GitHubGet access
FeaturesKryllOS SystemDrag-and-Drop EditorBacktest EngineMarketplaceKryllOS Skill
KRL TokenProduct Status
ResourcesBlogFAQWhitepaperMedia kitEtherscanCoinGeckoCoinMarketCap
GitHubGet access

A project is not a good fit on its own. It is one for a given strategy. Build yours once in KryllOS, run it on dozens of pairs with Multi Backtest, then open the reports to see which markets your rules hold up on.

A good fit depends on the strategy

The same coin can suit one strategy and wreck another. The useful question is not which token will take off, but which markets your rules were written for.

  • A trend strategy wants long runsIt rides a pair that climbs or falls for weeks, and keeps getting stopped out on one that zigzags sideways.
  • A rebound strategy wants a rangeBuying the dip works on a market that swings back and forth. In a steady slide, every dip has another one under it.
  • Each pair has its own temperamentVolatility, volume, the hours when it moves: two coins with similar charts can react very differently to the same rule. Only a backtest on each one shows how.

One strategy, every pair you want

No script, no spreadsheet. Multi Backtest is a KryllOS app: a queue of backtests that runs on your machine.

  1. 01Add to Multi BacktestFrom the Run menu of the editor, your strategy joins the queue with its settings.
  2. 02Duplicate, change the pairRight-click the row, Duplicate, then Edit launch params: exchange, pair, dates. One row per market you want to test.
  3. 03Run AllKryllOS runs four backtests at a time and queues the rest. Candles are downloaded once and stay on your machine.
  4. 04Open each reportDouble-click a finished row: the trades on the chart, the log of every decision, the drawdown.

“Test my Trend rider on every USDT pair, and show me where it holds.”

Say it to your KryllOS agent. With the skill built into KryllOS, Claude Code, Codex or the agent you already use lists the pairs on your exchange, launches the backtests on your machine and reads the reports for you. You get a shortlist, and the reports to check it.

Turn your AI into a KryllOS agent

Then tell a real fit from a fluke

One good report on one pair proves little. Before a token makes your list, send it back through the queue.

  • Another periodDuplicate the row and move the dates. A real fit holds up over more than one stretch of market, not only the one you picked.
  • Enough tradesThree trades are an anecdote. Keep the pairs where your rules fired often enough for the result to mean something.
  • A market you can tradeVolume and spread decide whether the fills in the report would happen for real. A thin pair can look great on candles alone.
  • Then paper tradingThe file you backtested runs as is in paper trading, on the live market, without real money. That is where a candidate earns its place.

Build once. Test it everywhere.

Install KryllOS, queue your strategy on every pair you are curious about, and keep the ones where it holds.

Get access to KryllOSHow Multi Backtest works