pith. sign in

arxiv: 2606.08232 · v1 · pith:44JZ7KPXnew · submitted 2026-06-06 · 💱 q-fin.TR · q-fin.CP· q-fin.RM· q-fin.ST

Hour-Aware Adaptive Risk Management for Autonomous Memecoin Trading: A Multi-Layer Intelligence Framework

Arati Uday Kamat This is my paper

Pith reviewed 2026-06-27 18:47 UTC · model grok-4.3

classification 💱 q-fin.TR q-fin.CPq-fin.RMq-fin.ST
keywords memecoin tradingSolana DEXrisk managementhour-of-day effectsfilter precisiondrawdown avoidanceautonomous trading
0
0 comments X

The pith

A rejection-tracking system shows memecoin trading filters avoided drawdowns that hit 17.9 percent of rejected tokens within 24 hours.

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

The paper deploys an autonomous trading system on Solana and tracks both executed trades and rejected tokens over 15 days. It finds that the filter stack prevented exposure to tokens that later experienced significant drawdowns in forward observations. Hour-of-day analysis shows directional but insignificant differences in performance. The system achieves positive cumulative returns but depends heavily on a few winning trades for profitability. These results suggest the rejection criteria provide a net benefit against market outcomes despite sample limitations.

Core claim

In a 15-day paper trade of 190 memecoin trades, a parallel rejection-tracking system collected 4,874 observations where 17.9 percent of rejected tokens reached 50 percent drawdown from reference within 24 hours, and the filter stack avoided these outcomes, indicating the criteria are net-positive. A Mann-Whitney test on hour effects was not significant, and removing the top three trades reverses profitability.

What carries the argument

The multi-layer intelligence framework consisting of hour-aware adaptive risk management and a rejection-tracking system for filter evaluation.

Load-bearing premise

The small 15-day sample and 190 trades sufficiently represent the effectiveness of the filters and hour effects despite in-sample selection and high sensitivity to individual trades.

What would settle it

Observing whether the rejection criteria continue to avoid drawdowns in an extended forward sample beyond the 15 days or whether the cumulative return remains positive after excluding the top three trades.

read the original abstract

This paper measures hour-of-day effects, filter precision, fragility, and realised yield in a 15-day paper-traded deployment of an autonomous memecoin trading system on Solana decentralised exchanges. The 190-trade sample (March 29 to April 12, 2026) shows a 40.5 percent win rate, mean per-trade return of +0.62 percent, cumulative +117.7 percent (net SOL +0.039), skewness -1.21, excess kurtosis 6.61. A Mann-Whitney U test of three poorest-performing UTC hours (2, 13, 23) against the others yields U = 1,274, p = 0.22; directional but not significant at n = 190. The three hours were selected in-sample, so the comparison is exploratory, not confirmatory. A parallel counterfactual rejection-tracking system collected 4,874 forward-sample observations across 184 distinct rejection events. Of those events, 17.9 percent reached a 50 percent drawdown from reference within 24 hours; 26.0 percent of forward samples recorded the rejected token below half-reference. The filter stack avoided these realised drawdowns: evidence that the rejection criteria are net-positive against forward-market outcomes. Fragility is the principal caveat. Removing the top three trades (1.6 percent of sample) flips cumulative return unprofitable. Profitability rests on a small number of large winners and is structurally fragile. The dataset and audit script are deposited under CC-BY-4.0 (Zenodo DOI 10.5281/zenodo.20043302).

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper reports results from a 15-day paper-traded deployment of an autonomous memecoin trading system on Solana DEXes. Across 190 trades it records a 40.5% win rate, +0.62% mean return, and +117.7% cumulative return. An exploratory Mann-Whitney U test on three in-sample poorest-performing UTC hours yields U=1,274, p=0.22. Counterfactual tracking of 4,874 forward observations from 184 rejection events shows 17.9% reaching 50% drawdown within 24 h and 26% falling below half-reference; the authors conclude that the filter stack is net-positive because it avoided these outcomes. The work flags fragility (cumulative return flips negative after removing the top three trades) and deposits the dataset and audit script under CC-BY-4.0.

Significance. If the central filter-effectiveness claim can be placed on firmer statistical footing, the work supplies one of the few open, forward-sample empirical records of risk-filter performance in high-volatility memecoin markets. The deposited dataset (Zenodo DOI 10.5281/zenodo.20043302) and audit script constitute a clear reproducibility strength that future researchers can use to test alternative filters or longer windows.

major comments (3)
  1. [Abstract] Abstract: the claim that the rejection criteria are net-positive rests on separate descriptive statistics for the 190 accepted trades and the 4,874 rejected observations; no direct statistical comparison (e.g., proportion test or Mann-Whitney on drawdown incidence) between the two cohorts is reported, leaving open whether the filter systematically avoids worse outcomes or merely captured a short lucky interval.
  2. [Abstract] Abstract: the Mann-Whitney U test on hour effects (U=1,274, p=0.22) is non-significant at conventional levels and the three hours were chosen in-sample, so the hour-aware component does not yet constitute confirmatory evidence for the multi-layer framework.
  3. [Abstract] Abstract: cumulative return of +117.7% is shown to flip sign after removal of the top three trades (1.6% of the sample); because this sensitivity directly affects the profitability claim that the filter stack is net-positive, additional robustness checks (e.g., trimmed means, bootstrap, or longer out-of-sample window) are required before the result can be treated as load-bearing.
minor comments (2)
  1. [Abstract] Abstract: the 15-day window and 190-trade sample are described only by calendar dates; adding the exact number of distinct tokens and the volatility regime statistics would help readers gauge external validity.
  2. [Abstract] Abstract: skewness (-1.21) and excess kurtosis (6.61) are reported without reference to the per-trade return distribution or to any benchmark (e.g., buy-and-hold memecoin index), reducing interpretability of the risk profile.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for the constructive feedback and for noting the reproducibility strengths of the deposited dataset. We address each major comment below, indicating revisions where we can strengthen the statistical presentation while remaining honest about the constraints of the 15-day deployment.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that the rejection criteria are net-positive rests on separate descriptive statistics for the 190 accepted trades and the 4,874 rejected observations; no direct statistical comparison (e.g., proportion test or Mann-Whitney on drawdown incidence) between the two cohorts is reported, leaving open whether the filter systematically avoids worse outcomes or merely captured a short lucky interval.

    Authors: We agree that a direct statistical comparison between the accepted and rejected cohorts would provide firmer evidence. In the revised manuscript we will add a two-proportion z-test on the 50% drawdown incidence rates and a Mann-Whitney U test on the forward-sample drawdown distributions, allowing a formal assessment of whether the filters systematically avoid worse outcomes. revision: yes

  2. Referee: [Abstract] Abstract: the Mann-Whitney U test on hour effects (U=1,274, p=0.22) is non-significant at conventional levels and the three hours were chosen in-sample, so the hour-aware component does not yet constitute confirmatory evidence for the multi-layer framework.

    Authors: The manuscript already describes the test as exploratory because the hours were chosen in-sample and the p-value is non-significant. We will revise the abstract to state this limitation more explicitly so that the hour-aware layer is not read as confirmatory evidence. revision: yes

  3. Referee: [Abstract] Abstract: cumulative return of +117.7% is shown to flip sign after removal of the top three trades (1.6% of the sample); because this sensitivity directly affects the profitability claim that the filter stack is net-positive, additional robustness checks (e.g., trimmed means, bootstrap, or longer out-of-sample window) are required before the result can be treated as load-bearing.

    Authors: We already report the fragility to removal of the top three trades. We will add trimmed-mean returns and bootstrap confidence intervals around the cumulative return in the revision. A longer out-of-sample window cannot be supplied without extending the live-trading period beyond the reported 15 days. revision: partial

standing simulated objections not resolved
  • Request for a longer out-of-sample window beyond the 15-day deployment

Circularity Check

0 steps flagged

No significant circularity; purely empirical reporting

full rationale

The manuscript reports direct empirical metrics from a 15-day paper-traded sample of 190 trades and 4,874 rejection observations, including win rate, mean return, cumulative return, skewness, kurtosis, and a Mann-Whitney U test (U=1,274, p=0.22). It explicitly flags the hour selection as in-sample and exploratory. No equations, derivations, fitted parameters, predictions by construction, or self-citations appear in the text. All claims rest on observed forward-market outcomes rather than any reduction to inputs or prior author work, making the derivation chain self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

This is an empirical performance report relying on observed trading data and statistical tests rather than theoretical axioms or new entities.

axioms (2)
  • domain assumption Paper-trading simulation accurately models real market execution without significant slippage or liquidity issues.
    The study uses paper-traded deployment, assuming no real-world trading frictions affect the results.
  • domain assumption The 15-day period and 190 trades provide a representative basis for evaluating hour-of-day effects and filter performance.
    The analysis and claims rest on this short timeframe despite acknowledged fragility.

pith-pipeline@v0.9.1-grok · 5843 in / 1568 out tokens · 35746 ms · 2026-06-27T18:47:40.495595+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

9 extracted references · 1 canonical work pages

  1. [1]

    Cryptocurrency trading: A survey,

    F. Fang, C. Ventre, M. Basios, L. Kanthan, D. Martinez-Rego, F. Wu, and L. Li, "Cryptocurrency trading: A survey," Financial Innovation, vol. 8, no. 1, pp. 1–59, 2022

    2022

  2. [2]

    A theory of intraday patterns: Volume and price variability,

    A. R. Admati and P. Pfleiderer, "A theory of intraday patterns: Volume and price variability," Review of Financial Studies, vol. 1, no. 1, pp. 3–40, 1988

    1988

  3. [3]

    Flash Boys 2.0: Frontrunning in decentralized exchanges, miner extractable value, and consensus instability,

    P. Daian, S. Goldfeder, T. Kell, Y. Li, X. Zhao, I. Bentov, L. Breidenbach, and A. Juels, "Flash Boys 2.0: Frontrunning in decentralized exchanges, miner extractable value, and consensus instability," in Proc. IEEE Symposium on Security and Privacy, 2020, pp. 910–927

    2020

  4. [4]

    Trading and returns under periodic market closures,

    H. Hong and J. Wang, "Trading and returns under periodic market closures," Journal of Finance, vol. 55, no. 1, pp. 297–354, 2000

    2000

  5. [5]

    Price discovery and trading after hours,

    M. J. Barclay and T. Hendershott, "Price discovery and trading after hours," Review of Financial Studies, vol. 16, no. 4, pp. 1041–1073, 2003

    2003

  6. [6]

    Common risk factors in cryptocurrency,

    Y. Liu, A. Tsyvinski, and X. Wu, "Common risk factors in cryptocurrency," Journal of Finance, vol. 77, no. 2, pp. 1133–1177, 2022

    2022

  7. [7]

    Does reject inference really improve the performance of application scoring models?

    J. Crook and J. Banasik, "Does reject inference really improve the performance of application scoring models?" Journal of Banking and Finance, vol. 28, no. 4, pp. 857–874, 2004

    2004

  8. [8]

    The offset tree for learning with partial labels,

    A. Beygelzimer and J. Langford, "The offset tree for learning with partial labels," in Proc. 15th ACM SIGKDD, 2009, pp. 129–138

    2009

  9. [9]

    Kamat, Post-Rejection Follow-up Sampling: A Methodology for Counterfactual Outcome Measurement in Algorithmic DEX Trading (v2 2026-05-30)

    A. Kamat, Post-Rejection Follow-up Sampling: A Methodology for Counterfactual Outcome Measurement in Algorithmic DEX Trading (v2, 2026-05-30). SSRN abstract_id 6607301; Zenodo DOI 10.5281/zenodo.20043516. [Kamat 2026b] Version History 2.0 (2026-05-30): Current version. Companion dataset at Zenodo DOI 10.5281/zenodo.20043302. 1.0 (2026-04-12): Superseded. ...

    work page doi:10.5281/zenodo.20043516 2026