gammerman, Professor of Computer Sci- ence. Presented at Royal Holloway, University of London, on 11th Decem- ber 1996. Royal Holloway, University of London, Egham, Surrey, 1997

  10. [16]

    Hedging predictions in ma- chine learning (with discussion).Computer Journal, 50:151–177, 2007

    Alexander Gammerman and Vladimir Vovk. Hedging predictions in ma- chine learning (with discussion).Computer Journal, 50:151–177, 2007

  11. [17]

    Cand` es

    Isaac Gibbs and Emmanuel J. Cand` es. Adaptive conformal inference under distribution shift. Advances in Neural Information Processing Systems, 34:1660–1672, 2021

  12. [18]

    Peter Gr¨ unwald, Rianne de Heide, and Wouter M. Koolen. Safe testing (with discussion).Journal of the Royal Statistical Society B, 86:1091–1171, 2024

  13. [19]

    Kolmogorov

    Andrei N. Kolmogorov. Grundbegriffe der Wahrscheinlichkeitsrechnung. Springer, Berlin, 1933. English translation:Foundations of the Theory of Probability. Chelsea, New York, 1950. 29

  14. [20]

    Kolmogorov

    Andrei N. Kolmogorov. Теория вероятностей. In Математика, ее со- держание, методы и значение, volume 2, pages 252–284. Издательство АН СССР, Moscow, 1956

  15. [21]

    Kolmogorov

    Andrei N. Kolmogorov. On tables of random numbers.Sankhy¯ a. Indian Journal of Statistics A, 25:369–376, 1963

  16. [22]

    количество информа- ции

    Andrei N. Kolmogorov. Three approaches to the quantitative definition of information. Problems of Information Transmission, 1:1–7, 1965. Russian original: Три подхода к определению понятия “количество информа- ции”

  17. [23]

    Logical basisfor information theoryand probability theory

    Andrei N.Kolmogorov. Logical basisfor information theoryand probability theory. IEEE Transactions on Information Theory, IT-14:662–664, 1968. Russian original: К логическим основам теории информации и теории вероятностей, published in Проблемы передачи информации

  18. [24]

    Kolmogorov

    Andrei N. Kolmogorov. Combinatorial foundations of information theory and the calculus of probabilities.Russian Mathematical Surveys, 38:29–40,

  19. [25]

    Kolmogorov and Vladimir A

    Andrei N. Kolmogorov and Vladimir A. Uspensky. Algorithms and ran- domness. Theory of Probability and Its Applications, 32:389–412, 1987. Russian original: Алгоритмы и случайность

  20. [26]

    The definition of random sequences.Information and Control, 9:602–619, 1966

    Per Martin-L¨ of. The definition of random sequences.Information and Control, 9:602–619, 1966

  21. [27]

    Transductive Confidence Machine is universal

    Ilia Nouretdinov, Vladimir V’yugin, and Alex Gammerman. Transductive Confidence Machine is universal. In Ricard Gavald` a, Klaus P. Jantke, and Eiji Takimoto, editors, Proceedings of the Fourteenth International Con- ference on Algorithmic Learning Theory, volume 2842 ofLectur...

  22. [28]

    Relations between randomness deficiencies

    Gleb Novikov. Relations between randomness deficiencies. Technical Re- port arXiv:1608.08246 [math.LO], arXiv.org e-Print archive, August 2016. Published inLecture Notes in Computer Science10307:338–350 (2017)

  23. [29]

    Hypothesis testing with e-values

    Aaditya Ramdas and Ruodu Wang. Hypothesis testing with e-values. Tech- nical Report arXiv:2410.23614 [math.ST], arXiv.org e-Print archive, May

  24. [30]

    Transduction with confidence and credibility

    Craig Saunders, Alex Gammerman, and Vladimir Vovk. Transduction with confidence and credibility. In Thomas Dean, editor,Proceedings of the Six- teenth International Joint Conference on Artificial Intelligence, volume 2, pages 722–726, San Francisco, CA, 1999. Morgan Kaufmann

  25. [31]

    Schervish.Theory of Statistics

    Mark J. Schervish.Theory of Statistics. Springer, New York, 1995. 30

  26. [32]

    The language of betting as a strategy for statistical and scientific communication (with discussion).Journal of the Royal Statistical Society A, 184:407–478, 2021

    Glenn Shafer. The language of betting as a strategy for statistical and scientific communication (with discussion).Journal of the Royal Statistical Society A, 184:407–478, 2021

  27. [33]

    Probability and Finance: It’s Only a Game! Wiley, New York, 2001

    Glenn Shafer and Vladimir Vovk. Probability and Finance: It’s Only a Game! Wiley, New York, 2001

  28. [34]

    The sources of Kolmogorov’sGrundbe- griffe

    Glenn Shafer and Vladimir Vovk. The sources of Kolmogorov’sGrundbe- griffe. Statistical Science, 21:70–98, 2006. Extended version: arXiv:1802. 06071 [math.HO]

  29. [35]

    Wiley, Hoboken, NJ, 2019

    Glenn Shafer and Vladimir Vovk.Game-Theoretic Foundations for Proba- bility and Finance. Wiley, Hoboken, NJ, 2019

  30. [36]

    Uspensky, and Nikolai Vereshchagin.Kol- mogorov Complexity and Algorithmic Randomness

    Alexander Shen, Vladimir A. Uspensky, and Nikolai Vereshchagin.Kol- mogorov Complexity and Algorithmic Randomness. American Mathemati- cal Society, Providence, RI, 2017

  31. [37]

    Shiryaev.Probability-1

    Albert N. Shiryaev.Probability-1. Springer, New York, third edition, 2016

  32. [38]

    Tibshirani, Rina Foygel Barber, Emmanuel J

    Ryan J. Tibshirani, Rina Foygel Barber, Emmanuel J. Cand` es, and Aaditya Ramdas. Conformal prediction under covariate shift. InAdvances in Neural Information Processing Systems 32 (NeurIPS 2019), 2019

  33. [39]

    Uspensky and Alexei L

    Vladimir A. Uspensky and Alexei L. Semenov. Algorithms: Main Ideas and Applications. Kluwer, Dordrecht, 1993

  34. [40]

    Vapnik.Statistical Learning Theory

    Vladimir N. Vapnik.Statistical Learning Theory. Wiley, New York, 1998

  35. [41]

    Grundlagen der Wahrscheinlichkeitsrechnung.Mathe- matische Zeitschrift, 5:52–99, 1919

    Richard von Mises. Grundlagen der Wahrscheinlichkeitsrechnung.Mathe- matische Zeitschrift, 5:52–99, 1919

  36. [42]

    Springer, Berlin, 1928

    Richard von Mises.Wahrscheinlichkeit, Statistik, und Wahrheit. Springer, Berlin, 1928. English translation:Probability, Statistics and Truth. William Hodge, London (1939)

  37. [43]

    On the concept of the Bernoulli property.Russian Math- ematical Surveys, 41:247–248, 1986

    Vladimir Vovk. On the concept of the Bernoulli property.Russian Math- ematical Surveys, 41:247–248, 1986. Russian original: О понятии бер- нуллиевости, another English translation with proofs: arXiv:1612.08859 (math.ST)

  38. [44]

    A logic of probability, with application to the foundations of statistics (with discussion)

    Vladimir Vovk. A logic of probability, with application to the foundations of statistics (with discussion). Journal of the Royal Statistical Society B, 55:317–351, 1993

  39. [45]

    Minimum description length estimators under the optimal coding scheme

    Vladimir Vovk. Minimum description length estimators under the optimal coding scheme. In Paul Vit´ anyi, editor,Computational Learning Theory, volume 904 ofLecture Notes in Computer Science, pages 237–251, Berlin,

  40. [46]

    Learning about the parameter of the Bernoulli model

    Vladimir Vovk. Learning about the parameter of the Bernoulli model. Journal of Computer and System Sciences, 55:96–104, 1997. EuroCOLT 1995 Special Issue. This paper is the journal version of [45]

  41. [47]

    Kolmogorov’s complexity conception of probability

    Vladimir Vovk. Kolmogorov’s complexity conception of probability. In Vincent F. Hendricks, Stig Andur Pedersen, and Klaus Frovin Jørgensen, editors, Probability Theory: Philosophy, Recent History and Relations to Science, pages 51–69. Kluwer, Dordrecht, 2001

  42. [49]

    Conformal e-prediction.Pattern Recognition, 166:111674,

    Vladimir Vovk. Conformal e-prediction.Pattern Recognition, 166:111674,

  43. [51]

    Testing exchangeability in the batch mode with e-values and Markov alternatives.Machine Learning, 114:99, 2025

    Vladimir Vovk. Testing exchangeability in the batch mode with e-values and Markov alternatives.Machine Learning, 114:99, 2025. Special Issue on Conformal Prediction and Distribution-Free Uncertainty Quantification

  44. [52]

    Inductive randomness predictors, On-line Compression Modellingproject(NewSeries), http://alrw.net, WorkingPaper44, May

    Vladimir Vovk. Inductive randomness predictors, On-line Compression Modellingproject(NewSeries), http://alrw.net, WorkingPaper44, May

  45. [53]

    Machine-learning applications of algorithmic randomness

    Vladimir Vovk, Alex Gammerman, and Craig Saunders. Machine-learning applications of algorithmic randomness. In Proceedings of the Sixteenth International Conference on Machine Learning, pages 444–453, San Fran- cisco, CA, 1999. Morgan Kaufmann

  46. [54]

    Algorithmic Learning in a Random World

    Vladimir Vovk, Alexander Gammerman, and Glenn Shafer. Algorithmic Learning in a Random World. Springer, Cham, second edition, 2022

  47. [55]

    On-linepredictive linear regression

    VladimirVovk, IliaNouretdinov, andAlexGammerman. On-linepredictive linear regression. Annals of Statistics, 37:1566–1590, 2009

  48. [56]

    Available as arXiv technical report arXiv:2503.02803 [cs.LG]

  49. [57]

    Kolmogorov’s contributions to the foun- dations of probability

    Vladimir Vovk and Glenn Shafer. Kolmogorov’s contributions to the foun- dations of probability. Problems of Information Transmission, 39:21–31, 2003. 32

  50. [58]

    A conversation with A

    Vladimir Vovk and Glenn Shafer. A conversation with A. Philip Dawid. Statistical Science, 40:148–166, 2025

  51. [59]

    Vladimir Vovk and Vladimir V. V’yugin. On the empirical validity of the Bayesian method. Journal of the Royal Statistical Society B, 55:253–266, 1993

  52. [60]

    Conformal e- testing

    Vladimir Vovk, Ilia Nouretdinov, and Alex Gammerman. Conformal e- testing. Pattern Recognition, 168:111841, 2025. Special Issue on Conformal Prediction and Distribution-Free Uncertainty Quantification

  53. [61]

    Confidence and discoveries with e-values

    Vladimir Vovk and Ruodu Wang. Confidence and discoveries with e-values. Statistical Science, 38:329–354, 2023. 33

  54. [64]

    E-values: Calibration, combination, and applications

    Vladimir Vovk and Ruodu Wang. E-values: Calibration, combination, and applications. Annals of Statistics, 49:1736–1754, 2021

  55. [1983]

    Russian original: Комбинаторные основания теории информации и исчисления вероятностей

  56. [1995]

    EuroCOLT 1995

    Springer. EuroCOLT 1995. 31

  57. [2025]

    Special Issue on Conformal Prediction and Distribution-Free Uncer- tainty Quantification

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.