Open data · Open source · Open track record

The intelligence terminal
that shows its work.

Bloomberg-grade analytics assembled from open data, and free because the data was already free. Bring your own sources, plug in any MCP server, compose the dashboard you actually want.

And every call our AI makes is timestamped before the outcome exists, then scored in public — win or lose. Because a track record you can’t audit isn’t a track record. It’s a screenshot.

16 of them need no API key. Free tier, no card, no trial timer.

The proof layer

A track record you can’t edit.

Every prediction our AI makes is written down the moment it’s made — entry, targets, stop, timestamp — before the outcome exists. Then it’s scored against real price data and published. Wins and losses.

That’s the whole trick, and it’s why nobody else does it. You cannot go back and collect a timestamp after the fact. The record either exists from t=0 or it doesn’t exist at all.

We run it on ourselves first. Next, we run it on everyone who sells you a call.

Audit the raw ledger
guard_engine · liverecording
Verified win rate46.8%Hit target before stop loss
Calls resolved2,381Scored, not cherry-picked
Average win+59.84%
Average loss46.11%We publish this one too

0 predictions are open right now. Their outcomes are already determined by targets we committed to before entry — we can’t move the goalposts once we see the chart.

Bring your own everything

Your API keys, your CSVs, your wallets, your private MCP servers. Your data never has to touch our infrastructure to be usable in our terminal — and when a provider's license forbids us from serving it to you, you can still bring your own key and see it anyway.

Compose, don't configure

Widgets snap into a grid you arrange. Every one of them is a plain HTTP endpoint returning JSON, so a widget you build is a widget anyone can run — including inside OpenBB Workspace, because we speak their spec rather than inventing our own.

Prove, don't claim

Every signal on the platform carries a timestamp from before its outcome was known, and a score from after. Ours included. If a number here has no receipt, it doesn't ship.

The open data layer

Everything below is already free.
Somebody just had to plug it in.

Bloomberg costs about $30,000 a year. A striking amount of what it sells you is assembled from sources that are public, keyless, and in several cases explicitly public domain. We list every one we speak to, along with what its license actually permits — including the ones we’re not allowed to serve you, so you can bring your own key.

27
Sources wired
16
Need no API key at all
15
We can serve you directly
6
Speak MCP natively

Prices, order books, funding, and open interest — straight from the exchanges, no vendor in between.

  • BinanceNo key

    Spot + perp klines, depth, funding, open interest

    Deepest free OHLCV history available anywhere.

  • CoinbaseNo key

    Spot, L2/L3 order book, candles

  • KrakenNo key

    Spot, OHLC, order book, spread

  • Perps, funding rates, open interest, options

  • HyperliquidNo keymcp

    Perp book, candles, and every address's live positions

    Positions are public per-address. Whale tracking that costs money elsewhere is free here.

  • CoinGeckoNo keymcp

    15k+ assets, market caps, categories, coin metadata

    Terms forbid redistribution. We use it for ID mapping and metadata, never as a data spine.

Plug and play

A widget is just a URL that returns JSON.

We didn’t invent a widget format. We adopted the one OpenBB already uses, which means a widget you build for SOLYZER also runs in OpenBB Workspace — and every backend built for OpenBB runs here.

No SDK, no lock-in, no language requirement. Describe your endpoint in a widgets.json, point us at it, and it’s a panel in your grid. That is the entire integration.

widgets.json
{
  "kol_track_record": {
    "name": "Caller Track Record",
    "description": "Scored calls, resolved vs open",
    "endpoint": "/widgets/kol-record",
    "type": "table",
    "params": [
      { "paramName": "handle", "type": "text" }
    ],
    "data": {
      "table": {
        "columnsDefs": [
          { "field": "called_at" },
          { "field": "token" },
          { "field": "outcome" },
          { "field": "pnl", "formatterFn": "percent" }
        ]
      }
    }
  }
}
bash
# Pull SOLYZER into your own agent
$ claude mcp add solyzer https://mcp.solyzer.ai/sse

  ✓ on_chain.holders          ✓ proof.win_rate
  ✓ on_chain.liquidity        ✓ proof.caller_record
  ✓ social.farcaster_signal   ✓ market.funding_rates

# ...or push your data into ours
$ solyzer connect mcp ./my-private-alpha-server
Model Context Protocol

Your agent is a first-class user.

MCP runs both ways here. Point Claude, Cursor, or your own agent at our server and it can query on-chain state, social signal, and the proof ledger as tools.

Or go the other direction: connect your MCP server — a private alpha feed, an internal database, a broker — and it becomes a source inside the terminal, sitting next to the public ones. It never leaves your machine unless you say so.

This is what “vibe trading” should have meant all along: you describe what you want to know, and the terminal assembles it from whatever it can reach.

What we build together

Price data is a commodity.
Memory is the moat.

Anyone with a credit card can buy the same prices we use. What can’t be bought is a record of who said what, and when, before anyone knew who was right. Incumbents pay analysts to label wallets by hand — which is exactly why they have pricing power, and exactly why an open set erodes it.

So we publish the labels. All of them, CC0, on Hugging Face. Giving them away is the strategy: it makes SOLYZER the place where the labeling happens, and a community’s archive compounds while a vendor’s moat depreciates.

  • Prediction Ledger

    live

    Every AI call we make, hashed and timestamped before the outcome exists, then scored against real price data.

    Ground truth is destroyed the moment it becomes obvious. This can only be collected forward, never backfilled.

  • Caller Track Records

    building

    The same scoring, applied to public influencers and alpha groups. Deleted calls stay on the record.

    Nobody sells this, because in the incumbent model the callers are the customers. Only the aggrieved party builds it.

  • Wallet & Entity Labels

    building

    Exchange wallets, market makers, funds, treasuries, bridges, MEV bots, insider allocations.

    Labeling is manual and cumulative. It's why Nansen has pricing power — and why an open set, seeded from public labels, erodes it.

  • Rug & Deployer Registry

    planned

    Not just 'is this a honeypot' — which deployer rugged before, funded by whom, audited by whom.

    Deployer recidivism graphs are devastatingly effective and nobody publishes them openly.

  • Unlock & Vesting Schedules

    planned

    Token emission and cliff schedules, read from the vesting contracts themselves.

    Paywalled by incumbents, yet the underlying data sits in public contracts. Tedious, not hard.

  • Narrative Taxonomy

    planned

    Which tokens belonged to which narrative — and, critically, when they were added.

    Every current sector map is survivorship-contaminated, which makes honest rotation backtests impossible. The 'when' is the whole asset.

Crypto is where we start,
not where we stop.

We began on-chain because that’s where the data is most open and the dishonesty is most expensive. But the method — open sources in, timestamped and scored claims out — isn’t about crypto at all.

Every SEC filing since 1993 is public domain and free. So is FRED, the ECB, the World Bank, the EIA. The same terminal that scores a caller on a memecoin can score an analyst on an earnings call.

Vibe trading is the first surface. Vibe analytics is the platform.

Free because it should be.

The terminal, the data catalog, the widgets, the MCP server, and the proof ledger are free and stay free. We’ll charge for depth, history, and enterprise deployment later — to people who can afford it, for things that genuinely cost us money. Not for the right to see public data.