{"id":"4b4da037-4708-4eed-ad4f-196f5ebfd294","arxiv_id":"2411.14058","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":2,"one_line_summary":"A standard wavelet power spectrum analysis of daily cryptocurrency and macro prices finds qualitative high-frequency 'hot spots' and a claimed XRP periodicity line, but provides no formal evidence for cyclical persistence or weak-form market inefficiency.","lead":"This paper applies a standard math tool, wavelet analysis, to daily prices of Bitcoin, Ethereum, XRP and several other assets, and reports visually identified patterns it calls 'hot spots' and a line-like periodicity for XRP. Because no statistical tests back these patterns, the main claim of predictable cyclical behavior is not demonstrated.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"XRP's 'line-like shape' is asserted as cyclical persistence and causality without any null-model or surrogate test; the visual feature may be a wavelet artifact of noise, so the central EMH claim currently rests on an untested interpretation.","rationale":"The reader's weakest_assumption identifies exactly the same gap: visual hot spots and line-like shapes are interpreted as evidence without statistical significance tests. I agree, and my proposed surrogate test is the minimal check that would settle it. The reader's REJECT verdict is appropriate: the paper's main contribution, the EMH/causality claim, is entirely dependent on an untested visual interpretation. I would not soften the verdict merely because the figures were unavailable in the provided text, since the absence of any null comparison is explicit in the prose; Section 3's descriptive account is the entire evidence. UNCHANGED means the verdict remains REJECT. If the surrogate test later passed, the paper would still need a predictive or structural test to support 'intrinsic causal relationship', but the first-order concern is whether the feature is signal at all. The paper does give credit for using standard wavelet methodology, but a standard tool does not by itself validate the interpretation placed on its output.","tokens_in":5085,"tokens_out":3415,"duration_ms":36144,"concrete_test":"Use the paper's daily XRP returns (July 10, 2017-December 31, 2022) and recompute the wavelet power spectrum with morl and cmor1.5-1.0 exactly as in Figures 3 and 10. Build 1,000 surrogate series: (a) Fourier-phase randomized surrogates preserving the periodogram, and (b) AR(1) or GARCH(1,1) fitted nulls. For each surrogate, compute the same wavelet power and extract the maximum power and the maximum-length contiguous ridge in the low-to-middle frequency band, restricting to times outside the cone of influence. If 5% or more of surrogates produce a ridge at least as long/strong as the XRP line, the line-like shape is not statistically significant and the claimed cyclical persistence/causality does not survive; if no surrogate matches it, the visual claim gains real support.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central conclusion (Section 4) is that red 'hot spots' and a 'line-like shape' in the wavelet power spectra (Figures 1-14) show 'cyclical persistence at different frequencies' and 'some intrinsic causal relationship', violating weak-form EMH. The load-bearing premise is that these visual features are signal, not noise. That premise is never tested. There are no significance contours (e.g., Torrence-Compo red-noise tests), no surrogate-data ensembles, no confidence intervals, and no sensitivity analysis for the mother wavelet (morl vs cmor1.5-1.0). This is not a mere presentational omission: the continuous wavelet transform is a redundant, locally autocorrelated representation, so even a single realization of iid or weakly autocorrelated returns produces patchy red regions and ridge-like structures; cone-of-influence boundary effects at the start/end of the sample can create near-linear arms. The XRP 'line' is hence exactly the kind of feature that a null process can generate. Additionally, the inference from a periodic-looking ridge to 'intrinsic causal relationship' is a non-sequitur: periodicity in returns would not by itself establish causality or predictability without an out-of-sample forecasting or predictive test. The text's 'Covid-19 peek in 2022' misdating (Section 3) is a concrete symptom of pattern-reading without date verification. Unless a null model can be rejected, the central claim is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper applies continuous wavelet transforms (Morlet and complex Morlet) to daily log returns of BTC, ETH, XRP, S&P500, gold, JPY/USD, and USD/EUR from July 10, 2017 to December 31, 2022, with the stated aim of testing the weak form of the efficient market hypothesis by detecting cyclical persistence. The authors visually interpret red 'hot spots' and a 'line-like shape' in the wavelet power spectra (Figures 1-14) as evidence of cyclical persistence and 'some intrinsic causal relationship' for certain investment horizons, especially for XRP. No statistical tests, confidence contours, null models, surrogate data, or out-of-sample checks are presented; the conclusions rest entirely on visual inspection of the spectral plots.","tokens_in":5351,"tokens_out":2613,"duration_ms":26724,"significance":"If the central claim were established with rigorous inference, the finding of predictable cyclical components in cryptocurrency returns would be relevant to the market efficiency literature and to practitioners. However, the manuscript provides no quantitative evidence supporting its conclusions: the visual features are never tested against a null model, the mother wavelet choices are not justified or varied for robustness, and the leap from spectral patterns to 'causal relationship' is not conceptually defended. Given this, the potential significance is not realized, and the paper currently offers little beyond a set of uncalibrated wavelet plots.","major_comments":[{"comment":"The central evidence for the paper's claim is visual inspection of wavelet power spectra, but no inferential procedure is applied. There are no significance contours against a red-noise or other null background (e.g., the standard Torrence-Compo approach), no surrogate data ensembles, and no confidence intervals. Because the continuous wavelet transform is a redundant, locally autocorrelated representation, even a single realization of i.i.d. or weakly autocorrelated returns produces patches of high power and ridge-like features; the reported 'hot spots' and the XRP 'line-like shape' could plausibly be such null-process artifacts. Without a statistical test, the central conclusion in Section 4 that these features indicate 'cyclical persistence' and 'intrinsic causal relationship' is unsupported.","section":"Section 3, Figures 1-14"},{"comment":"The inference from a visible line in a wavelet spectrum to 'some intrinsic causal relationship' is a non sequitur. Periodicity or persistence in returns, even if statistically confirmed, would not by itself establish causality; it could reflect time-varying risk premia, market microstructure effects, or other equilibrium dynamics. Furthermore, the statement in Section 3 that 'hot spots corresponds to the peaks in the standard, one-dimensional time series analysis' is tautological: the wavelet power spectrum is by construction a measure of local variance, so high power regions correspond to periods of high amplitude by definition. The paper does not provide any predictive test, out-of-sample evaluation, or causal model to support the EMH-violation claim.","section":"Section 4, Conclusion"},{"comment":"The text refers to 'the period of Covid-19 peek in 2022' (Section 3). The sample window covers 2017-2022, but the major COVID-19 market crash occurred in March 2020, not 2022. This factual misdating is symptomatic of the paper's pattern-reading approach: the visual identification of 'hot spots' is not anchored to verified calendar events, which further undermines confidence that the reported features correspond to real economic episodes rather than artifacts.","section":"Section 3, Covid-19 reference"}],"minor_comments":[{"comment":"The notation f ∈ H = L2(R) is slightly misleading; f is a function in L2(R) and the wavelet ψ is in L2(C). Clarify the domains and the admissibility condition on ψ.","section":"Section 2, Eq. (2)"},{"comment":"The wavelet coherence R2(σ,τ) is defined but never used in the analysis; either remove it or explain its intended role.","section":"Section 2, Eq. (5)"},{"comment":"There are several typographical errors: 'CryproCompare' should be 'CryptoCompare', 'peek' should be 'peak', 'the study examines the daily closing prices for three major cryptocurrencies (BTC, ETH and' is followed by an incomplete listing in the narrative before the parenthetical is closed, and 'It shows that if we focus' appears multiple times. A careful proofread is needed.","section":"Throughout"},{"comment":"The abstract mentions 'analysis for the probability distributions in the space of frequency and time variables', but no probability distributions are estimated or analyzed in the paper; this claim should be removed or implemented.","section":"Abstract and Section 2"},{"comment":"The figures (presumably color maps) are not described with axes labels or colorbar scales in the text; without such details, the reader cannot judge the magnitude of the reported 'hot spots' or the range of scales shown.","section":"Figures"}],"recommendation":"reject","confidential_remarks":"The paper is far from the standards of a serious empirical journal: the central claim rests on visual inspection without any statistical significance testing, and the conceptual link from wavelet spectra to causality and EMH violations is not established. Even a major revision would require substantial new analysis (null models, surrogate tests, robustness across wavelets, and a valid predictive or causal test) that is not a local fix but a fundamental reworking of the empirical strategy. I therefore recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a standard continuous wavelet transform applied to daily log returns of BTC, ETH, XRP, and a few traditional assets, using both Morlet and complex Morlet wavelets. What's genuinely new is the observation—if the figures really show it—of a continuous line-like ridge in XRP's wavelet power spectrum. That specific feature is not in the cited literature, and it could be a real phenomenon. The paper also gives a fair survey of prior wavelet work in crypto and economics. That is the extent of the credit I can give.\n\nThe soft spots are not minor. The central claim—cyclical persistence, an 'intrinsic causal relationship,' and a weak-form EMH violation—rests entirely on visual inspection of red regions in the figures. There are no significance contours (e.g., Torrence-Compo red-noise tests), no surrogate data ensembles, no confidence intervals, no sensitivity analysis for the mother wavelet, and no out-of-sample predictive test. That is not a presentational omission: the continuous wavelet transform of any single noisy realization produces patchy high-power regions, and cone-of-influence boundary effects can create near-linear arms. Even if the XRP line is real, periodicity in returns does not by itself imply causality or predictability without a forecasting or predictive test. The text's 'Covid-19 peek in 2022' is a concrete symptom of pattern-reading without verification—the main COVID crash was 2020, and 2022 was not the pandemic's peak in any standard sense.\n\nI agree with the reader's rejection verdict. The paper is not incoherent on its own terms—it does what it says, computes wavelet spectra, and cites relevant work—but the inference from a visible line to 'intrinsic causal relationship' is a logical leap, not a circular step. The author is thinking clearly about the tool but not about the statistics of what they are seeing.\n\nFor whom is this paper? A reader who wants a quick, unfiltered look at wavelet spectra for these assets might glance at the figures, but the analysis as it stands does not establish any of its conclusions. It deserves a desk reject, not referee time. If the author adds surrogate testing, significance contours, wavelet-parameter robustness checks, and an out-of-sample predictive test, the XRP ridge could become a legitimate follow-up. As submitted, the right move is to decline it and point the author toward those additions.","headline":"A visually driven wavelet analysis whose central EMH claim about XRP rests on untested visual features; methodologically unsupported but with a kernel of an observation worth rigorous follow-up.","tokens_in":5880,"tokens_out":1766,"would_cite":false,"duration_ms":19533,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Applying a continuous wavelet transform to daily log returns, this paper reports red 'hot spots' and a line-like feature in XRP's power spectrum, which it interprets as cyclical persistence at specific investment horizons.","keywords":["wavelet analysis","continuous wavelet transform","cryptocurrency returns","XRP","efficient market hypothesis","cyclical persistence","time-frequency analysis","Morlet wavelet"],"falsifier":"Take the XRP daily log-return series, generate a large ensemble of surrogate series with identical length and marginal distribution but no temporal structure (for example by random permutation or by phase randomization that preserves the periodogram), compute the same morl and cmor1.5-1.0 wavelet power spectra, and compare the frequency of line-like hot spots in the surrogates with the observed spectrum. If equivalently strong line-like features appear in a substantial fraction of surrogates, the claimed cyclical persistence is indistinguishable from noise; a second check is to recompute with a different mother wavelet (for example Paul or Daubechies) and mask the cone of influence to see whether the XRP line survives.","tokens_in":4822,"feed_emoji":"📈","tokens_out":9128,"duration_ms":84026,"temperature":0.7,"pith_summary":"The paper tries to establish that the time-frequency distribution of daily cryptocurrency log returns carries real cyclical structure rather than noise. Using continuous wavelet transforms with Morlet and complex Morlet mother wavelets, it reports stable low-frequency bands and red hot spots in the high-frequency range, and for XRP a line-like sequence of hot spots across low-to-middle frequencies. It reads this line as evidence of a unique persistent periodicity, and the Conclusion states that such cyclical persistence implies an intrinsic causal relationship for certain investment horizons—which, if correct, violates the weak form of the efficient market hypothesis for those returns. A sympathetic reader would care because a frequency-specific repeatable cycle in daily returns would be a concrete, horizon-dependent predictability signal.","feed_headline":"Crypto wavelet scan finds persistent price cycles, strongest in XRP","feed_subtitle":"Persistent hot spots in XRP's return spectrum imply a weak-form market inefficiency a trader could exploit.","key_machinery":"The key object is the continuous wavelet transform $W_f(\\sigma,\\tau)$, applied with the Morlet ('morl') and complex Morlet ('cmor1.5-1.0') mother wavelets. The transform maps the log-return series into a time-frequency plane, and the wavelet power spectra rendered as heat maps are the evidence carrier: red hot spots mark transient high-frequency bursts, while a line-like ridge marks a persistent frequency. The XRP line-like shape is the observable that carries the causal claim, and it is the only feature the Conclusion explicitly ties to cyclical persistence.","core_discovery":"On the paper's own terms, the central discovery is that daily log-return spectra are not featureless: low frequencies stay stable over time while high frequencies show bursty red hot spots, and XRP in particular exhibits sequential points 'almost like a line' in the low-to-middle frequency range—interpreted as a unique frequency or periodicity in XRP prices. The paper then draws the logical consequence that cyclical persistence at different frequencies means there exists some intrinsic causal relationship for the investment horizons defined by the sampling scales. Because such persistence would mean past price movements carry information about future price movements at those horizons, the author states that the wavelet analysis casts doubt on the weak form of the efficient market hypothesis.","pith_inferences":["An implicit extension is to estimate the dominant period of the XRP line-like feature and test out-of-sample directional forecasts; a working forecast rule would upgrade the spectral observation into an exploitable anomaly.","A wavelet-based Hurst exponent analysis could distinguish a deterministic cycle from long memory, two different mechanisms that would produce different predictability and risk implications.","Because no significance contours are reported, surrogate testing is the natural check: phase-randomized versions of the XRP returns should not reproduce the line-like feature if the persistence is real.","The COVID hot-spot pattern suggests a testable common-shock hypothesis: the same high-frequency band should activate across cryptocurrencies at the same dates, and removing that window should weaken the persistence claim."],"forward_implications":["If the XRP line-like feature is a genuine cycle, daily XRP log returns contain a frequency component that repeats, so past returns carry information about future returns at that horizon.","Weak-form market efficiency would fail at least for XRP daily data, since the price would not fully reflect all information contained in past prices.","The high-frequency hot spots shared by all cryptocurrencies during the 2022 COVID peak indicate a common high-frequency shock, not just asset-specific noise.","The stable low-frequency band across all series implies that long-horizon behavior is comparatively smooth, so the claimed violation is specific to particular investment horizons rather than to all frequencies."],"supporting_citations":[{"why":"supplies the preceding finding of cryptocurrency price stability and the motivation to look at high-frequency variance differences across cryptocurrencies.","marker":"T. Kikuchi, T. Onishi, K. Ueda (2021)"},{"why":"provides the precedent wavelet-coherency analysis of Bitcoin, gold, commodities and stocks that the paper extends by looking at power spectra of individual assets.","marker":"E. Bouri, S.J.H. Shahzad, D. Roubaud, L. Kristoufek, B. Lucey (2020)"},{"why":"shows wavelet analysis applied to Bitcoin bubbles and structural change, an antecedent for interpreting wavelet patterns in crypto prices.","marker":"W. Fruehwirt, L. Hochfilzer, L. Weydemann, S. Roberts (2021)"},{"why":"supplies a wavelet-based Hurst estimator for high-frequency crypto returns used to reason about long memory and COVID impacts.","marker":"M.B. Arouxet, A.F. Bariviera, V.E. Pastor, V. Vampa (2022)"},{"why":"offers wavelet coherence evidence on Bitcoin and COVID-19, the temporal benchmark for the paper's COVID hot-spot observation.","marker":"J.W. Goodell, S. Goutte (2021)"},{"why":"justifies wavelet multiscaling for extracting high-frequency components and seasonality in financial market data.","marker":"R. Gencay, F. Selcuk, B. Whitcher (2001)"},{"why":"establishes the economic use of wavelet analysis to uncover transient time-frequency relations, the methodological frame for interpreting cyclical persistence.","marker":"L. Aguiar-Conraria, N. Azevedo, M.J. Soares (2008)"}],"fun_headline_variants":["Wavelet analysis spots persistent crypto cycles, strongest for XRP","XRP price cycles persist in wavelet scan","Wavelet finds XRP's persistent cycles, hinting market inefficiency","Persistent crypto cycles found via wavelet analysis, XRP stands out","Wavelet analysis casts doubt on weak market efficiency via crypto cycles"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's conclusion depends on the assumption that the red hot spots and the line-like shape in the wavelet power spectra are true cyclical persistence and not artifacts of the Morlet mother wavelet, boundary effects at the edges of the sample, or the ordinary spectrum of a noisy return process, since no significance contours, confidence intervals, surrogate data, or robustness checks are reported.","fun_headline_variants_meta":{"raw":{"variants":["Wavelet analysis spots persistent crypto cycles, strongest for XRP","XRP price cycles persist in wavelet scan","Wavelet finds XRP's persistent cycles, hinting market inefficiency","Persistent crypto cycles found via wavelet analysis, XRP stands out","Wavelet analysis casts doubt on weak market efficiency via crypto cycles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001203,"raw_usage":{"total_tokens":4895,"prompt_tokens":818,"completion_tokens":4077,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":434,"completion_tokens_details":{"reasoning_tokens":3992}},"tokens_in":434,"tokens_out":4077,"duration_ms":28924,"temperature":1.0,"reasoning_tokens":3992,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:34:40.258750+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the XRP daily log-return series, generate a large ensemble of surrogate series with identical length and marginal distribution but no temporal structure (for example by random permutation or by phase randomization that preserves the periodogram), compute the same morl and cmor1.5-1.0 wavelet power spectra, and compare the frequency of line-like hot spots in the surrogates with the observed spectrum. If equivalently strong line-like features appear in a substantial fraction of surrogates, the claimed cyclical persistence is indistinguishable from noise; a second check is to recompute with a different mother wavelet (for example Paul or Daubechies) and mask the cone of influence to see whether the XRP line survives.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the preceding finding of cryptocurrency price stability and the motivation to look at high-frequency variance differences across cryptocurrencies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the precedent wavelet-coherency analysis of Bitcoin, gold, commodities and stocks that the paper extends by looking at power spectra of individual assets."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"shows wavelet analysis applied to Bitcoin bubbles and structural change, an antecedent for interpreting wavelet patterns in crypto prices."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"establishes the economic use of wavelet analysis to uncover transient time-frequency relations, the methodological frame for interpreting cyclical persistence."}],"review_version":1}