Candidate shape
Market, timeframe, strategy components, weights, parameters, and caller
metadata.
Server-owned profile
Datasets, execution policy, and scoring windows are fixed in competition
mode.
Private path
Protected submissions can use the encrypted intent runtime, separate from
public HTTP routes.
What a beta candidate contains
In the public API flow, a candidate is built from:marketId,- optional
timeframe, - strategy components,
- component weights,
- strategy parameters,
- caller-supplied correlation metadata such as
submissionId.
simulation-service. Evaluation lives in ranking-evaluation.
Parameter defaults
Strategy metadata endpoints can exposedefaultValue for optional parameters.
If a parameter has a surfaced default, callers may omit it and the producer
runtime applies the canonical default.
The router does not invent separate strategy defaults.
Competition profile
Current public competition mode is server controlled:Dataset and timeframe
Dataset and timeframe
Official hourly dataset by default, optional
timeframe = "daily", and
internal dataset resolution by market.Scored-window policy
Scored-window policy
Fixed scored-window policy plus fixed execution and scoring profiles.
Execution knobs
Execution knobs
Caller-supplied execution knobs are ignored when competition mode owns the
policy.
Private submission path
AlphaEngine also has an encrypted intent runtime for protected submissions. That path is separate from the public HTTP API. It uses on-chain intent records, an internal listener runtime, and confidential evaluation infrastructure. Use this framing:AlphaEngine keeps strategy inputs confidential during submission and evaluation,
while scores, evidence artifacts, and final effects can become public.
Submission boundaries
Do not treat a candidate as:- a promise of capital,
- a live trading mandate,
- a public endpoint for operator runtime behavior,
- a bypass around scoring and eligibility gates.
Related pages
How the arena works
See the full beta loop.
API simulations
Run candidates against official datasets.
Privacy model
Understand what stays protected.
