Skip to content
JevDirectory.orgTraffic count unavailable. View site traffic
CommunityJev Practices & Patterns

Jev As Quant

A research repository that wires Jev and its open-weight counterpart Laya into a look-ahead-free quant stack as the judgment layer, then reports seven experiments including the ones that failed.

See highlightsOpen github.com
Stars
6
Language
Python
License
APACHE-2.0
Published by
Community
Author
jiayylu
Last verified
Oct 2026
Tagscommunityawesomejevcatalog-import
Works forJev for Classification
Used inJev for Software & DevTools
Languages
Python

Highlights

  • Laya reads financial text well and numbers badly: zero-shot news macro-F1 0.71 once calibrated, but 0.52 balanced accuracy on raw figures.
  • Escalating only the least confident 18% to Claude lifts macro-F1 from 0.71 to 0.78, beating a TF-IDF baseline, for about $0.36 per thousand items.
  • On 8,677 real SEC filings the read matched the market's immediate reaction, but that move lands before a daily trader can trade; all readings lost after costs.
  • An alpha factory put 667 candidates through pre-registered gates and accepted exactly one; the useful output was the 666 it rejected.

Watch out

This is a negative-result research repository and the README says so: no broker interface, no order placement, nothing here is investment advice. The headline holdout result beat an equal-weight benchmark by +3.0%/yr at t=0.83, which is not significant, and trailed SPY by 3.8%/yr, with the stated conclusion that the market was not beaten. Laya weights are roughly 1.7 GB per checkpoint unless you use the released decision cache. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

jev_local
38GitHub stars
Local unofficial /v1/systemone adapter that serves Choice, Score, and Noul probabilities via first-token LLM judgments or a fine-tuned ModernBERT scorer, including Japanese docs.
Jev Practices & Patterns#community#awesomejev#catalog-import
CommunityArgos1111/jev_localPython
jev-on-a-laptop
24GitHub stars
Unofficial laptop study of parallel constrained decoding: prefill once, share the KV cache across fields, and decide all schema fields in one batched forward pass without writing JSON.
Jev Practices & Patterns#community#jevbest#catalog-import
Communityrorshopping/jev-on-a-laptopPython
jeval
22GitHub stars
Offline evaluator that checks whether classifier confidences match reality and converts mistake costs into a deployable threshold, emitting an HTML report plus machine-readable threshold files.
Jev Practices & Patterns#community#jevbest#catalog-import
Communityrlaope/jevalPython
jev-rag-benchmark
16GitHub stars
Vendor-neutral harness that tests whether Jev reranking improves a small RAG stack, using frozen hybrid BM25 plus BGE-M3 candidates and isolating retrieval, rerank, and generation stages.
Jev Practices & Patterns#community#jevbest#catalog-import
Communityerendikmenn/jev-rag-benchmarkPython
system-one
37GitHub stars
Minimal Python recipe that turns any causal LLM into a Jev-style classifier by prefilling choice_index and constraining sampling to legal indexes in one forward pass.
Jev Practices & Patterns#community#jevbest#catalog-import
Communitysgoedecke/system-onePython
system-one-open
39GitHub stars
Open replica of the typed decision API on openly available base models, answering choice, score, and true-probability questions in one forward pass with no decoding.
Jev Practices & Patterns#community#jevbest#catalog-import
Communitymithalouni/system-one-openPython

From the Jev guides

Original write-ups on what this entry is for.

Designing the State: What to Include and What to Drop
Most Jev accuracy problems are state problems. What belongs in a state, what is a pre-made judgment to leave out, and why naming the slice you ask about matters more than the wording.
Jev Practices & Patterns#state#design#primitives
Read article
Noul, End to End: One Probability and a Dozen Thresholds
A Noul returns one number and no separate confidence. How to read the value, pin its boundary with criteria, split compound questions, and set a per-action threshold that matches the cost of being wrong.
Jev Practices & Patterns#noul#primitives#confidence
Read article
Scoring Without Measuring: Bands, Rubrics, and Composite Triage
A Score is a weighted position, not a ruler. Why the number can land between levels, why levels are judged separately, and how to combine several atomic scores in code instead of one crowded rubric.
Jev Practices & Patterns#score#primitives#scoring
Read article
Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai