anvas turns market signals, conditions and actions into a visual canvas. Route a webhook, check the regime, size the risk — then send an alert or execute on the exchange. No scripting, no black box.
The builder and live executor run end to end on crypto futures and spot. Backtesting is in build. We are testing with a small closed group.
Starting with crypto perpetuals on one venue, done properly, before we widen the list.
We won't quote returns we haven't earned. Judge the tool on what it shows you, not on our numbers.
Drag sources, indicators, conditions and actions onto the board. Edges carry live values, so you watch a signal travel through your logic in real time instead of guessing from a log file.
Candle-based replay of the exact graph you built, with the venue's fee schedule and a slippage assumption you set yourself.
Every running graph, its open positions, its next decision and the exact node it is waiting on — on one screen. Graphs that only send requests show every call and its response code in the same place.
Most bots are a config screen bolted to someone else's opinion. anvas gives you the whole mechanism — and the receipts.
Preset bots hide the rule that lost you money. anvas shows the exact node, candle and value behind every fill.
Learn more →02Fees, slippage and walk-forward splits are on by default — the numbers you see are the numbers you get.
Learn more →03Drawdown ceilings, exposure caps and kill switches live in the graph — enforced by the engine, not by your discipline at 3am.
Learn more →Three shapes exist today: preset cloud platforms, code-in-the-cloud platforms, and self-hosted frameworks. Each trades away something. anvas is the fourth shape — visual authoring with the runtime visible.
| Preset cloud botstemplates and dropdowns | Code-in-cloudtheir editor, your Python | Self-hosted OSSyour VPS, your problem | anvasvisual graph, cloud runtime | |
|---|---|---|---|---|
| Authoring a strategy | Pick a preset, tune a few fields | Write and debug code | Code plus config files | Drag nodes; logic is the diagram |
| What you watch while it runs | A deal list and a status word | Your own print statements | Terminal logs over SSH | The graph itself, lighting up per candle |
| Answering “why did it buy?” | Support ticket, maybe never | Re-read your code, add logging, redeploy | Grep the logs you kept | Replayable trace: node, candle, value |
| Risk and guards | A stop-loss field per deal | Whatever you remembered to write | Whatever you remembered to write | Account-level guards enforced engine-side |
| Uptime and ops | Theirs | Theirs | Yours — VPS, restarts, upgrades | Ours, with a public status page |
| What it costs you | Tiers plus signal-marketplace upsells; some take a cut | Tiers by compute and data | Free software, paid server and your hours | Flat tier by workflows and venues. No cut, ever |
| Leaving | Strategy stays in their format | Your code, rewritten for their SDK | Nothing to leave | Export the graph as JSON; self-hosted build on the roadmap |
Categories, not brands — every platform differs and features move. If a specific comparison matters to you, ask us and we will answer straight, including where the other tool is the better choice.
You trade latency-sensitive or market-making strategies. We are candle and event driven, not colocated.
You need arbitrary Python with your own libraries in the loop. Write it and call us over a webhook instead.
Your keys must never leave your own hardware today. The self-hosted build is on the roadmap, not shipped.
The strategies below all need more than one behaviour, and need to decide for themselves when to change. That is the line between a bot and a graph.
We ask for trade permission and nothing else.
Keys are stored encrypted with a managed KMS, decrypted only inside the execution service, and we publish the egress IPs so you can allow-list us.
Drawdown ceilings and exposure caps are checked before an order leaves — not in a browser tab you closed.
TOTP on the account, per-device sessions you can revoke, and an email on every key change.
No. A graph only needs exchange credentials if one of its action nodes places an order. End the graph in an HTTP request instead — to a Telegram bot, a Discord webhook, your own server, anything with an API — and you never enter a key at all.
Three things: place or amend an exchange order, send an HTTP request — any method, any headers, any body, to any endpoint you can reach — or update state inside the graph that other branches read. The request node is how a graph talks to a Telegram bot, a Discord channel, your own service or anything else with an API. One graph can do several: alert first, then execute, or execute and report.
Yes, and it is a supported shape rather than a workaround. Point your TradingView alerts at an anvas inbound webhook, express the confirmation, cooldown and risk logic as nodes, and forward only what survives to whatever already executes for you. See signal routing and alerts.
Trade-only API keys, requested at the point you add an exchange action to a graph — never at signup. Read the detail on how it works.
No. The graph is the strategy. If you already have Pine or Python you keep it — anvas takes its output as a signal and adds the account-aware logic your script cannot see.
No. There are no DCA, grid or signal bot templates to configure. There are nodes, and the strategy is whatever you connect. A grid is something you can build, not a product you switch on.
Crypto perpetuals first, on one venue done properly, then the list widens — the roadmap has the order. Venue coverage limits the execution half only; signal routing works the same regardless of what you trade.
Private alpha through Q4 2026, with a small closed group. The waitlist is how you get in, and where you are in it is where you are in the queue.
One thing at a time, shipped to the alpha before the next one starts. No dates — you will see each step land.
The builder, the node library and live execution against real venues. Indicators, webhooks in and out, and the risk guards that sit in the graph.
Replay the same graph over OHLCV history with real fees and a slippage assumption you control. Walk-forward splits by default, and honest limits printed on every run.
Follow a running bot with your own sizing and your own guards — the graph stays visible, so you know what you are copying.
Tell us what you trade and how you automate it today. We open seats in small batches and start with the setups closest to what already works in the alpha.