Solution · Trading

Put every market signal in context.

A filing arrives, the market moves, and your team needs to reconstruct what was known before the decision. Minds helps trading researchers connect events, filings, and prior analysis in a dated record they can revisit.

A decision at a point in time
Illustrated workflow

A sample event clock

What was known at 10:30?

AS OF10:30
Earlier eventsAfter the cutoff
  1. 09:30
  2. 10:00
  3. 10:30
  4. 11:00
  5. 11:30
Filing AIncluded

Prior filing

Available before the selected event.

Published
10:05
Known to the agent
10:06
Both clocks precede 10:30.
Context at 10:30

Only what was available then.

A
Prior filingPublished 10:05 · known 10:06

Published ≤ cutoff
Known ≤ cutoff

Filing BExcluded

Later filing

It remains outside this earlier context.

Published
10:45
Known to the agent
10:46
Both clocks follow 10:30.
A record of the market and your reasoning

Keep the timeline behind the analysis.

Keep a filing’s publication time separate from when your system received it. Link it to the event, your working hypothesis, and the source material. That context helps a reviewer examine a historical decision without treating later information as if it had been available earlier.

01

Events

Process the incoming signal.

A spiking neural network processes changing inputs over time. Supply encoded market features and inspect the network’s responses.

02

Patterns

Inspect what the model is doing.

Read the output of each processing step and compare it with the input sequence as you evaluate an event-processing model.

03

Research

Bring filings into the analysis.

Query registered data sources with SQL and connect the results to relevant events and prior research.

04

Time

Separate the event from its arrival.

Track both when a fact applied and when it was recorded. Historical analysis needs to consider when each input became available.

05

Sources

Keep uncertainty visible.

Track unverified sources and inferred information through the reasoning record so a derived finding can be reviewed in context.

06

Updates

Review a change before relying on it.

Keep proposed knowledge separate during review and save checkpoints for supported learning components. Evaluate a proposed strategy update in your own research workflow before using it.

Developer example

Connect an event to your research workflow.

This HTTP example builds a network, processes sample input, and queries a registered research source. Enable SNN and analytics, and use a signed capability that permits those operations. The namespace must match that credential. Your application supplies the market feed and evaluation process.

Build and step a network · query research · TypeScript HTTPtypescript
// Run server-side with credentials scoped to this Mind.
const endpoint = process.env.AKASHA_URL;
const capability = process.env.AKASHA_CAPABILITY;
const namespace = process.env.AKASHA_NAMESPACE;
if (!endpoint || !capability || !namespace) {
  throw new Error("Configure your Mind connection and authorized namespace");
}

async function post(path: string, body: unknown) {
  const headers: Record<string, string> = {
    "Content-Type": "application/json",
    "x-akasha-capability": capability!,
  };
  if (process.env.AKASHA_TOKEN) {
    headers.Authorization = "Bearer " + process.env.AKASHA_TOKEN;
  }
  const response = await fetch(new URL(path, endpoint), {
    method: "POST", headers, body: JSON.stringify(body),
  });
  if (!response.ok) throw new Error("Mind request failed: " + response.status);
  return response.json();
}

// A small, untrained network to demonstrate the request sequence.
const network = await post("/v1/snn/network/build", {
  namespace,
  spec: {
    label: "market-input-example", seed: 42,
    ensembles: { input: { n_neurons: 100, dimensions: 1, radius: 1.0 } },
    connections: [],
  },
});
const identity = {
  network_id: network.network_id,
  namespace: network.namespace,
  version: network.version,
};
await post("/v1/snn/network/compile", identity);
const { output } = await post("/v1/snn/network/step", {
  ...identity, input: [0.5], // Replace with your encoded input.
});

// Assumes a registered resources table with both timestamp columns.
const { rows } = await post("/v1/analytics/sql", {
  sql: "SELECT title, published_at, known_at FROM resources " +
    "WHERE published_at <= '2026-03-15' AND known_at <= '2026-03-15' LIMIT 10",
});
Research controls

Make the inputs and changes reviewable.

Time-aware records and source tracking help you inspect how an analysis was produced. The workflow shown here illustrates research infrastructure; it provides no evidence of trading returns, predictive accuracy, or live trade execution.

At a glance
Event processing
Spiking neural network inputs and state
Historical context
Track valid time and when data was recorded
Learning review
Check new knowledge before promotion
Recovery
Checkpoints for supported learning components
HTTP requests on this page
Network
POST /v1/snn/network/build · compile · step
Research query
POST /v1/analytics/sql
Access
Signed capability and matching namespace
Time model
When a fact applied and when it was recorded

Build a clearer record behind your research.

Connect a Mind to your research workflow and preserve the events, sources, and decisions you want to revisit.