YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

QuantaBeta Core β€” Sovereign Deterministic Alpha Mining QuantaBeta Core Architecture

QuantaBeta Core

Sovereign Deterministic Alpha Mining

License: Sovereign Source v1.0 BSL Float Rust Haskell Lean 4 WORM Trust


LLMs generate coherent noise, not alpha.

This pipeline generates alpha from number theory.


What Is This?

QuantaBeta Core is a sovereign quantitative finance pipeline that replaces the standard "LLM research agent β†’ code gen β†’ backtest" loop with arithmetic invariant search β†’ proof-carrying code β†’ formally validated factors.

The central claim: market alpha is arithmetic structure, not statistical pattern. Ramanujan partition congruences, Hecke operator eigenvalues, and Rogers-Ramanujan identities are not metaphors. They are executable filters that select for genuine predictive structure in return series β€” structure that persists because it is grounded in number theory, not in learned correlations.

Every result is deterministic. Every computation is exactly rational. Every factor is WORM-sealed. Every acceptance criterion is a theorem, not a threshold.


The Arithmetic Invariant β€” No Floats. Ever.

The founding constraint of this codebase: f64 is banned at every layer.

This is not a style preference. It is a mathematical requirement.

Standard quant libraries (NumPy, pandas, VectorBT) use IEEE 754 floating-point. Float arithmetic is non-associative, non-commutative under rounding, and platform-dependent. Two machines running the same backtest can produce different results. A factor that "works" in development may fail in production because the rounding modes differ.

QuantaBeta uses:

Computation Type Library
Return series features rug::Rational GMP arbitrary-precision
PnL accounting rug::Integer GMP exact integer
Entropy computation rug::Float with Round::Down/Round::Up MPFR directed rounding
Sharpe ratio Rational interval [L, U] Exact bounds
Symbolic entropy SymLog2 { coeff: Rational, base: Integer } No evaluation

The result: given the same input, the pipeline produces bitwise-identical output on every machine, every run, forever.


The Mathematical Foundation

Ramanujan Partition Theory

The partition function p(n) counts the number of ways to write n as an ordered-indifferent sum of positive integers. It appears in three roles:

1. Complexity bound. p(n) bounds the search space for features of complexity n. Since p(n) ~ exp(Ο€βˆš(2n/3)) / (4n√3), the search space is super-polynomial but enumerable for small n.

2. Volatility measure. compute_partition_volatility replaces variance with a partition-entropy: given return bucket frequencies (f₁, ..., fβ‚–), the volatility is Ξ£ p(fα΅’)/p(window). Partition numbers measure "how many ways can this frequency distribution arise" β€” higher partition entropy means more combinatorial uncertainty.

3. Congruence filter. Ramanujan's exact congruences:

  • p(5k+4) ≑ 0 (mod 5) for all k β‰₯ 0 β€” verified in code, tested against OEIS A000041
  • p(7k+5) ≑ 0 (mod 7) for all k β‰₯ 0 β€” verified in code

A factor whose complexity index falls at a congruence residue is flagged as having low informational content. This is an arithmetic sieve, not a heuristic.

Hecke Operators

The Hecke operator T_n acts on a modular form f by:

(T_n f)_m = Ξ£_{d | gcd(n,m)} d^(k-1) * a_{nm/dΒ²}

In the pipeline, hecke_cross_correlation computes ⟨T_n(series_A), series_B⟩. If two return series arise from instruments related by an Eichler-Shimura construction β€” i.e., their L-functions share a newform β€” this inner product is large at the corresponding Hecke eigenvalue and small otherwise. This is the cross-predictability signal.

The Deligne bound |a_p(f)| ≀ 2p^((k-1)/2) (Fields Medal 1978) bounds the eigenvalues. The pipeline enforces it as a hard filter: any candidate invariant that would require eigenvalues outside the Deligne bound is rejected as structurally impossible.

Connection to PAR-011 (Jacobian Conjecture): The golden ratio Ο† = (1+√5)/2 that appears in the Jacobian proof via Jordan algebras also appears as the characteristic eigenvalue bound for the simplest Hecke operator T_2 on weight-2 forms. Four independent mathematical contexts, one structure. See: Zenodo 10.5281/zenodo.21727363.

Rogers-Ramanujan Identities

The first Rogers-Ramanujan identity:

Ξ£_{nβ‰₯0} q^(nΒ²) / (q;q)_n  =  Ξ _{nβ‰₯0} 1/((1-q^(5n+1))(1-q^(5n+4)))

This connects the combinatorial structure of sequences with gap constraints to the Ramanujan partition congruences β€” factors selected by the RamanujanCong(5, 4) invariant live precisely in the residue classes 5n+1 and 5n+4 of the product side.

True Entropy and the Ξ© = 0.21 Threshold

Shannon entropy: H(P) = logβ‚‚(N) - (1/N) Ξ£ cα΅’ logβ‚‚(cα΅’)

H is an algebraic number β€” a linear combination of logs of integers. The pipeline computes it three ways:

  • Point: MPFR at 256-bit precision, correctly rounded
  • Interval: Guaranteed bounds [L, U] with Round::Down / Round::Up
  • Symbolic: H = (1/N)logβ‚‚(N) + Ξ£(-cα΅’/N)logβ‚‚(cα΅’) β€” no evaluation, pure algebra

The entropy_coherent(Counts, 0.21) predicate in logic/entropy.pl gates every factor. A factor whose residuals have entropy below 0.21 bits concentrates β‰₯ 96.6% of its probability mass on a single outcome. This threshold mirrors the Ξ© field coherence gate in the SnapKitty constellation β€” the system stays coherent when its entropy is below 0.21.


Pipeline

Market Data
     |
     | rug::Rational β€” no f64 past this point
     v
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  LAYER 1: SYMBOLIC FEATURE ALGEBRA                             β”‚
β”‚  crates/quantabeta-core/src/features.rs                        β”‚
β”‚                                                                 β”‚
β”‚  compute_partition_volatility(returns, window)                  β”‚
β”‚    β†’ entropy of partition frequencies over return buckets       β”‚
β”‚    β†’ exact Rational output, deterministic                       β”‚
β”‚                                                                 β”‚
β”‚  hecke_cross_correlation(series_a, series_b, level)             β”‚
β”‚    β†’ ⟨T_n(series_A), series_B⟩ exact rational inner product    β”‚
β”‚    β†’ measures Hecke eigenvalue overlap between instruments      β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                               |
                               v
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  LAYER 2: ARITHMETIC INVARIANT SEARCH                          β”‚
β”‚  haskell/src/Quantabeta/InvariantSearch.hs                     β”‚
β”‚                                                                 β”‚
β”‚  Enumerates typed candidate invariants:                        β”‚
β”‚    HeckeCorr(level, weight)   β€” prime levels, even weights      β”‚
β”‚    PartitionVol(window)       β€” standard trading windows        β”‚
β”‚    RamanujanCong(modulus, residue) β€” mod 5, 7, 11               β”‚
β”‚                                                                 β”‚
β”‚  wellTyped filter: Hecke weights must be even, windows ≀ 252   β”‚
β”‚  Replaces: LLM research agent                                   β”‚
β”‚  Outputs: SGML <claim> tags for claimguard oracle              β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                               |
                               v
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  LAYER 3: FACTOR SYNTHESIS                                     β”‚
β”‚  logic/factor_synthesis.pl                                     β”‚
β”‚                                                                 β”‚
β”‚  Prolog DCG: invariant AST β†’ compilable Rust code              β”‚
β”‚  DCG grammars are provably correct β€” generated code is         β”‚
β”‚  structurally guaranteed syntactically valid                    β”‚
β”‚  Content-addressed factor ID from AST hash                     β”‚
β”‚  Emits Bifrost JSON audit manifest                             β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                               |
                               v
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  LAYER 4: DETERMINISTIC BACKTEST                               β”‚
β”‚  crates/quantabeta-core/src/backtest.rs                        β”‚
β”‚                                                                 β”‚
β”‚  Lamport logical clock β€” not wall time. Order is provable.     β”‚
β”‚  PnL = Ξ£(pos_t Γ— (price_{t+1} - price_t)) - fees              β”‚
β”‚  All arithmetic: rug::Integer (exact)                          β”‚
β”‚  Sharpe = Rational interval [L, U] β€” not a point estimate      β”‚
β”‚  SHA-256 audit hash seals exact PnL + Sharpe bounds            β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                               |
                               v
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  LAYER 5: FORMAL VALIDATION                                    β”‚
β”‚  lean/Quantabeta/Validation.lean                               β”‚
β”‚                                                                 β”‚
β”‚  IsRobust(f, baseline, Ξ΅) :=                                   β”‚
β”‚    βˆ€ noise : |noise_i| ≀ Ξ΅.epsilon,                            β”‚
β”‚      pnl(f, baseline + noise) > 0                              β”‚
β”‚                                                                 β”‚
β”‚  A universally quantified statement over ALL perturbations.    β”‚
β”‚  Not Sharpe > 1.5. A theorem.                                  β”‚
β”‚  Ramanujan congruence axiom + Deligne bound axiom included.    β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
                               |
                               v
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚  LAYER 6: WORM FACTOR REGISTRY                                 β”‚
β”‚  crates/quantabeta-core/src/worm.rs                            β”‚
β”‚                                                                 β”‚
β”‚  Each FactorArtifact carries:                                  β”‚
β”‚    arithmetic_invariant β€” the number-theoretic basis           β”‚
β”‚    proof_hash β€” Lean 4 proof term hash                         β”‚
β”‚    code_hash β€” Rust WASM hash                                  β”‚
β”‚    sharpe_interval β€” [L, U] rational bounds                    β”‚
β”‚    entropy_signature β€” true entropy of residuals               β”‚
β”‚    operator β€” "Ahmad_Ali_Parr"                                 β”‚
β”‚    previous_seal β€” SHA-256 chain link                          β”‚
β”‚                                                                 β”‚
β”‚  verify_chain() checks entire chain in O(n)                    β”‚
β”‚  β†’ Connects to snap-os/bifrost for Blake3+Ed25519 sealing      β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Cross-Cutting: True Entropy

crates/true-entropy is used across all layers as the exact entropy primitive.

// Point estimate β€” MPFR 256-bit, correctly rounded
let h = shannon_entropy_exact([3u64, 1, 2, 4], 256);

// Guaranteed interval β€” directed rounding
let (lo, hi) = shannon_entropy_interval([3u64, 1, 2, 4], 256);
// Invariant: lo ≀ true_entropy ≀ hi, always

// Symbolic β€” no evaluation, pure algebra
let sym = shannon_entropy_symbolic([3u64, 1, 2, 4]);
// Returns: [SymLog2{coeff: 1/10, base: 10}, SymLog2{coeff: -3/10, base: 3}, ...]
// H = (1/10)logβ‚‚(10) + (-3/10)logβ‚‚(3) + (-1/10)logβ‚‚(1) + ...

The entropy_coherent(Counts, 0.21) Prolog predicate calls this layer and gates the entire pipeline.


What Is Built

Layer File What It Does Tests
1 crates/quantabeta-core/src/features.rs Partition volatility + Hecke cross-correlation, exact rational OEIS A000041 p(0..10), determinism
1 crates/ramanujan-ops/src/partition.rs HRR partition p(n), Ramanujan congruences mod 5 and 7 OEIS A000041 p(0..20), congruences
1 crates/ramanujan-ops/src/hecke.rs T_n double-coset formula, Deligne bound T_1 identity, Deligne bound
1 crates/ramanujan-ops/src/qseries.rs q-integers, q-Pochhammer, Rogers-Ramanujan RR identity at q=1/10
cross crates/true-entropy/src/lib.rs Exact/interval/symbolic Shannon entropy, MPFR Uniform=1bit, certain=0, interval contains point
cross haskell/src/Verified/Entropy.hs Symbolic entropy HOC, rational logβ‚‚ intervals, partition entropy Type-checked
2 haskell/src/Quantabeta/InvariantSearch.hs Typed invariant enumeration, Deligne+IC checks, SGML output wellTyped filter
3 logic/factor_synthesis.pl Prolog DCG β†’ Rust code gen, Bifrost manifest Hecke + partition synthesis
3 logic/entropy.pl Bifrost FFI bridge, Ξ© coherence gate, WORM audit Integration (requires FFI)
4 crates/quantabeta-core/src/backtest.rs Lamport clock, integer PnL, rational Sharpe interval Determinism test
5 lean/Quantabeta/Validation.lean Formal robustness βˆ€ Ξ΅-bounded noise trivially_robust_increasing
6 crates/quantabeta-core/src/worm.rs SHA-256 append-only WORM chain Chain integrity

Run

cargo test --workspace

Tests verify:

  • p(0)..p(20) match OEIS A000041 exactly
  • p(5k+4) ≑ 0 (mod 5) holds for k=0..10 (Ramanujan)
  • p(7k+5) ≑ 0 (mod 7) holds for k=0..5 (Ramanujan)
  • Deligne bound |a_2| ≀ 64 satisfied for Delta function
  • Rogers-Ramanujan identity verified at q=1/10 to order 20
  • Shannon entropy [1,1] = exactly 1 bit at 256-bit precision
  • Interval [L,U] always contains point estimate
  • Backtest determinism: same ticks β†’ same PnL β†’ same audit hash
  • WORM chain integrity verified after 2 appends

Connection to SnapKitty Stack

Repo Role
snapkitty-clojure-lisp-bridge claimguard oracle gates every factor claim via SGML before WORM seal
snap-os/bifrost Production WORM β€” upgrade worm.rs SHA-256 to Blake3+Ed25519
the-49th-call Abjad-Swarm Born rule weighting uses Ο†^(-i) β€” same Ο† as Hecke bounds
jacobian-formal PAR-011 Jordan operator uses the same Ο†. Four independent contexts, one structure.
gkn-i4-e7-lean Iβ‚„ quartic invariant structure mirrors partition function algebra

The Ο† Convergence

The golden ratio Ο† = (1+√5)/2 appears independently in four formal contexts across this constellation:

Context How Repo
PAR-011 Jordan fixed-point operator T(ρ) = φ⁻¹UρU† + φ⁻²ρ, drives commutativity jacobian-formal
Hecke eigenvalue bound Characteristic eigenvalue of T_2 on weight-2 forms quantabeta-core
Abjad-Swarm Born rule Agent weighting Ο†^(-i), golden ratio decay per level the-49th-call
Iβ‚„ quartic invariant E₇ symmetry structure gkn-i4-e7-lean

This is not numerology. It is convergence across independent formal derivations. Each is machine-verifiable.


Prior Art

Record DOI Date
Jordan Spectral Transformer (Ο† operator origin) 10.5281/zenodo.21443609 2026-07-19
PAR-011: Jacobian Conjecture via Jordan Algebras 10.5281/zenodo.21727363 2026-07-31

WORM anchor: github.com/SNAPKITTYWEST/quantabeta-core


License

Sovereign Source License v1.0 β€” Business Source License variant.

  • Non-production use: Free. Research, education, evaluation, personal projects.
  • Production use (live or paper trading, capital > $1,000): Requires commercial license until 2029-01-01.
  • After 2029-01-01: AGPL-3.0.

The IP is held by Bel Esprit D'Accord Irrevocable Trust (EIN 42-697643). Unauthorized commercial use is interference with trust property.

See LICENSE for full terms including WORM chain integrity clause, namespace protection, and prior art anchors.

Commercial licensing: ahmedparr93@gmail.com | collectivekitty.com


Built by: Ahmad Ali Parr + Claude Code
Trust: Bel Esprit D'Accord Irrevocable Trust
Constellation: SNAPKITTYWEST

Ω = TRUST ∧ CODE

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Space using Snapkitty/quantabeta-core 1

Collection including Snapkitty/quantabeta-core