Closed beta · invite only

An autonomous trading engine that shows its work.

16 specialized agents scan markets, build consensus, and trade your Alpaca account around the clock. You see every decision, every rationale, and every parameter the system tunes from its own results.

16
Specialized agents
5
Strategy tiers
100%
Decision audit trail
Live agent decision streamNow
  • BiotechAgentBUYGILD91%FDA PDUFA in 8d, ALPHA tier
  • TechAgentBUYNVDA88%ER beat, +12% premarket
  • PortfolioMgrTRIMMRK72%Tier 2 profit target hit
  • SentimentAgHOLDTSLA54%Mixed signal, awaiting cluster
  • QuantumAgEXITXYZ83%Conviction dropped <20
Every decision is logged, attributed, and reviewable.
Why us

Four things every other autopilot is missing.

01 — Engine

16 agents, not 1

Tech, Biotech, Healthcare, Energy, Finance, Bond, Commodity, Transport, Sentiment, Options Flow, Insider, Inverse-ETF, Alternative Data, Mean Reversion, Reinforcement Learning, and Quantum. Each specializes. Trades only fire on 3+ agent consensus, or a single agent at 75%+ confidence.

02 — Catalyst

ALPHA tier auto-trigger

When an upcoming FDA decision or earnings event lines up with multi-agent consensus, the engine opens a catalyst-sized position in the ALPHA tier. Risk-capped to 5% per position, 10% aggregate.

03 — Sizing

Dynamic Capital Reallocator

Scans every open position every 15 minutes, scores conviction 0-100 from agent confidence + P&L slope + tier strength + freshness. Proposes consolidations — trim losers, scale winners — never fragments capital.

04 — Learning

Self-tuning calibration loop

Replays closed trades through the same conviction model. If top-bin picks aren't winning enough, the system raises its own thresholds — written back to the adaptive layer as operator-attributed overrides. Closes the loop.

How it works

Connect Alpaca. Watch the bot. Approve its tuning.

  1. 1

    Connect your Alpaca account

    Paper or live. Credentials are validated against Alpaca, encrypted, and stored in Supabase. Switch modes any time from Settings — the engine restarts cleanly on either.

  2. 2

    The engine scans, decides, trades

    16 agents evaluate every signal every cycle. Consensus gates trades to your tier policy. Positions sized via Kelly, stops set broker-side as GTC orders.

  3. 3

    You watch, you tune, you override

    Every decision streams to the Overview page. Every position shows the agent that opened it and the bot's current view (HOLD / ADD / TRIM / EXIT). One click overrides anything.

  4. 4

    The loop closes itself

    Weekly AI post-mortem narrates what worked. Calibration validator proposes threshold adjustments. You approve — the engine retunes itself for next week.

What it isn't

We are honest about who this is for.

For
  • Sophisticated retail traders who already use Alpaca
  • Traders who want autonomy + auditability, not a black box
  • Traders comfortable with API-driven brokers + multi-week holds
  • Traders who'll review tuning recommendations weekly
Not for
  • Day traders chasing intraday scalps
  • People who want a copy-trading social feed
  • Anyone expecting guaranteed returns — markets don't work that way
  • Users without an Alpaca brokerage account
Status

Closed beta. One operator, one live account, real money.

Vaspera/Trade is in active development by the founder, trading on a personal live Alpaca account. We will not publish performance statistics until they reflect at least 90 days of post-calibration trading. The engine, the agent code, and the UI you would use are all live today — what's not live yet is a public, audited track record.

If you want in, write the founder directly with a sentence about how you trade today and what you'd use the system for. Access is gated by capacity, not arbitrary signups.

Want to see the bot's brain?

Closed beta. Email the founder, get a tour, decide.

Request access →