Skip to content
中文
Seris is open sourceRelease notes

Docs

Strategies and backtests

Ask Seris for a TypeScript strategy, review the source before it is saved, then backtest it locally on real historical candles from Binance spot and Hyperliquid perpetuals, including HIP-3 markets like xyz:NVDA. Each run reports return, drawdown, fees, trade markers and an equity curve, records its data venue, candle range, fees and slippage, and stays in Strategies.

Draft a strategy

Describe the idea in Chat, for example: Write a BTC strategy that enters on a moving-average crossover and uses an ATR trailing stop. The agent drafts strategy.ts with its parameters and a SKILL.md that documents it. Strategies import types and indicator helpers from @seris/strategy.

Review and save

Saving a strategy goes through approval: the card in Chat shows the source before any file is written. Seris compiles and smoke-runs the strategy as it saves it. Saved strategies live in skills/strategies/ in your Seris data directory and carry over when you upgrade.

Seris includes ma-trail-stop, a moving-average strategy with an ATR trailing stop, so you can try a backtest right away.

Run a backtest

Ask in Chat, for example: Backtest ma-trail-stop on BTCUSDT 1h for the last 30 days, or: Backtest ma-trail-stop on Hyperliquid xyz:NVDA 1h for the last 14 days.

  • Binance spot: pairs such as BTCUSDT or ETHUSDT
  • Hyperliquid perpetuals: crypto such as BTC or ETH, and HIP-3 markets such as xyz:NVDA

Anything you leave out uses a default: the strategy's own timeframe, 90 days of history, 10,000 in starting cash, and 5 bps each for fees and slippage. Mention a different value in your request to change it, including parameter overrides.

Only closed candles enter a backtest. While a run is in progress, the agent can follow its status and tell you when it finishes.

Read the results

  • Total return, Sharpe and Sortino
  • Max drawdown
  • Win rate and profit factor
  • Number of trades and fees paid
  • Trade markers on the price chart and the equity curve
  • Data venue, actual candle range, and the fee and slippage assumptions used

Open Strategies to see each strategy's code, documentation, parameters and every backtest run. Ask the agent to compare earlier runs before you change a parameter.