For external readers: This document is one of 10 curated case-study artefacts from Project Aṣe, a quantitative crypto-trading research project that produced a documented negative result with mechanism across four strategy attempts plus a literature audit, all under pre-committed criteria (Sharpe > 1.0 AND profit-factor > 1.4 OOS, called “§14” internally) that held without bar erosion across 19 ADRs. The full project repo is private (scope B of the Phase 5e shipping spec). Start with ./00-readme.md for the reading guide. Read order context: The 2-day literature audit that asked “is §14 empirically achievable on this surface at all?” — produces Criterion D. The substance of the document below is unchanged from the internal version; only this framing block and personal/identifying content scrubs are added.

Pre-Phase-5 Literature Audit

§0. Metadata

§1. Mission

Answer the binding question from Phase 5 scope §4.1:

“What Sharpe ratios and profit factors do published / open-source crypto trading strategies actually achieve on retail-accessible surfaces, particularly BTC/ETH spot long-only at various timeframes?”

The audit’s role is to provide the evidence base on which §5.1’s pre-committed decision criteria are mechanically applied. The audit does NOT propose strategies, revise criteria, or recommend bar adjustments. It surfaces what the literature documents and lets the criteria fire.

§2. Methodology

§2.1 Source priorities (operationalised per scope §4.3)

Tier Definition Qualifies for which criteria?
1 Peer-reviewed academic journal articles All criteria including A
2 SSRN / arXiv working papers from established academics; AQR / Two Sigma / institutional research All criteria including A
3 Crypto-native institutional research (Glassnode, Coinmetrics, Kaiko, Galaxy, Grayscale) with methodology Criteria B, C, D, E only
4 Open-source community evidence (Freqtrade, GitHub repos with documented OOS) Criteria B, C, D, E only
5 Practitioner blog posts Caveat-heavy, only cited as pointer

§2.2 Search execution (12 queries across two waves)

Wave 1 (surveys + institutional discovery):

  1. “cryptocurrency algorithmic trading literature review academic survey”
  2. “Bitcoin Ethereum trading strategies survey peer-reviewed Sharpe”
  3. “crypto quantitative strategy systematic backtest meta-analysis”
  4. “AQR cryptocurrency systematic research paper”
  5. “Two Sigma Renaissance Bitcoin cryptocurrency quantitative trading”
  6. “Quantpedia cryptocurrency strategy database backtest results”
  7. “technical trading cryptocurrencies Annals Operations Research Sharpe out-of-sample”
  8. “Galaxy Digital Coinmetrics institutional crypto strategy performance research”

Wave 2 (criterion-targeted): 9. “high frequency momentum trading cryptocurrencies Sharpe ratio Bianchi” 10. “cryptocurrency cross-sectional momentum Liu Tsyvinski Sharpe long only” 11. “Bitcoin daily timeframe trend following strategy Sharpe profit factor out-of-sample academic” 12. “cryptocurrency factor momentum long-only winners Sharpe 1.28 weekly rebalance” 13. “Cryptocurrency factor momentum transaction costs gross net Sharpe profit factor” 14. “cryptocurrency momentum strategy net transaction costs Sharpe profit factor friction-adjusted” 15. “Cryptocurrency momentum Grobys Shahzad illusion Sharpe spurious” 16. “Bitcoin Ethereum long only spot trend following annual Sharpe 1.0 retail accessible” 17. “cryptocurrency 4 hour timeframe trading academic Sharpe out-of-sample BTC ETH” 18. “Hudson Urquhart technical trading cryptocurrencies daily 15000 rules data”

Wave 3 (methodology / friction) was rolled into Waves 1–2 because friction-adjustment evidence surfaced naturally in the factor-momentum query thread (especially Baldi-Lanfranchi 2024 and Cryptocurrency Market Risk-Managed Momentum).

§2.3 Inclusion criteria for citation

A source is cited in this report only if at least one of the following is true:

§2.4 Discipline rules applied

Per scope §4.5:

  1. No cherry-picking. The Grobys & Shahzad dissent on crypto momentum Sharpe reliability is cited alongside the Liu/Tsyvinski/Wu / Quantitative Finance positive findings. The Hudson & Urquhart Bitcoin-OOS-fails-predictability finding is cited alongside generally positive results for other cryptocurrencies.
  2. Friction-adjusted figures preferred. Where only gross is reported, the gross figure is flagged and the source’s Tier is capped at 4 for §14-bar evaluation.
  3. IS vs OOS distinction explicit. Multiple sources report IS-only Sharpe; these are noted as such.
  4. Publication bias caveat. Section §6 of this report applies a qualitative confidence adjustment.
  5. Reddit/Twitter excluded as primary evidence. All cited sources are journals, working papers, or institutional research pages.
  6. Honest verdict. Section §7’s Decision Mapping applies the criteria mechanically; no soft-pedalling to favour a preferred path.

§2.5 Judgment call (pre-execution, recorded for auditability)

Per the open question raised before search began: if a Tier-1 paper reports §14-clearing OOS performance on daily BTC/ETH and notes “results extend to higher-frequency operation”, this audit routes that evidence to Criterion B (timeframe shift), NOT Criterion A (strict 4h). The “extends to” claim is a claim, not an OOS result on 4h itself. The operator approved this routing implicitly by saying “proceed” without challenge.

§3. Achievability Matrix

The matrix maps (timeframe × asset universe × strategy class) → Sharpe range. Empty cells with N=0 are themselves findings — they inform Criterion E.

Cell Timeframe Asset universe Strategy class Sharpe (net OOS reported) PF §14? N sources Top tier
A1 4h BTC/ETH long-only spot Trend / breakout N=0 direct evidence — — 0 —
A2 4h BTC/ETH long-only spot Mean-revert N=0 direct evidence — — 0 —
A3 4h BTC/ETH long-only spot Multi-signal / factor N=0 direct evidence — — 0 —
A4 4h BTC + alts long-only Cross-sectional N=0 direct evidence — — 0 —
B1 8h BTC/ETH long-only spot Trend / breakout N=0 direct evidence — — 0 —
B2 8h BTC + alts long-only Cross-sectional N=0 direct evidence — — 0 —
C1 Daily BTC/ETH long-only spot Trend / momentum BTC OOS: no predictability (Hudson & Urquhart 2021); intraday TS momentum gross Sharpe ~1.6 unverified net (Shen 2022) Not reported No 2 T1
C2 Daily/Weekly BTC + alts long-only Cross-sectional momentum Long-only winners Sharpe 1.28 (Cryptocurrency Factor Momentum 2023); risk-managed momentum Sharpe 1.42 (2025) Not reported in abstracts Possibly (caveats) 3 T1
D1 4h Perp futures top-N Long-only rotation N=0 direct evidence — — 0 —
D2 Daily Perp futures top-N Long-only rotation Implied by C2 (universe broader) Not reported Possibly 0 direct, 2 adjacent T1

§3.1 Reading the matrix

§4. Per-Source Citation Table

# Source Tier Year Strategy class Asset universe Timeframe Sharpe PF Friction-adj? IS/OOS Notes
1 Hudson & Urquhart, Annals of Operations Research 1 2021 Technical trading (5 classes, ~15,000 rules) 2 BTC markets + 3 other cryptocurrencies Inferred daily (not directly verified) BTC: no OOS predictability; other cryptos: positive Not reported Breakeven costs higher than market typical (yes friction-aware) Both IS and OOS reported Tier-1 peer-reviewed. The key negative finding on BTC specifically.
2 Cryptocurrency Factor Momentum, Quantitative Finance Vol 23 No 12 1 2023 Cross-sectional factor momentum, long-only winners 3,900+ coins Weekly rebalanced 1.28 annualised (long-only winners) Not reported in abstract Profitable after short-sale constraints; transaction cost treatment in body not verified 2014–2022 sample, full-sample Profitable on long-only leg specifically.
3 Liu, Tsyvinski, Wu (likely Review of Financial Studies or Journal of Finance) 1 2022 Three-factor model (market + size + momentum) Cryptocurrency universe Weekly Cross-sectional Sharpe varies; long-leg concentrated; “worsens monotonically with volume” Not reported in abstract Not specified in retrieved summaries Mixed Theoretical foundation; long-leg performance is volume-dependent — implies BTC/ETH (highest volume) get the weakest momentum effect.
4 Cryptocurrency Market Risk-Managed Momentum Strategies, ScienceDirect 1 2025 Risk-managed (volatility-targeted) momentum Cryptocurrency universe Weekly 1.42 (risk-managed); 1.12 (plain) annualised Not reported in abstract “Robust to transaction costs” (claim, not verified) Backtested with robustness checks Strongest positive case for Criterion C.
5 Grobys & Shahzad, International Journal of Finance & Economics 1 2024 Cross-sectional momentum (long-short) Cryptocurrencies Variable Sharpe mathematically undefined (claim: power-law variance) N/A N/A N/A The dissenting view: crypto momentum Sharpe is unreliable.
6 Cryptocurrency momentum has (not) its moments, Financial Markets and Portfolio Management 1 2025 Momentum Cryptocurrencies Variable Pro/con discussion of Grobys & Shahzad N/A N/A N/A Follow-up academic debate.
7 Mann, SSRN WP “Quantitative Alpha in Crypto Markets: Systematic Review” 2 2025 Meta-analysis 24+ peer-reviewed studies, 2018–2025 Various Reviews findings; identifies cross-exchange arb, factor investing, on-chain signals Various Various Both Meta-review confirming the strategy classes that show statistical alpha. Paper paywalled — only abstract verified.
8 Shen et al., Financial Review — “Bitcoin intraday time series momentum” 1 2022 Time-series momentum, intraday Bitcoin Half-hour intervals (not 4h) ~5.4× buy-and-hold (gross, derived ~2.86) Not reported Gross — friction adj not verified Both High-frequency intraday; not directly applicable to 4h cell.
9 Caferra & Vidal-Tomás, CentAUR Reading repository (working) 2 2021 Bitcoin intraday TS momentum Bitcoin Intraday Improved Sharpe vs B&H Not reported Not verified Both Working paper version.
10 Multi-timeframe trend strategy on Bitcoin (Quantpedia) 3 2024 Daily filter + hourly entries Bitcoin Multi-TF Sharpe 0.33 → 0.80 (improvement) Not reported Yes (with fees) OOS Tier 3 vendor research. Demonstrates timeframe-shift improvement but neither figure clears §14.
11 Baldi-Lanfranchi, “Transaction-cost-aware Factors” SSRN/Lancaster WP 2 2024 Factor models with TCA construction Equities + general; references crypto Various Net Sharpe collapses ~95% gross→net for high-turnover factors Not reported Yes (the paper’s whole point) N/A KEY methodological point: high-turnover factor strategies (momentum) suffer extreme gross-to-net Sharpe degradation. Applied to equities directly; implication for crypto is severe.
12 Grobys & Sapkota — Economics Letters 1 2019 Momentum trading Many cryptocurrencies Daily/weekly Negative — no positive abnormal returns N/A N/A 2014–2018 Earlier negative result on crypto momentum.
13 Korajczyk & Sadka, Journal of Finance 1 2004 Equity momentum, trading-cost robustness Equities (NOT crypto) Daily Robust to trading costs at modest position sizes; degrades at scale Reported (mostly survives) Yes OOS Reference methodology for friction-adjustment in momentum strategies.
14 Trend Following Strategies (Grayscale Research) 3 2025 Trend / momentum on Bitcoin Bitcoin Daily Trend signals manage Bitcoin volatility Not reported Generally yes (institutional context) OOS Tier 3 institutional. Suggests trend signals improve risk-adjusted outcomes for BTC but doesn’t claim §14 clearance.

§4.1 Notable absences (themselves findings)

§5. Findings by §5.1 Criterion

§5.1 Criterion A — §14 achievable on BTC/ETH 4h spot long-only

Requirement (verbatim): “≥ 2 independent published studies (peer-reviewed or established quant firms) demonstrating Sharpe > 1.0 AND PF > 1.4 on BTC/ETH 4h spot long-only, with reasonable friction adjustments and out-of-sample evidence.”

Evidence found: Zero qualifying studies. Matrix cells A1–A4 are all empty. The most directly relevant study (Hudson & Urquhart 2021, Tier 1, Annals of Operations Research) — a comprehensive test of ~15,000 technical trading rules on Bitcoin and four other cryptocurrencies — finds no out-of-sample predictability for Bitcoin specifically. Predictability does persist for other cryptocurrencies in their sample, but BTC is the most-relevant asset for Criterion A and the result is decisive.

Verdict: Criterion A is NOT satisfied.

§5.2 Criterion B — Timeframe shift materially improves achievability

Requirement: “Multiple studies showing Sharpe improvement on 8h, 12h, or daily timeframes for spot long-only strategies on majors.”

Evidence found:

The pattern: timeframe shift toward daily/weekly does NOT clearly improve §14-cleared performance on single-asset BTC/ETH long-only. Improvements exist in directional sense, but no Tier 1/2 study on majors long-only at 8h/12h/daily clears Sharpe > 1.0 AND PF > 1.4 OOS net of fees.

Verdict: Criterion B has PARTIAL evidence (timeframe shift improves Sharpe directionally) but does NOT meet the threshold of “multiple studies showing Sharpe > 1.0 net OOS on majors long-only at 8h+ timeframes.”

§5.3 Criterion C — Cross-sectional approaches show stronger evidence

Requirement: “Published evidence that long-only cross-sectional ranking strategies across N perps achieve §14, with mechanism reasonably attributable to the cross-sectional structure rather than long-short asymmetry.”

Evidence found:

Counter-evidence:

Unverified critical detail: Profit factor (Criterion C requires PF > 1.4 alongside Sharpe > 1.0). I could not find PF figures in any abstract or open-text source for the cross-sectional momentum papers. For a Sharpe-1.28-to-1.42 long-only momentum strategy, PF would plausibly exceed 1.4 — but this is interpretation, not citation. Criterion C’s PF requirement is therefore unverified, not confirmed satisfied.

Friction adjustment: “Robust to transaction costs” appears in the 2025 risk-managed paper’s abstract but the magnitude of friction tested was not retrieved. Baldi-Lanfranchi 2024 (Tier 2) shows that high-turnover factor strategies (specifically momentum) can suffer up to 95% gross-to-net Sharpe degradation in equities — a methodological warning that may apply to crypto momentum if turnover is high. Weekly rebalancing on 3,900 coins likely implies HIGH turnover. This is a substantive caveat.

Verdict: Criterion C has STRONG SHARPE EVIDENCE for long-only cross-sectional crypto momentum (Sharpe 1.12–1.42 across multiple Tier 1 sources, mechanism cross-sectional rather than long-short) but PF unverified and friction-adjustment magnitude unverified. Dissenting Tier 1 view exists. This is partial-positive evidence, not unambiguous §14 clearance.

§5.4 Criterion D — Mixed or inconclusive evidence

Requirement: “Cells in the matrix show partial evidence but no clear winner.”

Evidence: Yes — the matrix has:

Verdict: Criterion D’s threshold IS met. No criterion fires cleanly; multiple cells have partial evidence pointing in different directions.

§5.5 Criterion E — No published evidence §14 achievable on retail-accessible long-only crypto spot

Requirement: “Multiple studies surveyed across timeframes and asset counts, none demonstrating §14 on long-only crypto spot.”

Evidence: Criterion E is NOT triggered. Criterion C has positive Sharpe evidence (1.12–1.42 across multiple Tier 1 sources) for long-only cross-sectional momentum. Although unverified for PF and full friction magnitude, this is “positive evidence” sufficient to clear Criterion E’s “no evidence anywhere” threshold.

Verdict: Criterion E is NOT triggered. This is a positive finding — the literature is not unanimously negative on retail-accessible long-only crypto strategies.

§5.6 Verification update (open-source second pass, 2026-05-12 post-operator-review)

The operator raised three substantive caveats on the v1 audit: (a) 2-hour-vs-2-day compression made abstract-summaries load-bearing, (b) PF unverified across all Criterion C sources, (c) Path D’s top-N perps universe doesn’t match the literature’s 3,900-coin universe. They authorised a 1-2 day open-source verification pass before locking the decision. This subsection records what verification revealed.

Sources where verification refined the v1 finding:

  1. Cryptocurrency Factor Momentum (Fieberg, Liedtke, Metko, Zaremba, Quantitative Finance 2023):
    • Authors confirmed (Tier 1 attribution solid).
    • Sample: 3,900+ coins, 2014–2022, weekly rebalancing.
    • Headline Sharpe 1.28 verified as the long-only winner-portfolio figure under one specification.
    • NEW FINDING: the paper explicitly states “the momentum effect prevails in larger cryptocurrencies but incurs substantial trading costs and extracts alphas largely from short positions” (in the long-short variant). This directly contradicts the v1 audit’s reading that long-only winners cleanly clear §14. The gross Sharpe 1.28 figure faces significant friction degradation that the paper itself flags. Net Sharpe is not retrievable from open sources but is materially below 1.28.
    • PF remains unverified.
  2. Cryptocurrency Market Risk-Managed Momentum (ScienceDirect 2025):
    • Headline Sharpe 1.42 (risk-managed) and 1.12 (plain) verified.
    • NEW FINDING: while the paper claims robustness to transaction costs, it also acknowledges “risk management does not change the tail risk of cryptocurrency momentum” and “the strategy is subject to considerable uncertainty that implies greater risk than previous cryptocurrency research has suggested.” These are internal caveats that align with the Grobys & Shahzad dissent on Sharpe applicability rather than contradicting it.
    • Friction magnitude tested still unverified.
    • PF still unverified.
  3. Liu, Tsyvinski, Wu — Common Risk Factors in Cryptocurrency (Journal of Finance 2022):
    • CRITICAL NEW FINDING — Sharpe is specification-dependent. At the 1-week horizon, the top quintile portfolio yields weekly return 11.22% with Sharpe 0.45 (NOT 1.28). The 1.28 figure from the Fieberg et al paper is one specification’s result; the same general approach in Liu/Tsyvinski/Wu gives 0.45 for the same direction at 1-week horizon. Whether crypto momentum clears Sharpe > 1.0 depends materially on formation period, holding period, and rebalancing frequency.
    • This means the v1 audit’s “Sharpe 1.12 to 1.42 across multiple Tier 1 sources” framing was misleading. The range across specifications is more like 0.45 to 1.42, with 0.45 being a Tier-1 published figure on cross-sectional crypto momentum that the v1 audit did not cite.
    • The long-leg-degrades-with-volume finding stands. Top-N perps would land at the weakest part of the momentum cross-section.

  4. Hudson & Urquhart (2021): verification confirms 14,919 trading rules across 2010–2017 sample period. Timeframe still inferred daily but consistent with the sample range and rule count. The headline finding (no BTC OOS predictability) is unchanged.

  5. Baldi-Lanfranchi (2024): verified that the 95% gross→net Sharpe degradation finding was on equity factor momentum, not crypto. Applies as a methodological warning to crypto momentum given similar high-turnover profile, but is not direct crypto evidence.

Net effect of verification on the matrix:

Cell C2 (Daily/Weekly cross-sectional long-only) in §3 should be re-read as:

Metric v1 audit reading Post-verification reading
Sharpe range (gross) 1.12 to 1.42 across “multiple sources” 0.45 to 1.42 across specifications within Tier 1 sources
Net of fees Robustness claimed Friction degradation acknowledged by sources themselves (“substantial trading costs” — Fieberg et al)
PF Unverified, assumed plausible Unverified, and sources’ internal caveats imply PF reliability is itself questioned
Tail risk Not discussed Sources acknowledge tail risk + “considerable uncertainty… greater risk than previous research has suggested”
Methodological dissent One Tier-1 dissent (Grobys & Shahzad) Same dissent, now reinforced by the positive papers’ own caveats

Refined verdict on Criterion C (post-verification):

Under the strictest reading of Criterion C’s evidence threshold — “published evidence that long-only cross-sectional ranking strategies across N perps achieve §14” where §14 = Sharpe > 1.0 AND PF > 1.4 OOS net of fees — the literature does NOT provide evidence sufficient to fire Criterion C. The Sharpe-passing claims are:

The v1 audit’s “partial-positive” framing of Criterion C is charitable. A stricter reading lands closer to “Sharpe yes in some specifications, PF unknown, friction degradation acknowledged, tail risk live, metric itself contested” — which is less Criterion D and more Criterion E than the v1 audit suggested.

Refined verdict on Criterion D vs E (post-verification):

However: the strength of Criterion C’s evidence is materially weaker than the v1 audit’s matrix suggested. The path-forward logic should reflect this:

This is what the operator requested in their recommendation 2: “Pre-commit a tighter Path D gate now, before Path C runs.” This subsection IS that pre-commit. The Path D scope document, when written, must include the three gates above as explicit pre-conditions.

Standard caveats apply with calibration for the audit’s findings:

  1. Successful strategies are over-represented. Papers reporting positive Sharpe ratios are more publishable than papers reporting negative results. The negative-evidence findings in this audit (Hudson & Urquhart on BTC OOS; Grobys & Shahzad on Sharpe undefinability) carry disproportionate weight because negative findings clear a higher publication bar.

  2. Many published Sharpe figures are gross, not net. Where friction-adjustment is not explicit in the abstract, the realized net Sharpe is likely 30–95% lower (per Baldi-Lanfranchi 2024’s equity finding, methodologically applicable to high-turnover crypto strategies). Cross-sectional momentum’s gross Sharpe 1.28–1.42 should be considered an upper bound; net Sharpe could easily be 0.6–0.9.

  3. The cited sources skew toward 2018–2025 publications and BTC-heavy data. Pre-2018 backtests may not generalise (small universe, different regime). Post-2022 OOS holdout is rare in the cited sample.

  4. “Robust to transaction costs” claims in abstracts are almost always under-specified. When a paper claims robustness without specifying the friction magnitude tested, the claim has weak qualitative force.

  5. The author of this audit ran three strategies (MFD, BMR, FCMFD) on the BTC/ETH 4h spot long-only surface and observed three §14 failures. This is a sample of 3 from the audited surface and the failures are direct evidence that this surface is empirically hard — independent of literature claims. The literature audit and the operator’s own trajectory are mutually reinforcing.

Net effect of publication bias on this audit’s verdict: It tilts evidence toward false-positive Criterion C clearance (positive results are over-published). Mechanically applied caveats land Criterion C in “partial-positive” rather than “clearly clears §14” territory, which is the correct calibration.

§7. Decision Mapping

Applying the pre-committed §5.1 criteria mechanically to the §5 findings:

Criterion A check (strict 4h BTC/ETH long-only spot):
  Required: ≥ 2 Tier 1/2 studies, Sharpe > 1.0 AND PF > 1.4 OOS friction-adj
  Found: 0 qualifying studies.
  Result: FAIL → A does NOT fire.

Criterion B check (timeframe shift on 8h/12h/daily for majors long-only):
  Required: multiple studies showing Sharpe IMPROVEMENT on 8h+ for BTC/ETH long-only
  Found: directional improvement (0.33 → 0.80) but no Tier 1/2 study clears §14
        on majors long-only spot at any non-4h timeframe.
  Result: PARTIAL but does NOT meet "multiple studies showing improvement [to §14]"
          threshold. → B does NOT fire cleanly.

Criterion C check (long-only cross-sectional perps achieving §14):
  Required: published evidence + mechanism cross-sectional (not long-short asymmetry)
  Found: Sharpe 1.12–1.42 across 3+ Tier 1 sources on long-only winners.
        Mechanism is cross-sectional (verified).
        Profit factor UNVERIFIED in abstracts (Criterion C requires PF > 1.4).
        Friction-adjustment claimed but magnitude unverified.
        Dissenting Tier 1 view exists (Grobys & Shahzad).
  Result: PARTIAL-POSITIVE but PF/friction unverified. → C does NOT fire
          unambiguously per the strict reading of "achieves §14".

Criterion D check (mixed/inconclusive across A/B/C):
  Required: cells with partial evidence, no clean winner.
  Found: A FAIL, B partial, C partial-positive. → D FIRES.

Criterion E check (no evidence anywhere):
  Found: C has positive Sharpe evidence, so E does NOT fire.

“Path C first (smaller change, cheaper test). If C fails its own SIA or §14, escalate to Path D.”

Path C — universe revision to 8h timeframe; re-run MFD on 8h data as cheap test of the timeframe hypothesis. Per scope §6 Path C: 1–2 weeks estimated duration.

§7.2 Why this routing is honest given the evidence

§7.4 Path D escalation gates (pre-committed post-verification)

The §5.6 verification update revealed that the v1 audit’s “partial-positive Criterion C” framing was charitable. Cross-sectional crypto momentum Sharpe is specification-dependent (range 0.45–1.42 within Tier 1 sources), gross-of-fees with sources themselves acknowledging substantial trading costs, PF universally unverified, and methodologically contested. To prevent the Path C → Path D escalation from committing 4–8 weeks on weaker-than-claimed evidence, the following gates are pre-committed in this audit and must be satisfied before any Path D scope document is written:

Gate 1 — Full-text PF verification. The Fieberg/Liedtke/Metko/Zaremba 2023 Quantitative Finance paper must be obtained in full text. Profit factor for the long-only winner portfolio (or its closest reported analogue) must be extracted. If PF < 1.4 net of fees, Criterion C strictly fails on the AND-clause of §14, and the decision re-routes to Criterion E (Path E), not Path D.

Gate 2 — Universe match. The Path D scope as currently drafted (top-N liquid perps, BTC/ETH/SOL/BNB/XRP minimum) does not match the literature’s evidentiary universe (3,900+ coins). Either: (a) The Path D scope is expanded to a top-20 or top-50 perp universe with explicit liquidity gates (matching better, larger build), OR (b) Full-text retrieval of the Fieberg et al / Liu-Tsyvinski-Wu papers confirms the cross-sectional momentum effect persists when restricted to top-10 by volume (matching Path D’s scope). Without one of these, Path D’s empirical foundation is inadequate.

Gate 3 — Net-of-fees Sharpe verification. The Cryptocurrency Market Risk-Managed Momentum 2025 paper’s “robust to transaction costs” claim must be quantified. Specifically: what friction magnitude was tested (1 bp? 10 bp? 50 bp?), and what is the actual net Sharpe at realistic crypto retail-accessible fees (typically 10–20 bp round-trip on Binance spot)? If the paper tested friction at < 10 bp, the result does not generalise to retail Binance and Path D’s foundation weakens further.

Gate 4 — Methodological dissent resolution. The Grobys & Shahzad (2024 IJFE) finding that crypto momentum Sharpe is mathematically undefined due to power-law variance must be addressed. Either the Path D plan acknowledges that the strategy is built on a metric the literature itself contests, OR the plan substitutes a metric that doesn’t require finite second moments (e.g., Calmar, Sterling, max-drawdown-relative measures). The first option is acceptable as long as it is explicit; the second is preferable.

If gates 1–4 cannot be satisfied: Path D is not authorised, and the decision re-routes to Path E (stop in-universe, ship the SIA framework as standalone infrastructure). This is bar erosion in reverse — closing a possible path because the evidence supporting it didn’t hold up under scrutiny, not because we lowered the bar to dismiss it.

These gates are the operationalised version of the operator’s recommendation 2 (“Pre-commit a tighter Path D gate now, before Path C runs”). They become binding constraints on any future Path D scope document.

§8. References

Tier 1 — Peer-reviewed academic journals

  1. Hudson, R., & Urquhart, A. (2021). Technical trading and cryptocurrencies. Annals of Operations Research, 297(1), 191–220. https://link.springer.com/article/10.1007/s10479-019-03357-1 [paywalled; open version: https://centaur.reading.ac.uk/85715/8/Hudson-Urquhart2019_Article_TechnicalTradingAndCryptocurre.pdf]
  2. Cryptocurrency factor momentum. (2023). Quantitative Finance, 23(12), 1853–1869. https://www.tandfonline.com/doi/abs/10.1080/14697688.2023.2269999
  3. Cryptocurrency market risk-managed momentum strategies. (2025). Finance Research Letters. https://www.sciencedirect.com/science/article/abs/pii/S1544612325011377
  4. Grobys, K., & Shahzad, S. J. H. (2024). Cryptocurrency momentum: Is it an illusion? International Journal of Finance & Economics. https://onlinelibrary.wiley.com/doi/10.1002/ijfe.70036
  5. Cryptocurrency momentum has (not) its moments. (2025). Financial Markets and Portfolio Management. https://link.springer.com/article/10.1007/s11408-025-00474-9
  6. Liu, Y., Tsyvinski, A., & Wu, X. (2022). Common Risk Factors in Cryptocurrency. SSRN. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3379131
  7. Shen, D., et al. (2022). Bitcoin intraday time series momentum. Financial Review. https://onlinelibrary.wiley.com/doi/10.1111/fire.12290
  8. Grobys, K., & Sapkota, N. (2019). Cryptocurrencies and momentum. Economics Letters, 180, 6–10. https://ideas.repec.org/a/eee/ecolet/v180y2019icp6-10.html

Tier 2 — Working papers from established academics / institutions

  1. Mann, W. (2025). Quantitative Alpha in Crypto Markets: A Systematic Review of Factor Models, Arbitrage Strategies, and Machine Learning Applications. SSRN. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5225612 [paywalled; LinkedIn summary at https://www.linkedin.com/posts/william-mann-cfa_quantitative-alpha-in-crypto-markets-a-systematic-activity-7321138680465666051-Jv9l]
  2. Baldi-Lanfranchi, F. (2024). Transaction-cost-aware Factors. SSRN/Lancaster Univ. WP. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4737166

Tier 3 — Institutional / vendor research

  1. Grayscale Research. (2025). The Trend is Your Friend: Managing Bitcoin’s Volatility with Momentum Signals. https://research.grayscale.com/reports/the-trend-is-your-friend-managing-bitcoins-volatility-with-momentum-signals
  2. Quantpedia. (2024). How to Design a Simple Multi-Timeframe Trend Strategy on Bitcoin. https://quantpedia.com/how-to-design-a-simple-multi-timeframe-trend-strategy-on-bitcoin/
  3. Two Sigma. Risk Analysis of Crypto Assets. https://www.twosigma.com/articles/risk-analysis-of-crypto-assets/

Tier 4 — Open-source / community

  1. (Searched but no documented forward-test results meeting the audit’s citation threshold were identified.)

Tier 5 — Practitioner blogs

  1. (Excluded by audit discipline §4.5.5; used only as pointers to find Tier 1–3 sources.)

Notable absences (themselves findings)

§9. Summary one-liner

Criterion D fires. The §5.1 decision routes to Path C first (re-run MFD on 8h as cheap timeframe-shift test). Path D escalation requires satisfying the four post-verification gates in §7.4 — failing which the decision re-routes to Path E (stop in-universe, ship the SIA framework). ADR-0017 documents this decision next.

End of audit.