Quick answers about AlphaFlowSeven's AI council, portfolio simulation, pricing, and the platform's data infrastructure.
Yes. New users get up to three AI councils on a curated set of tickers — completely free. No credit card required. You only pay if you scale beyond the free tier, and even then billing is usage-based rather than a subscription.
No. AF7 is a signal platform with a built-in portfolio simulation. Every signal can be routed into a simulated position so you can evaluate the council's calls without committing real capital. There is no on-exchange execution and no exchange account is required.
Each of the six agents in a council can run on a different model. Supported providers are Anthropic Claude (Sonnet, Opus, Haiku), OpenAI (GPT-4o and the GPT-5 family), Google Gemini (2.5 Flash and 2.5 Pro), xAI (Grok 4), DeepSeek (V3.2), and Moonshot AI (Kimi K2.6).
Three reasoning tiers. Tier 1 — Technical, Momentum, and Risk analysts — read the market in parallel. Tier 2 — Bull and Bear advocates — build the strongest case for each direction from Tier 1 findings. Tier 3 — the Lead Trader — synthesises everything plus its own last four prior decisions and renders the final BUY / SELL / HOLD / CLOSE call with a conviction score.
Configurable per council. Common cadences are 15 minutes, 1 hour, 4 hours, and 1 day. Between full council meetings, a lightweight "check-in" mode runs the Lead Trader on its own at a shorter interval for moment-to-moment HOLD / CLOSE calls within the directional envelope set by the last full meeting.
A full meeting runs all six agents across three tiers. A check-in runs only the Lead Trader, briefed with the standing guidance from the last full meeting plus the latest market data, to catch moment-to-moment changes that don't warrant a full re-deliberation.
Live OHLC streams from Binance International over WebSocket, with automatic Kraken fallback after three failed connections or a region block (HTTP 451). REST backfills cover any gaps. RSI, ADX, +DI, -DI, and ATR are computed on the same timeframe the council is configured for.
A freshness guard fails the council run loudly if the newest OHLC candle for the requested (ticker, interval) is more than three interval-lengths old. Stale data never reaches the agents — the run is marked failed and the next scheduled run picks up once data is flowing again.
Any ticker available on Binance International or Kraken today, with broader coverage (including tokenised equities) on the roadmap. The free tier ships with curated councils; paid users can spin up a council on any ticker and timeframe combination.
Each simulated position carries a take-profit percentage and a trailing-stop percentage. The trailing stop ratchets in the favorable direction only — it never moves against the position. Optional emergency stops close the position regardless of trailing-stop state. Monitoring runs every 60 seconds.
Accuracy depends on the configured agents, models, and market conditions — there is no fixed number. The Council Evaluator runs an LLM auditor over a council's recent calls and proposes prompt edits, so you can iterate. AF7 publishes every simulated outcome in the dashboard, including losing ones.
Yes. Every completed run is stored as an AICouncilExecution row with the final call, conviction score, agent responses, and links to any simulated positions it generated. Run the Evaluator to get an LLM-generated assessment with recommended prompt changes.
Four: AI Council calls (Lead Trader output), candle pattern matches, grid-level crossings, and external TradingView webhooks. All four sources flow through the same processing pipeline, so handling is identical regardless of the source.
Yes. Every model — strategies, councils, positions, signals — is scoped to your user via a tenant system that auto-injects user_id on insert. Other users cannot see your configurations, calls, or simulated positions.
Account top-ups are processed in-app. See the Account page once logged in. Costs are debited per council run, per check-in, and per evaluation, based on the provider's per-token rate plus a small platform margin.
The fastest way to evaluate AF7 is to sign up and run a council yourself — free for three councils, takes a couple of minutes, no card required.