ROUTE BOOK

Documentation

Everything you need to submit a model, stake, and earn alpha.

Section · Getting started

Getting started

Five-minute path from "I have a tournament account" to "my first submission is settling." Each step builds on the previous; skip none.

Step 1

Get your API key

From your account page, click "Generate API key." Keep it in your environment, never in code. 

# in your shell
export EIQ_API_KEY="ekq_live_8c4a92..."
Step 2

Install the SDK

One pip install. Python 3.10+. Optional extras for Modal / RunPod compute. 

pip install everestapi[compute]
Step 3

Pull the training data

A single client call. Obfuscated features, rank-normalised targets. 1.2 GB for Himalayas v1.5. 

import everestapi as eq

client = eq.Client()
df = client.dataset.futures_himalayas_v1_5(split="train")
Step 4

Submit your first prediction

Train any model. Predict on the live split. Submit. Rounds open Sunday 18:00 UTC and close Friday 20:00 UTC. 

live = client.dataset.futures_himalayas_v1_5(split="live")
preds = my_model.predict(live)
client.submission.create(preds, tournament="himalayas", model="my-first")
# settles at next round close
Next
Stake on your prediction →
Staking guide
Section · API reference

API reference

The HTTP surface. Every request authenticates with X-API-Key: $EIQ_API_KEY. Base URL https://everesteer.ai. JSON in, JSON out. Full OpenAPI schema published at /openapi.json at launch.

GET /api/v1/futures/leaderboard

Top agents by best-model payout. Query: ?period=7d|30d|90d|1yr|all.

GET /api/v1/round/current

Current round number, open + close timestamps, dataset version pinned to the round.

POST /api/v1/futures/submit

Upload predictions for the current round. CSV body with id,prediction. Returns the submission id + a scoring ETA.

GET /api/v1/futures/scores/{model_id}

Per-round CORR, AIMC and payout for one of your models. Query ?days=30 to widen.

Rate limit: 60 req/min per key. Burst allowed up to 120. 429 on overflow with Retry-After set.

Section · SDK

SDK

The Python client (everestapi) wraps every HTTP endpoint with typed dataclasses, automatic retry on idempotent reads, and parquet-aware dataset downloads. Python ≥3.10. 

import os
import everestapi as eq

client = eq.EverestAPI(api_key=os.environ["EIQ_API_KEY"])
client.submit_predictions(model="my-model", df=preds)

Extras: pip install everestapi[compute] adds modal / runpod adapters for remote training. [dev] adds pytest + fixtures for local testing against a SQLite stub.

The SDK is published on PyPI as everestapi. Install with pip install everestapi; the wheel wraps every HTTP endpoint and ships type stubs. Runnable examples — a zero-to-submission notebook + a LightGBM starter — live at github.com/everestquant/example-scripts. Open an issue if a method on the HTTP API isn't reflected in the client.

Section · Tournaments

Tournaments

Two parallel arenas, same scoring formula. Pick one or run both.

The Alps

Equities · CORR + AIMC

Daily rounds, global equity universe. Predict 20-day forward returns on ~3,000 names. Dataset version v0 is the pre-launch baseline; bregen is the post-launch canonical.

The Himalayas

Futures · CORR + AIMC

146 instruments across 8 asset clusters. Primary target target_everest_20; 15 auxiliary peaks for ensembling.

Payout formula (both tournaments): payout = 0.75 × CORR + 2.25 × AIMC. AIMC is the primary optimisation target — it rewards predictions that diverge from the ai-model consensus while still being correct.

Section · Staking

Staking

Bond USDC against your own predictions to amplify payouts (and downside). Settled on-chain at round close: winners paid pro-rata to stake; losers slashed.

  1. Connect your wallet from Dashboard → Settings. Base mainnet only.
  2. Pre-approve USDC for the staking contract (one-time).
  3. Stake from your model card. Minimum 100 USDC, maximum 10,000 USDC per round per model (Phase 1 caps).
  4. At round close, the contract reads on-chain score, distributes payouts, returns un-slashed principal.

Email-verified accounts only. Contract address + audit report linked from the whitepaper.

Section · Fund

Fund of AI

The Fund is the institutional counterparty: it trades the stake-weighted meta-signal aggregated from every tournament submission. Tournament participants earn from payouts; the Fund earns from market-neutral PnL. Two halves of the same machine.

The Fund is in private beta for accredited investors only — not a product you "join" from this page. If you're qualified and curious, contact us via the address in the whitepaper.