TL;DR: A no-code trading strategy builder lets you describe when to enter, how much to risk and when to exit without writing code. In Trigr, that means wiring one trigger, optional filters, a signal and a risk node into a graph, then backtesting it on point-in-time data with explicit costs before moving to a paper agent and, if it still holds up, a live Hyperliquid agent.
What is a no-code trading strategy builder?
A no-code strategy builder is a visual editor that turns trading rules into something a backtest engine and an execution system can run. Instead of writing Python or Pine Script, you pick conditions from a menu, connect them, and set parameters.
The value is not just speed. When the structure is fixed, whole classes of bugs disappear: you cannot accidentally reference tomorrow's close, forget to apply fees, or write an exit that never fires. The tradeoff is expressiveness. A good builder is honest about what it can and cannot represent.
Trigr's builder, called Studio, is aimed at perpetual futures on Hyperliquid: 43 certified crypto perps plus TradFi markets such as gold and the S&P 500 via Hyperliquid HIP-3. The same graph you draw is what gets backtested, what runs on a paper agent, and what a live agent executes. (In the current beta, TradFi markets can be backtested and run on paper agents, but not traded live.)
How does Trigr's node graph work?
Every Trigr strategy is a small pipeline of nodes. Each node has a structural kind, which says where it sits, and a data source, which says what it reads.
| Node kind | Role | How many |
|---|---|---|
| TRIGGER | The primary condition that fires the entry pipeline | Exactly one |
| FILTER | Extra conditions that must also hold (trend, volatility, time, cross-asset confirmation) | Zero or more |
| SIGNAL | Combines trigger and filters into a long or short signal | One |
| RISK | Size, leverage, take-profit, stop-loss, direction, trailing and time exits | One |
| EXECUTE | Turns the resolved signal into an order on the chosen venue | One |
The data sources are technical indicators (RSI, EMA, SuperTrend and many more) and API data: funding rate, open interest, long/short ratio, liquidations, DVOL, hashrate, Santiment on-chain and social series, FRED macro series, VIX and its term structure, rates, EIA oil stocks, CFTC positioning, the crude futures curve and ORATS implied volatility for the TradFi underlyings. The capability catalog lists every asset, timeframe, indicator and data source with its exact parameters.
Supported timeframes run from 5 minutes to 1 month: 5m, 15m, 30m, 1H, 4H, 1D, 1W and 1M.
Cross-asset nodes
Any node can read a different market than the one you trade. You might trade SOL but only allow longs when BTC is above its 200-period EMA, or gate an ETH entry on BTC funding. The engine resolves these references point-in-time: if the other input is on a coarser timeframe, its last completed value is carried forward and never read ahead.
How do you choose a trigger and filters?
The trigger answers "what event makes me consider a trade?" Filters answer "under what conditions do I trust that event?" Keeping the two separate makes a strategy easier to reason about and to test.
Picking a trigger
Good triggers are specific events rather than states. "RSI crosses below 30" is an event; "RSI is below 30" is a state that can stay true for many bars. Common trigger families:
- Mean reversion: an oscillator reaching an extreme, or price closing outside a Bollinger band.
- Momentum and breakout: a moving-average cross, a Donchian-channel breakout, a SuperTrend flip.
- Derivatives data: funding reaching an unusual level, open interest jumping, or a liquidation spike. See funding-rate and open-interest strategies for ideas to test.
Because there is exactly one trigger per strategy, you are forced to decide what the core idea is. If you find yourself wanting three triggers, you probably have three strategies.
Adding filters
Filters reduce the number of trades. That is the point: most raw triggers fire in regimes where they do not work. Useful filter types include:
- Trend regime: price above or below a long moving average on a higher timeframe.
- Volatility regime: ATR or DVOL within a range, so you skip dead markets or chaotic ones.
- Cross-asset confirmation: BTC's trend for an altcoin strategy, or the VIX level for a risk-on idea.
- Macro context: slow-moving series such as financial conditions or credit spreads. The article on macro data in crypto strategies covers the point-in-time pitfalls.
Each filter you add is also another parameter you could have chosen differently. That matters later, when you judge whether a backtest result is skill or search.
How do signal, direction and risk turn into positions?
The SIGNAL node combines the trigger and filters into one long or short decision per bar. The RISK node then decides what that decision means in practice. Its main fields:
- Size: 1 to 100% of capital per position, committed as margin. Leverage multiplies that margin into notional exposure, so 10% size at 2x controls about 20% of capital in notional.
- Leverage: 1 to 50x.
- Max concurrent positions: 1 to 20.
- Take-profit and stop-loss: optional percentage barriers from 0.2 to 50%.
- Direction: long, short or both.
- Opposite-signal handling: reverse the position, ignore the new signal, or allow it only to reduce.
- ATR trailing stop: a stop that follows price by a multiple of the Average True Range.
- Time stop: close after a maximum number of bars.
- Exit on signal flip: close when the entry signal reverses.
Exits live in the RISK node by design. Trigr does not support custom exit graphs, so every exit is one of these well-defined controls. That keeps backtest and live behavior identical and avoids the common bug where a hand-written exit condition references data the live system does not have.
Direction deserves attention. A long-only momentum strategy and a both-sides reversing strategy can share the same trigger and behave completely differently in a sideways market. The companion guide on position sizing and risk settings goes through each control in detail, including how leverage interacts with liquidation on Hyperliquid.
An illustrative example
Here is an idea to test, not a recommendation:
- Trigger: SOL 1H RSI(14) crosses up through 30.
- Filter: BTC 4H close above its 200 EMA (cross-asset).
- Filter: SOL 1H ATR(14) as a percentage of price below a chosen ceiling.
- Signal: long when trigger and both filters are true.
- Risk: 10% size, 2x leverage, long only, 3% stop-loss, ATR trailing stop at 2.5x, time stop at 48 bars.
Nothing about this graph is known to be profitable. It is a reasonable structure for a dip-buying hypothesis, and the backtest is where you find out whether the hypothesis survives costs and a skeptical reading of the results.
How should you backtest a no-code strategy?
The builder makes strategies quick to create, which also makes it easy to create hundreds of variants and keep the one that looks best. The backtest engine and the way you read it are what protect you from that.
Trigr's engine is built around a few rules (the backtesting docs cover the cost and intrabar details):
- Point-in-time data. Every feature uses only information available at the bar's close. The point-in-time backtesting guide explains why this matters so much.
- Next-bar-open fills. A signal on bar N fills at the open of bar N+1, not at the close that produced the signal.
- Fees always on. Trading fees and the builder fee are applied to every run.
- Slippage and funding are opt-in. They are off by default, so a first result is labeled gross. You can add a flat slippage in basis points per fill and funding from a historical series or a flat per-8h rate. Slippage is a flat assumption, not an order-book or market-impact model.
- Conservative intrabar handling. If a take-profit and stop-loss are both touched inside one bar, the stop wins.
- No invented data. Missing inputs are never filled with made-up values.
A standard backtest shows net return, raw Sharpe, drawdown and a heuristic robustness score, and marks the Deflated Sharpe Ratio and the Probability of Backtest Overfitting as not yet computed. An ML optimization run fills them in: DSR, a method from Bailey and López de Prado, discounts a Sharpe ratio for the number of trials and the sample's statistical properties. You also get an equity curve, monthly returns, the full trade log and an audit trail flagging which inputs were real and which simulated.
One point is easy to misread: Studio holds out the last 30% of a standard backtest as out-of-sample by default. It is unseen only until you act on it. Standard backtests use the full available history, and once you have looked at that out-of-sample stretch while tweaking a strategy, it is part of your research data. A final out-of-sample check comes from forward testing on a paper agent after you freeze the strategy, or from the anchored, purged walk-forward evaluation in ML optimization.
A standard backtest costs 50 credits (1,000 credits equal $1), and re-running an unchanged strategy is a free cache hit.
Can an AI assistant build the graph for you?
Yes, in two ways. Studio includes an AI copilot that drafts or edits a strategy from plain English, for example "long ETH on a 4H SuperTrend flip, only when BTC funding is below its 30-day average, 5% size, 3x leverage".
You can also bring your own assistant. Trigr runs a remote MCP server that ChatGPT, Claude, Claude Code, Codex and other MCP clients can connect to through guided setup. The assistant can search the capability catalog, draft and save strategies, run credit-metered backtests with optional slippage and funding, and read stored runs and trade logs. It cannot start execution, place a live trade or withdraw funds, and consequential actions go through a single-use approval bound to the exact tool arguments.
Either way, the output is the same node graph, subject to the same validation. An AI can write a strategy faster than you can; it cannot make an overfit one less overfit.
How do you go from backtest to a live agent?
Once a strategy survives a net-of-costs backtest and a skeptical read, the next step is a paper agent. Paper agents fill at the current Hyperliquid price and deduct a base-tier trading fee and the builder fee, but model no slippage and no funding, so treat them as a check on signal behavior and timing rather than on exact costs.
An agent runs on Paper, Hyperliquid, or Propr, with up to 52 strategy slots, one per market. Allocation selects an equity share; effective strategy size determines target margin inside that share. Allocation does not reserve cash. Every venue uses current account equity, and totals can exceed 100%. An entry whose margin does not fit is skipped; a fee-only shortage can reduce its size. Setup shows a combined backtest and a risk panel, and a frontier marker separates the backtest from the forward record so you never confuse the two.
For live trading on Hyperliquid, the connection is non-custodial. Trigr generates a trade-only agent key (a Hyperliquid API wallet), you approve it with your master wallet, and Trigr stores it encrypted so the agent can run while your browser is closed. The key can place and cancel orders but cannot withdraw or send funds to another address. You keep your master wallet key and custody of your funds. The deployment guide walks through the steps, and Hyperliquid's documentation on margining is worth reading before you choose leverage.
Backtests are not guarantees; perps are leveraged and can lose more than expected.
What are common mistakes with no-code strategies?
| Mistake | Why it hurts | What to do instead |
|---|---|---|
| Stacking many filters until the curve looks smooth | Each filter is a fitted choice; the result is often noise | Keep filters that have a reason to exist; check DSR in an ML run |
| Judging on the gross result | Funding and slippage can erase thin edges | Turn on slippage and funding before deciding |
| Tuning until the out-of-sample split looks good | You have already looked at it | Freeze the strategy, then forward-test on paper |
| Using high leverage to rescue a weak edge | Leverage scales losses and liquidation risk too | Fix the edge first; size modestly |
| Ignoring funding on long holding periods | Perp funding accrues while you hold | Read Hyperliquid's funding mechanics and include funding in the test |
| Adding a second trigger idea to the same graph | Mixes hypotheses and muddles attribution | Build separate strategies and combine them in one agent |
A useful habit is to write down the hypothesis before building: what should happen, in which regime, and why. If the best version of the strategy needs conditions you cannot explain, it is probably fitted to the past.
Next steps
Open the strategy builder docs for the full field reference, then build one simple graph and backtest it with slippage and funding turned on before changing anything else.