inference_example.py

"""Minimal inference snippet for jc-builds/polymarket-edge-bot.

Install: pip install lightgbm scikit-learn sentence-transformers numpy
"""
import json
import pickle
import numpy as np
import lightgbm as lgb
from sklearn.isotonic import IsotonicRegression
from sentence_transformers import SentenceTransformer
from huggingface_hub import hf_hub_download

REPO = "jc-builds/polymarket-edge-bot"

# --- load artifacts ---
spec = json.loads(open(hf_hub_download(REPO, "feature_spec.json")).read())
gbm = lgb.Booster(model_file=hf_hub_download(REPO, "lightgbm_model.txt"))
iso = pickle.load(open(hf_hub_download(REPO, "isotonic_calibrator.pkl"), "rb"))
encoder = SentenceTransformer(spec["embedding_model"])

# --- build a tabular row matching feature_spec.json ---
# In production, derive these from the live Polymarket Gamma API payload.
tabular_row = {col: 0.0 for col in spec["tabular_columns_in_order"]}
tabular_row["first_yes_price"] = 0.30          # current YES price
tabular_row["first_yes_price_log_odds"] = float(np.log(0.30 / 0.70))
tabular_row["first_yes_price_distance_from_half"] = 0.20
tabular_row["log_total_usd"] = float(np.log1p(50_000))
tabular_row["duration_days"] = 7

# Question embedding
question = "Will Bitcoin close above $100,000 on Friday?"
emb = encoder.encode([question], normalize_embeddings=True)[0]

# Concatenate
x = np.concatenate([
    np.array([tabular_row[c] for c in spec["tabular_columns_in_order"]],
             dtype=np.float32),
    emb,
])[None, :]

# --- score ---
p_raw = gbm.predict(x)[0]
p_yes = float(iso.predict([p_raw])[0])
edge = p_yes - tabular_row["first_yes_price"]
print(f"P(YES) = {p_yes:.3f}    market = {tabular_row['first_yes_price']:.3f}    "
      f"edge = {edge:+.3f}   {'BUY YES' if edge > 0.05 else 'BUY NO' if edge < -0.05 else 'skip'}")