Spaces:
Running
Running
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,10 +1,61 @@
|
|
| 1 |
---
|
| 2 |
title: README
|
| 3 |
-
emoji:
|
| 4 |
colorFrom: blue
|
| 5 |
colorTo: yellow
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
---
|
|
|
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
title: README
|
| 3 |
+
emoji: 📚
|
| 4 |
colorFrom: blue
|
| 5 |
colorTo: yellow
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
+
# Binomial Technologies
|
| 10 |
|
| 11 |
+
**Open-source ML specialists for finance.**
|
| 12 |
+
|
| 13 |
+
We build small (≤500M parameter) task-specific models for finance under Apache 2.0 — engineered for sub-second CPU inference, public eval tables, and drop-in compatibility with the pipelines quant teams actually run.
|
| 14 |
+
|
| 15 |
+
## Thesis
|
| 16 |
+
|
| 17 |
+
For narrow finance tasks, small specialists beat:
|
| 18 |
+
|
| 19 |
+
- **Frontier LLMs** on cost and latency by two orders of magnitude
|
| 20 |
+
- **Dictionary methods** (Loughran-McDonald, FinBERT) on context-awareness and number of dimensions captured per article
|
| 21 |
+
- **Closed bespoke fine-tunes** on auditability — every model card here ships with eval tables, methodology, and explicit limitations
|
| 22 |
+
|
| 23 |
+
Nobody has open-sourced this stack at this fidelity. That's the gap we fill.
|
| 24 |
+
|
| 25 |
+
## The model zoo
|
| 26 |
+
|
| 27 |
+
Six task-specialists named after thinkers in quantitative finance. One per quarter through 2027.
|
| 28 |
+
|
| 29 |
+
| Model | Task | Status |
|
| 30 |
+
| --------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------- | -------------------------- |
|
| 31 |
+
| **[binomial-marks-1](https://huggingface.co/BinomialTechnologies/binomial-marks-1)** | Earnings-call NLP scoring — 23 outputs (10 topics × {mention, direction}, 3 tone) | Shipped (v1.1, April 2026) |
|
| 32 |
+
| binomial-shannon-1 | Financial news characterizer | In progress |
|
| 33 |
+
| binomial-godel-1 | Realized volatility forecasting | In design |
|
| 34 |
+
| binomial-mandelbrot-1 | Market regime classification | In design |
|
| 35 |
+
| binomial-simons-1 | Order-flow / microstructure | In design |
|
| 36 |
+
| binomial-bachelier-1 | Vol surface dynamics | v2 cycle |
|
| 37 |
+
|
| 38 |
+
All models Apache 2.0. All run under 100 ms on CPU (most under 30 ms).
|
| 39 |
+
|
| 40 |
+
## What we publish
|
| 41 |
+
|
| 42 |
+
- **Weights** on this org's HF Hub
|
| 43 |
+
- **Runtime helpers** as PyPI packages — `pip install binomial-marks`
|
| 44 |
+
- **Source, training scripts, eval harnesses** — [github.com/Binomial-Capital-Management/binomial-ai-research](https://github.com/Binomial-Capital-Management/binomial-ai-research)
|
| 45 |
+
- **Model cards** — full eval tables, panel comparisons, tier (1 / 2 / 3) declared upfront
|
| 46 |
+
|
| 47 |
+
## Tier system
|
| 48 |
+
|
| 49 |
+
Each model card declares one of three tiers honestly:
|
| 50 |
+
|
| 51 |
+
| Tier | Definition |
|
| 52 |
+
| ----------- | -------------------------------------------------------------------------------------------------- |
|
| 53 |
+
| **1** | Production-validated against measurable outcomes (returns, realized vol). Tradeable as a feature. |
|
| 54 |
+
| **2** | Research preview. Eval against an LLM panel + held-out test sets. Use as input to your own models. |
|
| 55 |
+
| **3** | Experimental. |
|
| 56 |
+
|
| 57 |
+
We do not host inference. Weights are yours to deploy.
|
| 58 |
+
|
| 59 |
+
## Contact
|
| 60 |
+
|
| 61 |
+
ilay@binomialtec.com
|