Is your edge real?
See for yourself.

Active traders who want to trade systematically, save their time and automate the routine work.

Releases — 1–2 times a month. The roadmap is shaped together with active users, requests from tickets go into the work. The newest major addition — already live — is Veskald Orion, three AI strategy agents: Bellatrix (a global assistant on every page of the platform — answers any question about the product, your strategies, trading and your stats, and reality-checks existing strategies), Rigel (quick prototypes with pro safeguards), Betelgeuse (deep research in a dialogue with you — it asks rather than assumes, builds and validates the strategy, and digs until it finds durable edge — or tells you plainly the strategy is dead, instead of burning tokens).

No. Everything works through the visual builder — you pick indicators and conditions from a menu. If you want to go deeper — there are AND/OR groupings and multi-timeframe rules. No code at any stage. Missing an indicator? Ask — we build it, and it becomes part of the platform for everyone.

Technically — no, and by design. We only accept read-only API keys; any key that carries trading or withdrawal rights is rejected on submission and never stored. Even if you wanted to authorize us, you could not: on most exchanges a trading key only works from a whitelisted IP, and our servers sit behind Cloudflare with no public address to whitelist. And for automated execution the trading keys never reach us at all — they run inside the open-source bot on your own server.

Not "better" — complementary. By hand you will try 10–20 combinations. The algorithm goes through thousands. But the main thing is that it optimizes not for maximum profit (one lucky trade means nothing) but for the smoothness of the equity curve. That protects against the overfitting almost every self-made quant experiment falls into.

On trades up to $300 the gap with real trading is under 0.5%. What is accounted for: commission (you set it yourself based on your exchange tier), funding every 8 hours, slippage from liquidity, and on large volumes the impact of order size on execution price. The backtest works with market orders, and — importantly — the backtester core is open: parameters can be tuned to your trading profile, you see exactly how each trade is calculated. You can compare expectation with reality in the Execution Log.

Strategies are stored in the Veskald database — that’s what lets backtests and the optimizer run on our compute. If you want your logic unreadable, enable strategy encryption: a two-key scheme, your key + our key, either one useless on its own. Your key is never written to disk. An encrypted strategy cannot be decrypted without you — including by the Veskald team.

Stable today: Binance, Bybit, Kraken, KuCoin. In development: OKX, Coinbase. If you need another exchange (Bitget, BingX, Hyperliquid, dYdX and others) — write to us, we usually ship it in 1–3 weeks.

A saved strategy you can use as a starting point. Take a ready template from the library or save one of your own. When you build a new strategy on top of a template, you work with a copy — the original stays intact. Experiment without fear of breaking what works. The public library is strategies we research and publish ourselves — the logic is fully open: verify it, change it, make it yours.

On servers in the EU (Hetzner, Germany). GDPR applies in full — details are in the Privacy Policy. Strategy encryption with a two-key scheme is available for your trading logic; exchange trading keys are not passed to Veskald at all.

No lock-in. On the Trader plan and above, every strategy emits real-time events over a WebSocket: you register your IP, get an IP-bound token, and our backend streams events straight to your endpoint — build any execution system of your own on top of Veskald. Alternatively, automated execution runs through an open-source bot you host yourself. Trading keys never reach us either way. Integration options for teams are on the B2B page.

Yes: rebuild its rules in the visual builder — most sold strategies are a handful of conditions — then run the full validation: honest costs, IS/OOS split, Monte Carlo, cross-asset test. You’ll see what it’s actually made of. Bellatrix will walk you through the results.

No. The public template library comes from our own research agent, which works around the clock on the platform side. It remembers every strategy it has ever built — the ones that survived and the ones that died — so it knows what not to repeat; it tunes on in-sample data and spends the out-of-sample run once, never iterating on it; we decide which of its strategies to publish, always with the logic open. Its knowledge base takes in everything that clears a basic sanity check, profitable or not: what failed is worth as much as what worked, and a base built from survivors alone would teach survivorship bias. Betelgeuse is a different agent with a different job: it researches together with you, inside your account, reading the same knowledge base — it writes nothing back to it. Your strategies play no part in any of this: they belong to you, not to us, and the platform touches them for one reason only — running the service for you: backtests, optimization, launch. Turn on encryption, and no one can read them without your key — including us.

No. Validation is an in-sample / out-of-sample split with an untouched holdout, a cross-asset run on up to 4 other assets with their own date windows, Monte Carlo (trade permutation) and bootstrap. The optimizer splits in-sample / out-of-sample itself and targets a smooth equity curve rather than the peak.

Not as a script: strategy logic here is rules, not code or a model, so every strategy stays readable and testable. If an indicator you need is missing, ask — we build it, and it becomes a regular indicator of the platform for everyone. 100+ indicators today: trend, momentum, volatility, volume, candle patterns. Your own code lives after the signal: the bot is open source and runs on your machine, so you can filter its signals with any model of yours or replace the bot with your own executor.