Pay-per-call · USDC · no account, no API key
A fixed rule set shortlists S&P 500 names on two documented setups — post-earnings drift and momentum. A language model with live web search then reads the news on each and ranks five for the next five sessions. Each one carries the setup it expresses, a written thesis, the risk most likely to break it, a confidence level, and the sources it actually read. Your agent pays a cent and gets the JSON.
GET /v1/picks/today HTTP/1.1 Host: screen.rateitall.co.za
HTTP/1.1 402 Payment Required PAYMENT-REQUIRED: eyJ4NDAyVmVyc2lvbiI6Miwi… // 0.01 USDC · exact scheme
HTTP/1.1 200 OK { five ranked names, with reasoning and sources }
What you get
A single call returns the whole day's screen. Each pick names the company, its sector, the thesis in two to four sentences, the single most material risk to that thesis, a confidence level the model sets honestly, and every source URL it consulted. The response also reports how the five split across sectors, so concentration is visible rather than buried.
{
"run_date": "2026-10-01",
"sector_breakdown": { "Industrials": 2, "Health Care": 2, "Financials": 1 },
"picks": [
{
"rank": 1,
"ticker": "XXXX", // always from the published universe
"company": "Example Corp",
"sector": "Industrials",
"thesis": "Setup: momentum — <2–4 sentences, every claim drawn from a cited source>",
"key_risk": "<the specific thing most likely to make this wrong>",
"confidence": "medium", // low | medium | high
"sources": [ { "title": "…", "url": "https://…" } ],
"entry_price": 000.00,
"entry_asof": "2026-10-01T20:00:00.000Z"
}
// … four more
],
"method": { /* see below */ },
"disclaimer": "…"
}Also available: /v1/picks/{date} for any past session, and /v1/performance for the full per-pick history behind the figures below.
Method
Every weekday after the US close, a fixed rule set builds a short list from the S&P 500 using two setups with decades of published evidence at a multi-week horizon: post-earnings drift (a positive earnings surprise reported in the last three sessions) and intermediate momentum (the top fifth of the universe by 13-week return, trading within 10% of its 52-week high). Names from the previous five sessions are removed so the list rotates. That candidate list is deterministic and stored for every session.
A language model with live web search then reads the recent news on each candidate — the part the numbers cannot see — and ranks five for the next five trading sessions, stating which setup each one expresses. Two independent passes are run; names both agree on rank first, and a name only one pass chose cannot be marked high confidence. The ranking is not deterministic. Every run is logged with its candidate set, model passes and cited sources, so any list can be traced back to what produced it.
Live method: loading…
Setup: post-earnings drift — or Setup: momentum —, so you can see what each pick is an expression of.The full machine-readable method statement is free to read at /v1/method.
The record so far
| Session | Picks | Average return |
|---|---|---|
| Loading… | ||
Returns are simple price changes from the close recorded on the session date to the most recent mark. They exclude dividends, costs, slippage, taxes and position sizing. A session whose entry prices were not taken at the close is excluded entirely rather than counted, because counting it would flatter this table. Per-pick detail, including any excluded session and the reason, is available to buyers at /v1/performance. The aggregates on this page come from /v1/stats, which anyone can read.
How to buy
There is no account to open here and no API key to look after. The listing is sold through PowaHub, and your assistant pays per call in USDC at the moment it asks.
If the card does not load, go straight to the listing on PowaAgent.
Works with ChatGPT, Claude, Gemini and other assistants that support tool use. Your assistant pays per call in USDC.
Integration
The method description and this record are deliberately free to read, so you can judge the service before you pay for it.