By Keel Research Team · Updated Sep 19, 2026
21 explainers on how Hyperliquid perpetuals work, how to backtest a strategy without fooling yourself, how to size a position, and where AI agents fit. Each one quotes its numbers from the calculator it links.
How perps, funding and liquidation work on the venue.
What perps are, how funding anchors the price, and how they differ from spot, options and dated futures.
What a funding rate is, why perps need one, and how to read positive and negative rates on Hyperliquid.
The isolated-margin formula, maintenance-margin tiers, cross-margin behavior, and how to stay out of the zone.
The tests that separate a real edge from a fitted one, with the calculator for each.
What a backtest measures, the mistakes that fake performance, and how to validate before risking capital.
Validate strategy parameters across rolling out-of-sample windows instead of one split.
Train/test/holdout vs k-fold vs walk-forward, and why OOS Sharpe matters more than in-sample.
Block-bootstrap resampling and confidence intervals on Sharpe and max drawdown.
The two failure modes, the PBO and DSR diagnostics that catch them, and a red-flag checklist.
Combinatorially-symmetric cross-validation: the chance your best in-sample strategy underperforms out of sample.
Adjust a backtest Sharpe for the number of trials and the shape of the return distribution.
Same formula, different risk denominator: when each is the right metric and what their ratio says.
Ten items to clear before running a crypto strategy, from walk-forward to live parity.
How much to put on, and how to read the damage when it goes wrong.
Risk-percent, Kelly and volatility-targeted sizing compared, with the math behind each.
The formula, what it computes, and why fractional Kelly is what traders actually run.
Scale exposure inversely to recent volatility: what it does for Sharpe and drawdowns, and what it costs.
The largest peak-to-trough loss an equity curve has taken: formula, why it beats volatility, real examples.
Single-stock and index perpetuals on Hyperliquid via HIP-3.
Where an LLM helps with systematic trading, and where it does not.
The open standard that lets AI agents call external tools, and why it matters for trading.
An AI agent that builds and runs a strategy with you: where it works, where it fails.
LLMs propose strategies well and validate them badly; the workflow that keeps the engine in charge.