ZenHodl Python SDK

Calibrated fair win probabilities and edge signals for prediction markets — Polymarket, Kalshi, and sportsbooks — across 8 sports (NBA, WNBA, NCAAMB, NCAAWB, CFB, NFL, NHL, MLB).

Zero dependencies (standard library only). Built-in timeouts and retry with exponential backoff.

Install

pip install zenhodl

Try it without an API key

The sample endpoint is free and needs no signup:

from zenhodl import ZenHodl

client = ZenHodl()               # keyless — sample + health only
sample = client.edges_sample()   # delayed sample of the live edge feed
print(sample["count"], "signals")

Full access

Get a key from zenhodl.net/pricing — the free Developer tier (500 requests/month, 5-minute delay, no card) or a paid plan for real-time data.

from zenhodl import ZenHodl

client = ZenHodl("sk_live_...")

# Games where the model disagrees with the market by ≥10 cents
edges = client.edges(min_edge=10)
for e in edges["signals"]:
    print(e)

# All live games with fair win probabilities
games = client.games(sport="MLB")

# Available sports + model metadata
sports = client.sports()

# Predictions CSV (bytes)
csv_bytes = client.predictions("latest")

# Backtests (Pro tier and above)
result = client.backtest(sport="NBA", min_edge=8)

Configuration

client = ZenHodl(
    "sk_live_...",
    timeout=15.0,   # per-request timeout, seconds
    retries=2,      # retry on 429/5xx/network errors (exponential backoff)
)

Errors raise zenhodl.ZenHodlError with .status and .detail.

What this is (and isn't)

ZenHodl publishes calibrated model probabilities and compares them with live venue prices, with public methodology and a results ledger that includes losses. It is a research and monitoring API — not a source of guaranteed profitable picks, and nothing here is financial advice.

• Docs: zenhodl.net/docs
• Methodology: zenhodl.net/methodology
• Results ledger: zenhodl.net/results

License

MIT