> ## Documentation Index
> Fetch the complete documentation index at: https://docs.hizz.io/llms.txt
> Use this file to discover all available pages before exploring further.
# AI Trading strategies
> How Hizz strategies are structured, backtested, and deployed, plus the current published catalog.
A Hizz strategy is a versioned set of deterministic rules: signal filters,
entry conditions, risk controls, sizing, allowed sides, and an execution-cost
assumption. The backtester and deployed agent compile the same definition, so
the live agent does not replace the strategy with an opaque chat response.
## Published strategy catalog
| Strategy | Market / clock | Style | Core rules | Intended use |
| ----------------------------- | -------------- | --------------------- | ------------------------------------------------------------------------- | --------------------------------------------------------------- |
| **Trend Trail · HYPE Runner** | HYPE-USD · 4h | Medium frequency · 2x | EMA10/30 trend, 30bps chop guard, 120bps hard stop, 150bps trailing stop | Stay aligned with a larger HYPE trend and trail favorable moves |
| **Micro Trail · BTC 15m** | BTC-USD · 15m | High frequency · 2x | EMA5/20 trend, 5bps chop guard, 35bps hard stop, 45bps trail, 24h timeout | Verify the full live pipeline quickly with a small allocation |
Micro Trail is deliberately fast and fee-sensitive. It is useful for
checking the arm → run → open → manage → close pipeline, not evidence of a
durable long-term edge. Tight stops can churn in a ranging market.
## How the strategy blocks work
Indicators define when a market is eligible and when an entry triggers.
Examples include EMA separation, RSI regime limits, MACD confirmation, and
a chop guard that rejects flat conditions.
Each strategy defines its initial stop and any take-profit, trailing stop,
reversal, or maximum-hold exit. These are strategy controls, not guarantees
against gaps, slippage, liquidation, or API failure.
Allocation is USD margin per trade. Notional is allocation multiplied by
leverage, so both fee dollars and market exposure scale with leverage.
The current built-in backtests model **10bps per fill**: 5bps Hizz plus a
5bps base-tier Strike taker estimate. Live accounting replaces that estimate
with the actual Strike fee and the Hizz builder fee attached to each fill.
## Reading a backtest
The strategy detail page reports return, trade count, win rate, maximum
drawdown, Sharpe ratio, and the historical equity curve. Use them together:
* **Return** without drawdown can hide an unacceptable path to the result
* **Win rate** says nothing about the size of wins versus losses
* **Trade count** matters because a tiny sample is easy to over-interpret
* **Sharpe** summarizes historical risk-adjusted consistency, not future safety
* **Fee assumptions** must be close to the fills you expect in live mode
Backtests are recalculated from the current strategy definition and market
series. They are historical simulations, not a promise that the same fills,
slippage, liquidity, funding, or return will occur live.
## Create a private AI strategy
Open **My Strategy → Create AI strategy** and describe the market, direction,
clock, signal, exits, size, and leverage you want. If you omit risk rules, Hizz
adds conservative defaults for review.
A useful prompt is specific:
```text theme={null}
Trade BTC-USD on the 1h clock. Go long only when EMA20 is above EMA50
and RSI is below 70. Allow shorts on the reverse condition. Use 2x leverage,
a 1.2% hard stop, a 1.8% trailing stop, and no more than 25% of capital.
```
Always inspect the compiled rules and backtest before choosing **Deploy**.