Pith. sign in

REVIEW 4 major objections 4 minor 96 references

AlphaG-OPD: Reliability-Gated Sibling Counterfactuals for On-Policy Distillation in Symbolic Alpha Factor Discovery

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that reliability-gated sibling counterfactuals turn terminal factor scores into local action guidance for GFlowNet alpha discovery, improving results under a fixed evaluator budget.

desk verdict A genuinely new local-credit mechanism for GFlowNet alpha search, clearly derived and honestly scoped; the empirical edge is real but thinner than the abstract claims, and the reliability gate needs fresh-suffix validation before I would bet on it. read the letter →

arxiv 2608.01303 v1 pith:2XUFY5JN submitted 2026-08-02 cs.LG

Yaoyu Su This is my paper · ORCID
classification cs.LG
keywords symbolicalphafactordiscoveryGFlowNetsTrajectoryBalanceon-policydistillationcounterfactualcreditassignmentreliabilitygatingquantitativefinancereinforcementlearning
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

Symbolic alpha factor discovery can score a finished expression but cannot say which construction decision caused the score. This paper claims that terminal evaluations can be converted into local action guidance by comparing three grammar-valid sibling actions at partial abstract-syntax-tree states that the search policy actually visits. The comparison is trusted only when four matched completion suffixes agree on a winner and the estimated advantage has a positive lower confidence bound, and accepted guidance is capped by KL radius, replay expiry, and gradient balancing. Under a fixed 10,000 physical-score budget, the full method improves factor quality on CSI300, CSI500, CSI1000, and S&P 500 relative to the structure-aware GFlowNet baseline, without changing the terminal reward, backward policy, grammar, or factor-pool rules.

What carries the argument

The load-bearing mechanism is the reliability-gated paired counterfactual teacher. Each accepted teacher is built from a 4×3 matched completion matrix with credits $C_{ki} = \log R(x_{ki}) - \log q(u_k | s, A_s)$, so the shared suffix proposal cancels and the row comparison depends only on sibling rewards under identical continuations. An exponential tilt of the current sibling distribution is anchored with a target KL radius $\delta = 0.03$, and the teacher is admitted only when winner agreement $\gamma \ge 0.75$ and the empirical lower confidence bound $\mathrm{LCB}(s) = \bar{\Delta} - z\,\mathrm{SE}(\Delta)$ is positive; accepted rows are replayed under a score-indexed expiry and scaled by a measured gradient ratio so the auxiliary update stays bounded relative to Entropy-TB.

What would settle it

Take the accepted sibling rows from a trained run, draw many fresh grammar-valid suffixes, say 50, for each row, compute the sibling with the highest mean log terminal reward under those suffixes, and check how often it matches the sibling that the paper's gate preferred; if the agreement rate is not clearly above chance, the K=4 reliability gate is not identifying genuinely better actions.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that Trajectory Balance's trajectory-level credit can be profitably augmented with a local, reliability-gated distillation channel that supplies missing action-level labels. At a partial AST state sampled by the current forward policy, the method forms a sibling triple, evaluates all three siblings under four shared completion suffixes, and builds a KL-bounded target distribution only when at least three suffixes pick the same winner and the paired-improvement lower confidence bound is positive. The auxiliary forward-KL update then reallocates probability among those siblings while the terminal reward, backward policy, grammar, and factor-pool rules stay untouched. The paper claims this turns sparse terminal feedback into reusable structural knowledge and that the resulting factors transfer to portfolio metrics across multiple markets and random seeds.

Load-bearing premise

The entire benefit rests on the assumption that four shared random completions are enough to tell which of three sibling construction steps is genuinely better; if those four suffixes are unrepresentative of the completions the policy will meet, the accepted teacher is noise and the KL update can distort the search.

Editorial extensions

If this is right

  • The paired teaching intervention alone, before reliability gating, already improves over the base model on the CSI500 ablation, so local sibling comparisons carry information beyond the trajectory-level objective.
  • Adding the reliability gate shifts gains toward decision-relevant metrics, improving RankIC, return, Sharpe, and maximum drawdown over the un-gated paired teacher.
  • Adding bounded consolidation improves all seven reported ablation metrics over the gated one-use version, so persistence and strength control matter.
  • Because none of the terminal reward, backward policy, grammar, or pool rules change, the method can be attached to an existing GFlowNet factor miner as an additive training channel.

Reading between the lines

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

  • A natural extension the paper does not test is whether the same matched-sibling comparison transfers to other GFlowNet program synthesis settings, such as symbolic regression or molecule generation, where terminal reward similarly arrives only at completion.
  • The K=4 suffix count is the main hidden dial: the reliability gate's agreement and LCB statistics are computed from exactly four completions, and a sensitivity sweep over K would reveal whether the method depends on that specific number.
  • Because replay expiry is measured in physical score calls, one could vary the 1,000-score lifetime to probe the trade-off between freshness and credit persistence without changing any other component.
  • Held-out suffix verification could turn the gate's internal confidence statistic into a calibrated test: accepted rows should have their preferred sibling confirmed under fresh suffixes at a rate matching the gate's confidence.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper introduces AlphaG-OPD, an on-policy distillation method layered on top of an Entropy-TB GFlowNet for symbolic alpha factor discovery. The method has three components: a structural interface that selects grammar-valid sibling actions at partial AST states visited by the current policy; a reliability-gated paired counterfactual teacher that evaluates three siblings under four shared suffixes, admits a KL-bounded target only if winner agreement and a positive empirical lower confidence bound are satisfied; and a bounded consolidation mechanism that stores accepted targets in a replay queue with score-indexed expiry and scales the auxiliary gradient relative to the Entropy-TB gradient. Experiments on CSI300, CSI500, CSI1000, and S&P 500 compare the method against AlphaSAGE and other baselines, and an equal-physical-score ablation on CSI500 claims to isolate the contributions of paired teaching, reliability gating, and consolidation. The paper includes a detailed appendix with derivations, physical-score accounting, implementation configuration, and metric definitions.

Significance. If the empirical claims were fully supported, the paper would make a useful contribution: it offers a budget-aware mechanism for converting terminal factor rewards into local structural credit for GFlowNet search, with a clear separation of where, what, and how long to teach. The manuscript is strong on transparency: the derivations in the appendix are internally consistent, the physical-score ledger is explicit, and the implementation details (optimizer, architecture, hyperparameters, data splits, metric formulas) are sufficiently complete to make reproduction feasible. The central claim, however, is empirical, and the current evidence is underpowered: only three seeds are used, no significance tests are reported, the main comparison uses literature-reference numbers rather than matched re-runs, and the reliability gate is evaluated on the same suffixes used to construct the teacher. These issues are load-bearing for the paper's main message, and they require additional analysis rather than simple presentation fixes.

major comments (4)
  1. [Component II, Eqs. (5)-(7)] The reliability gate is computed on the same K=4 suffixes that are used to construct the teacher, so a passing LCB demonstrates consistency on those four completions, not generalization to the suffix distribution the policy will encounter. With K=4, z=1, and gamma_min=0.75, the gate is permissive: the standard error in Eq. (7) has only three degrees of freedom, and at least three of four suffixes must agree. The manuscript explicitly states that 'the main method uses no held-out verification suffixes,' so this is a deliberate design choice; however, the central claim that the gate selects reliable local preferences needs support. Please report the acceptance rate, the distribution of LCB values, and a sensitivity analysis over K or a held-out-suffix check showing that accepted teachers reproduce under fresh completions.
  2. [Experiments, Table 2] The additive ablation does not cleanly support the claimed benefit of reliability gating. In Table 2, Components I-II (Gate On) has lower IC than Components I-II (Gate Off): 3.730 versus 4.040. The paper says the gate 'shifts the gain toward decision-relevant metrics,' but no standard deviations or significance tests are reported for Table 2, so this shift could reflect noise. The text also notes that Gate Off already improves over Base, which leaves open the possibility that part of the gain comes from the different score allocation or from probe-induced exploration rather than from gated guidance. Please provide per-seed results for the ablation, significance tests or confidence intervals, and acceptance statistics for the gated arm.
  3. [Experiments, Table 1 and Results and Analysis] The cross-market improvements over AlphaSAGE are not statistically supported. Only three random seeds are used, and the AlphaSAGE rows are literature-reference values rather than matched re-runs under the same 10K physical-score budget and protocol. On CSI1000, AlphaG-OPD's IC is lower than the reference (5.89 versus 6.04), and on S&P 500 its IC is lower (0.56 versus 0.89). The blanket statement of 'strong cross-market performance' is therefore stronger than Table 1 warrants. Please provide matched re-runs of the GFlowNet baselines under the same budget, paired significance tests across seeds, or explicitly qualify the claims on the mixed CSI1000 and S&P 500 results.
  4. [Component III, Eq. (10)] The gradient-balancing rule fixes rho=0.10, lambda_max=1e4, replay capacity 512, and score TTL 1000 without sensitivity analysis. Because Component III is credited with a large improvement in Table 2 (IC moving from 3.730 to 6.018), the result depends on these constants. The measured reference gradient g_ETB also includes the entropy-bonus contribution, which is a design choice but not the only possible one. Please add a sensitivity study over rho, replay TTL, and capacity, or otherwise justify these fixed values as robust rather than tuned.
minor comments (4)
  1. [Abstract and references] The abstract and several references have missing spaces, e.g., 'InInternationalConference' in the Biggio reference and similar spacing issues elsewhere; a full copyedit would improve readability.
  2. [Table 1 caption] The caption says 'bold marks the better completed result within each matched GFlowNet pair,' but the AlphaSAGE rows are literature-reference values, not matched re-runs; the word 'completed' also appears to be a typo for 'computed.'
  3. [Figure 3] Panel (B) reports '+45%, +90%, +150%' improvements without showing per-seed scatter or confidence intervals; adding these would make the figure more informative.
  4. [Appendix G] The phrase 'positive maximum drawdown magnitude' is odd; maximum drawdown is already a positive magnitude in the reported convention, and the wording should be simplified.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the OPD teacher is explicitly self-generated from the same terminal reward, and the paper's claims rest on external benchmarks and controlled ablations rather than on a self-referential derivation.

full rationale

The paper's central derivation is empirical and self-contained. The OPD teacher is deliberately constructed from the same terminal reward R(x) used by Trajectory Balance; this is the announced mechanism ('turns terminal factor evaluations into local action guidance'), not a hidden re-import of the conclusion. The reliability gate is computed from the same four suffixes as the teacher, but the paper explicitly labels it an 'empirical abstention rule over four paired construction samples, not a 95% confidence guarantee' and states that 'the main method uses no held-out verification suffixes.' No claim is made that the gate provides independent out-of-sample validation, so the shared-suffix design is a stated limitation rather than a circular derivation. The auxiliary objective (Eq. 9) is a forward KL bounded against Entropy-TB by Eq. (10), and the reported gains are supported by multi-seed comparisons on four universes and an equal-physical-score ablation ladder. No load-bearing self-citation, imported uniqueness theorem, or renamed known result appears in the derivation chain. The self-referential aspects of the teacher and gate are disclosed and do not reduce the central empirical claim to its inputs by construction.

Assumptions & free parameters 11 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical entities. It relies on standard GFlowNet/RL machinery and a set of hand-picked hyperparameters. The main unverified background assumptions are the representativeness of the four-suffix paired evaluation and the validity of IC as a training reward.

free parameters (11)
  • K (shared suffixes) = 4
    Number of matched completions per probe; directly determines the sample size for agreement and LCB computation.
  • delta (target KL radius) = 0.03
    KL trust region for the anchored teacher; fixed without sensitivity analysis.
  • gamma_min (winner agreement threshold) = 0.75
    Requires at least 3 of 4 suffixes to agree on the preferred sibling.
  • z (LCB multiplier) = 1.0
    Multiplier for the standard error in the lower confidence bound; not a 95% confidence level.
  • rho (target OPD-to-TB gradient ratio) = 0.10
    Scales the auxiliary gradient relative to Entropy-TB.
  • b_exit (exit logit bias) = 1.5
    Bias added to the exit action in the shared suffix proposal.
  • budget split (ordinary/probe scores) = 8008/1992
    Allocates 166 probes of 12 scores each within the 10K budget.
  • replay TTL = 1000
    Row expiration age in physical score calls.
  • replay capacity = 512
    Maximum number of active rows in the replay queue.
  • eta_H (entropy coefficient) = 0.01
    Entropy bonus weight in the Entropy-TB objective.
  • lambda_max (gradient cap) = 1e4
    Upper bound for the auxiliary loss scaling factor.
assumptions (5)
  • standard math Trajectory Balance objective with learned partition function Z is a valid global learning objective for GFlowNets.
    The base algorithm relies on TB (Eq. 1) as the global search objective, unchanged from Malkin et al. 2022.
  • domain assumption The mean cross-sectional Pearson IC on training data, R(x)=max(|IC|, exp(-10)), is a meaningful reward for factor quality.
    The reward is used both for TB and for the OPD teacher; if IC on train is a poor proxy for out-of-sample factor quality, the local guidance is misdirected.
  • domain assumption The four shared suffixes sampled from the reward-blind proposal are representative enough to evaluate sibling actions.
    Component II builds both the teacher and its gate from the same K=4 suffixes; no held-out suffixes are used.
  • standard math The proposal term q(u_k|s,A_s) cancels in all quantities, so the teacher depends only on reward differences.
    This is shown in Eq. C.7; the row winner and Delta_k are invariant to the proposal probabilities.
  • domain assumption The on-policy state distribution from the evolving forward policy is the appropriate place to provide local supervision.
    The method restricts teaching to states visited by the current policy, assuming this avoids distribution mismatch and yields useful credit.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AlphaG-OPD: Reliability-Gated Sibling Counterfactuals for On-Policy Distillation in Symbolic Alpha Factor Discovery." pith.science (2026). https://pith.science/paper/2XUFY5JN

@misc{pith2026260801303,
  author       = {Pith},
  title        = {Pith review of: AlphaG-OPD: Reliability-Gated Sibling Counterfactuals for On-Policy Distillation in Symbolic Alpha Factor Discovery},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2XUFY5JN}},
  note         = {Machine review of arXiv:2608.01303}
}
read the original abstract

Symbolic alpha factor discovery can score a completed expression, but it provides no direct label for the structural decisions that produced it. Generative flow networks (GFlowNets) preserve a diverse, reward-proportional distribution over complete expressions, yet their trajectory-level objective does not compare unchosen sibling actions at an intermediate state. We introduce AlphaG-OPD, a structural on-policy distillation framework that turns terminal factor evaluations into local action guidance. Its design separates three decisions. Component I determines where to teach by exposing grammar-valid siblings at partial abstract-syntax-tree (AST) states visited by the current forward policy. Component II determines what is reliable enough to teach: it evaluates three supported siblings under four shared suffixes and admits a KL-bounded target only when their matched comparisons exhibit sufficient winner agreement and a positive empirical lower confidence bound (LCB). Component III determines how strongly and for how long to teach by consolidating accepted targets through bounded replay, score-indexed expiry, and forward-gradient balancing, without additional factor evaluations. Terminal reward, Trajectory Balance, the backward policy, grammar, and factor-pool rules remain unchanged. An equal-physical-score four-arm ablation tests paired teaching, reliability gating, and consolidation. Across China's CSI300, CSI500, and CSI1000 and the U.S. S&P 500, the complete method delivers strong cross-market performance over multiple random seeds.

Figures

Figures reproduced from arXiv: 2608.01303 by the authors.

Figure 1
Figure 1. Trajectory-level TB and local OPD. The conceptual [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of AlphaG-OPD through its three questions. Component I determines where OPD may teach by exposing supported structural actions at an on-policy partial AST. Component II determines what is reliable enough to teach by comparing siblings under four common suffixes. Component III controls how strongly and for how long accepted credit persists through bounded replay and measured forward-gradient balancing. All v… view at source ↗
Figure 3
Figure 3. CSI500 direction transfer and frozen OOS qual [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

96 extracted references · 27 canonical work pages

  1. [1]

    arXiv preprint arXiv:2509.25055 , year=

    AlphaSAGE: Structure-Aware Alpha Mining via GFlowNets for Robust Exploration , author=. arXiv preprint arXiv:2509.25055 , year=

    arXiv

  2. [2]

    Proceedings of the 29th ACM SIGKDD conference on knowledge discovery and data mining , pages=

    Generating synergistic formulaic alpha collections via reinforcement learning , author=. Proceedings of the 29th ACM SIGKDD conference on knowledge discovery and data mining , pages=

  3. [3]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Alphaforge: A framework to mine and dynamically combine formulaic alpha factors , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  4. [4]

    Advances in neural information processing systems , volume=

    Attention is all you need , author=. Advances in neural information processing systems , volume=

  5. [5]

    Neural computation , volume=

    Long short-term memory , author=. Neural computation , volume=. 1997 , publisher=

    1997

  6. [6]

    arXiv preprint arXiv:1412.3555 , year=

    Empirical evaluation of gated recurrent neural networks on sequence modeling , author=. arXiv preprint arXiv:1412.3555 , year=

    arXiv

  7. [7]

    Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining , pages=

    Xgboost: A scalable tree boosting system , author=. Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining , pages=

  8. [8]

    arXiv preprint arXiv:2410.15474 , year=

    Optimizing backward policies in GFlownets via trajectory likelihood maximization , author=. arXiv preprint arXiv:2410.15474 , year=

    arXiv

Show all 96 references
  1. [9]

    arXiv preprint arXiv:2404.18400 , year=

    Llm-sr: Scientific equation discovery via programming with large language models , author=. arXiv preprint arXiv:2404.18400 , year=

    arXiv

  2. [10]

    International Conference on Machine Learning , pages=

    Better training of gflownets with local credit and incomplete trajectories , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  3. [11]

    Advances in neural information processing systems , volume=

    Lightgbm: A highly efficient gradient boosting decision tree , author=. Advances in neural information processing systems , volume=

  4. [12]

    International Conference on Machine Learning , pages=

    Learning gflownets from partial episodes for improved convergence and stability , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  5. [13]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: System Demonstrations , pages=

    Alpha-gpt: Human-ai interactive alpha mining for quantitative investment , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: System Demonstrations , pages=

    2025

  6. [14]

    Advances in Neural Information Processing Systems , volume=

    A unified framework for deep symbolic regression , author=. Advances in Neural Information Processing Systems , volume=

  7. [15]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Factorvae: A probabilistic dynamic factor model based on variational autoencoder for predicting cross-sectional stock returns , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  8. [16]

    Proceedings of the 2021 International conference on management of data , pages=

    Alphaevolve: A learning framework to discover novel alphas in quantitative investment , author=. Proceedings of the 2021 International conference on management of data , pages=

    2021

  9. [17]

    IEEE Transactions on Signal Processing , year=

    QuantFactor REINFORCE: mining steady formulaic alpha factors with variance-bounded REINFORCE , author=. IEEE Transactions on Signal Processing , year=

  10. [18]

    International conference on machine learning , pages=

    Towards understanding and improving gflownet training , author=. International conference on machine learning , pages=. 2023 , organization=

    2023

  11. [19]

    Forty-second International Conference on Machine Learning , year=

    Alphaqcm: Alpha discovery in finance with distributional reinforcement learning , author=. Forty-second International Conference on Machine Learning , year=

  12. [20]

    Journal of Machine Learning Research , volume=

    Gflownet foundations , author=. Journal of Machine Learning Research , volume=

  13. [21]

    Advances in Neural Information Processing Systems , volume=

    Trajectory balance: Improved credit assignment in gflownets , author=. Advances in Neural Information Processing Systems , volume=

  14. [22]

    arXiv preprint arXiv:1707.06347 , year=

    Proximal policy optimization algorithms , author=. arXiv preprint arXiv:1707.06347 , year=

    arXiv

  15. [23]

    European semantic web conference , pages=

    Modeling relational data with graph convolutional networks , author=. European semantic web conference , pages=. 2018 , organization=

    2018

  16. [24]

    Advances in neural information processing systems , volume=

    Graph transformer networks , author=. Advances in neural information processing systems , volume=

  17. [25]

    Communication, Simulation, and Intelligent Agents: Implications of Personal Intelligent Machines for Medical Education

    Clancey, William J. Communication, Simulation, and Intelligent Agents: Implications of Personal Intelligent Machines for Medical Education. Proceedings of the Eighth International Joint Conference on Artificial Intelligence (IJCAI-83)

  18. [26]

    Classification Problem Solving

    Clancey, William J. Classification Problem Solving. Proceedings of the Fourth National Conference on Artificial Intelligence

  19. [27]

    , title =

    Robinson, Arthur L. , title =. 1980 , doi =. https://science.sciencemag.org/content/208/4447/1019.full.pdf , journal =

    1980

  20. [28]

    New Ways to Make Microcircuits Smaller---Duplicate Entry

    Robinson, Arthur L. New Ways to Make Microcircuits Smaller---Duplicate Entry. Science

  21. [29]

    Clancey and Glenn Rennels , abstract =

    Diane Warner Hasling and William J. Clancey and Glenn Rennels , abstract =. Strategic explanations for a diagnostic consultation system , journal =. 1984 , issn =. doi:https://doi.org/10.1016/S0020-7373(84)80003-6 , url =

    1984 doi

  22. [30]

    and Rennels, Glenn R

    Hasling, Diane Warner and Clancey, William J. and Rennels, Glenn R. and Test, Thomas. Strategic Explanations in Consultation---Duplicate. The International Journal of Man-Machine Studies

  23. [31]

    Poligon: A System for Parallel Problem Solving

    Rice, James. Poligon: A System for Parallel Problem Solving

  24. [32]

    Transfer of Rule-Based Expertise through a Tutorial Dialogue

    Clancey, William J. Transfer of Rule-Based Expertise through a Tutorial Dialogue

  25. [33]

    The Engineering of Qualitative Models

    Clancey, William J. The Engineering of Qualitative Models

  26. [34]

    2017 , eprint=

    Attention Is All You Need , author=. 2017 , eprint=

    2017

  27. [35]

    Pluto: The 'Other' Red Planet

    NASA. Pluto: The 'Other' Red Planet

  28. [36]

    Wilmott , volume=

    101 formulaic alphas , author=. Wilmott , volume=. 2016 , publisher=

    2016

  29. [37]

    Cell systems , volume=

    A generative neural network for maximizing fitness and diversity of synthetic DNA and protein sequences , author=. Cell systems , volume=. 2020 , publisher=

    2020

  30. [38]

    the Journal of Finance , volume=

    The cross-section of expected stock returns , author=. the Journal of Finance , volume=. 1992 , publisher=

    1992

  31. [39]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Factorvae: A probabilistic dynamic factor model based on variational autoencoder for predicting cross-sectional stock returns , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  32. [40]

    Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages =

    Yu, Shuo and Xue, Hongyan and Ao, Xiang and Pan, Feiyang and He, Jia and Tu, Dandan and He, Qing , title =. Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages =. 2023 , isbn =. doi:10.1145/3580305.3599831 , abstract =

    2023

  33. [41]

    arXiv preprint arXiv:2308.00016 , year=

    Alpha-GPT: Human-AI Interactive Alpha Mining for Quantitative Investment , author=. arXiv preprint arXiv:2308.00016 , year=

  34. [42]

    arXiv preprint arXiv:2002.08245 , year=

    AutoAlpha: An efficient hierarchical evolutionary algorithm for mining alpha factors in quantitative investment , author=. arXiv preprint arXiv:2002.08245 , year=

    2002 arXiv

  35. [43]

    Proceedings of the 2021 International Conference on Management of Data , pages =

    Cui, Can and Wang, Wei and Zhang, Meihui and Chen, Gang and Luo, Zhaojing and Ooi, Beng Chin , title =. Proceedings of the 2021 International Conference on Management of Data , pages =. 2021 , isbn =. doi:10.1145/3448016.3457324 , abstract =

    2021

  36. [44]

    IEEE Transactions on Smart Grid , volume=

    Online detection of stealthy false data injection attacks in power system state estimation , author=. IEEE Transactions on Smart Grid , volume=. 2016 , publisher=

    2016

  37. [45]

    Proceedings of the 2021 International Conference on Management of Data , pages=

    AlphaEvolve: A Learning Framework to Discover Novel Alphas in Quantitative Investment , author=. Proceedings of the 2021 International Conference on Management of Data , pages=

    2021

  38. [46]

    Renewable and Sustainable Energy Reviews , volume=

    A review on time series forecasting techniques for building energy consumption , author=. Renewable and Sustainable Energy Reviews , volume=. 2017 , publisher=

    2017

  39. [47]

    arXiv preprint arXiv:2006.14171 , year=

    A closer look at invalid action masking in policy gradient algorithms , author=. arXiv preprint arXiv:2006.14171 , year=

    2006 arXiv

  40. [48]

    , title =

    Ehsani, Sina and Linnainmaa, Juhani T. , title =. The Journal of Finance , volume =. doi:https://doi.org/10.1111/jofi.13131 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1111/jofi.13131 , abstract =

    doi

  41. [49]

    Proceedings of the 2023 9th International Conference on Industrial and Business Engineering , pages =

    Zhang, Yaoyu and Lv, Lin , title =. Proceedings of the 2023 9th International Conference on Industrial and Business Engineering , pages =. 2023 , isbn =. doi:10.1145/3629378.3629454 , abstract =

    2023

  42. [50]

    Proceedings of the 5th International Conference on Big Data and Education , pages =

    Zhao, Heng and Zuo, Xumin and Lin, Peisong , title =. Proceedings of the 5th International Conference on Big Data and Education , pages =. 2022 , isbn =. doi:10.1145/3524383.3524396 , abstract =

    2022

  43. [51]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Spatio-temporal self-supervised learning for traffic flow prediction , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  44. [52]

    Scientific Reports , volume=

    Customer churn prediction using composite deep learning technique , author=. Scientific Reports , volume=. 2023 , publisher=

    2023

  45. [53]

    2019 , institution=

    Revisiting Stock Alpha Mining Based On Genetic Algorithm , author=. 2019 , institution=

    2019

  46. [54]

    arXiv preprint arXiv:2004.05388 , year=

    You impress me: Dialogue generation via mutual persona perception , author=. arXiv preprint arXiv:2004.05388 , year=

    2004 arXiv

  47. [55]

    arXiv preprint arXiv:2111.00053 , year=

    Symbolic regression via neural-guided genetic programming population seeding , author=. arXiv preprint arXiv:2111.00053 , year=

    arXiv

  48. [56]

    Petersen and Mikel Landajuela and T

    Brenden K. Petersen and Mikel Landajuela and T. Nathan Mundhenk and Cl. Deep symbolic regression: Recovering mathematical expressions from data via risk-seeking policy gradients , booktitle =. 2021 , url =

    2021

  49. [57]

    2007 , publisher=

    Quantitative equity portfolio management: modern techniques and applications , author=. 2007 , publisher=

    2007

  50. [58]

    International Conference on Machine Learning , pages=

    Learning equations for extrapolation and control , author=. International Conference on Machine Learning , pages=. 2018 , organization=

    2018

  51. [59]

    Nature , volume=

    Mastering atari, go, chess and shogi by planning with a learned model , author=. Nature , volume=. 2020 , publisher=

    2020

  52. [60]

    Journal of biomedical informatics , volume=

    How are you feeling?: A personalized methodology for predicting mental states from temporally observable physical and behavioral information , author=. Journal of biomedical informatics , volume=. 2017 , publisher=

    2017

  53. [61]

    2019 , publisher=

    Finding Alphas: A quantitative approach to building trading strategies , author=. 2019 , publisher=

    2019

  54. [62]

    arXiv preprint arXiv:2106.14131 , year=

    Symbolicgpt: A generative transformer model for symbolic regression , author=. arXiv preprint arXiv:2106.14131 , year=

    arXiv

  55. [63]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    DeepTrader: a deep reinforcement learning approach for risk-return balanced portfolio management with market conditions Embedding , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  56. [64]

    arXiv preprint arXiv:2110.13716 , year=

    Hist: A graph-based framework for stock trend forecasting via mining concept-oriented shared information , author=. arXiv preprint arXiv:2110.13716 , year=

    arXiv

  57. [65]

    Proceedings of the Web Conference 2021 , pages=

    Rest: Relational event-driven stock trend forecasting , author=. Proceedings of the Web Conference 2021 , pages=

    2021

  58. [66]

    arXiv preprint arXiv:2009.11189 , year=

    Qlib: An ai-oriented quantitative investment platform , author=. arXiv preprint arXiv:2009.11189 , year=

    2009 arXiv

  59. [67]

    Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining , pages=

    Stock price prediction via discovering multi-frequency trading patterns , author=. Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining , pages=

  60. [68]

    International Conference on Machine Learning , pages=

    Neural symbolic regression that scales , author=. International Conference on Machine Learning , pages=. 2021 , organization=

    2021

  61. [69]

    Proceedings of the National Academy of Sciences , volume=

    Data-driven discovery of coordinates and governing equations , author=. Proceedings of the National Academy of Sciences , volume=. 2019 , publisher=

    2019

  62. [70]

    Proceedings of the 12th annual conference on Genetic and evolutionary computation , pages=

    Age-fitness pareto optimization , author=. Proceedings of the 12th annual conference on Genetic and evolutionary computation , pages=

  63. [71]

    Parallel Problem Solving from Nature-PPSN XII: 12th International Conference, Taormina, Italy, September 1-5, 2012, Proceedings, Part I 12 , pages=

    Geometric semantic genetic programming , author=. Parallel Problem Solving from Nature-PPSN XII: 12th International Conference, Taormina, Italy, September 1-5, 2012, Proceedings, Part I 12 , pages=. 2012 , organization=

    2012

  64. [72]

    Proceedings of the Genetic and Evolutionary Computation Conference 2016 , pages=

    Epsilon-lexicase selection for regression , author=. Proceedings of the Genetic and Evolutionary Computation Conference 2016 , pages=

    2016

  65. [73]

    Proceedings of the 2014 Annual Conference on Genetic and Evolutionary Computation , pages=

    Multiple regression genetic programming , author=. Proceedings of the 2014 Annual Conference on Genetic and Evolutionary Computation , pages=

    2014

  66. [74]

    Proceedings of the genetic and evolutionary computation conference , pages=

    Where are we now? A large benchmark study of recent symbolic regression methods , author=. Proceedings of the genetic and evolutionary computation conference , pages=

  67. [75]

    A Unified Framework for Deep Symbolic Regression , url =

    Landajuela, Mikel and Lee, Chak Shing and Yang, Jiachen and Glatt, Ruben and Santiago, Claudio P and Aravena, Ignacio and Mundhenk, Terrell and Mulcahy, Garrett and Petersen, Brenden K , booktitle =. A Unified Framework for Deep Symbolic Regression , url =

  68. [76]

    arXiv preprint arXiv:1905.10947 , year=

    Graph neural networks exponentially lose expressive power for node classification , author=. arXiv preprint arXiv:1905.10947 , year=

    1905 arXiv

  69. [77]

    arXiv preprint arXiv:1511.06295 , year=

    Policy Distillation , author=. arXiv preprint arXiv:1511.06295 , year=

    arXiv

  70. [78]

    2024 , url=

    Gu, Yuxian and Dong, Li and Wei, Furu and Huang, Minlie , booktitle=. 2024 , url=

    2024

  71. [79]

    The Twelfth International Conference on Learning Representations , year=

    On-Policy Distillation of Language Models: Learning from Self-Generated Mistakes , author=. The Twelfth International Conference on Learning Representations , year=

  72. [80]

    arXiv preprint arXiv:2604.00626 , year=

    A Survey of On-Policy Distillation for Large Language Models , author=. arXiv preprint arXiv:2604.00626 , year=

    arXiv

  73. [81]

    arXiv preprint arXiv:2604.13016 , year=

    Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe , author=. arXiv preprint arXiv:2604.13016 , year=

    arXiv

  74. [82]

    arXiv preprint arXiv:2605.10889 , year=

    Unmasking On-Policy Distillation: Where It Helps, Where It Hurts, and Why , author=. arXiv preprint arXiv:2605.10889 , year=

    arXiv

  75. [83]

    arXiv preprint arXiv:2603.25562 , year=

    Revisiting On-Policy Distillation: Empirical Failure Modes and Simple Fixes , author=. arXiv preprint arXiv:2603.25562 , year=

    arXiv

  76. [84]

    Decoupling

    Zhao, Anhao and Xin, Haoran and Fan, Yingqi and Tong, Junlong and Li, Wenjie and Shen, Xiaoyu , journal=. Decoupling

  77. [85]

    Zhao, Hanyang and Chen, Haoxian and Lin, Han and Winata, Genta Indra and Yao, David and Tang, Wenpin , journal=

  78. [86]

    arXiv preprint arXiv:2605.13643 , year=

    Prefix Teach, Suffix Fade: Local Teachability Collapse in Strong-to-Weak On-Policy Distillation , author=. arXiv preprint arXiv:2605.13643 , year=

    arXiv

  79. [87]

    Liang, Kun and Tang, Chenming and Bai, Clive and Liu, Weijie and Yang, Saiyong and Wu, Yunfang , journal=

  80. [88]

    arXiv preprint arXiv:2602.12275 , year=

    On-Policy Context Distillation for Language Models , author=. arXiv preprint arXiv:2602.12275 , year=

    arXiv

  81. [89]

    arXiv preprint arXiv:2605.09725 , year=

    On-Policy Distillation with Best-of-N Teacher Rollout Selection , author=. arXiv preprint arXiv:2605.09725 , year=

    arXiv

  82. [90]

    Zhong, Qiyong and Zheng, Mao and Song, Mingyang and Lin, Xin and Sun, Jie and Jiang, Houcheng and Wang, Xiang and Fang, Junfeng , journal=

  83. [91]

    arXiv preprint arXiv:2601.18734 , year=

    Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models , author=. arXiv preprint arXiv:2601.18734 , year=

    arXiv

  84. [92]

    arXiv preprint arXiv:2605.06387 , year=

    Asymmetric On-Policy Distillation: Bridging Exploitation and Imitation at the Token Level , author=. arXiv preprint arXiv:2605.06387 , year=

    arXiv

  85. [93]

    Yang, Zhicheng and Guo, Zhijiang and Song, Yifan and Xu, Minrui and Wang, Yongxin and Wang, Yiwei and Liang, Xiaodan and Tang, Jing , journal=

  86. [94]

    arXiv preprint arXiv:2606.01249 , year=

    Trust Region On-Policy Distillation , author=. arXiv preprint arXiv:2606.01249 , year=

    arXiv

  87. [95]

    arXiv preprint arXiv:2606.02684 , year=

    Filter, Then Reweight: Rethinking Optimization Granularity in On-Policy Distillation , author=. arXiv preprint arXiv:2606.02684 , year=

    arXiv

  88. [96]

    AlphaAgent:

    Tang, Ziyi and Chen, Zechuan and Yang, Jiarui and Mai, Jiayao and Zheng, Yongsen and Wang, Keze and Chen, Jinrui and Lin, Liang , booktitle=. AlphaAgent:. 2025 , doi=

    2025

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.