TL;DR: Position sizing decides how much a single wrong trade can cost you, which matters more for survival than any entry signal. For crypto perps, the settings that do the most work are size as a share of capital, leverage, a stop defined in advance, a volatility-aware trailing stop, a time stop and a cap on concurrent positions. In Trigr all of these live in one RISK node, so the backtest and a live agent apply them identically.
Why does position sizing matter more than entries?
Two traders can take identical signals and end the year in completely different places because of sizing. One risks a small slice of equity per trade and survives a losing streak. The other uses high leverage and a wide stop, and a handful of consecutive losses ends the account.
Losing streaks are normal. Even a strategy with a genuine edge will have runs of losses much longer than intuition suggests. Sizing should be chosen so the worst plausible streak is survivable, psychologically and financially.
Perps add two complications: leverage makes it easy to take more exposure than you intended, and liquidation can close a position before your own stop does.
Which risk settings can you control in Trigr?
In Trigr's no-code Studio, every strategy has exactly one RISK node. It turns a signal into a position and decides when that position ends. The no-code strategy builder guide shows where it sits in the graph.
| Setting | Range or options | What it controls |
|---|---|---|
| Size | 1–100% of capital | The share of capital a position commits as margin |
| Leverage | 1–50x | Multiplies that margin into notional exposure |
| Max concurrent positions | 1–20 | How many positions the strategy can hold at once |
| Take-profit / stop-loss | 0.2–50% each, optional | Fixed percentage exits |
| Direction | Long, short or both | Which sides the strategy may take |
| Opposite-signal handling | Reverse, ignore, reduce-only | What happens when a signal fires against an open position |
| ATR trailing stop | ATR multiple and period | A volatility-scaled trailing exit |
| Time stop | Max bars | Close after a fixed number of bars |
| Exit on signal flip | On or off | Close when the entry signal reverses |
A position closes when any active exit trips first. There are no custom exit graphs: exits are these defined controls, which keeps the logic identical between backtest and live execution. The strategy builder docs have the full field reference.
How do size and leverage interact?
Think in terms of notional exposure and loss at the stop, not leverage alone.
In Trigr, size is the share of capital a position commits as margin, and leverage multiplies that margin into notional exposure. Notional is therefore roughly capital times size times leverage. A $1,000 account using 20% size and 3x leverage commits $200 of margin and controls about $600 of notional. If the stop-loss sits 4% away, a stopped-out trade costs about $24 before fees and slippage, or 2.4% of the account.
Now run the same account at 50% size and 10x leverage: notional is about $5,000, and the same 4% stop costs about $200, or 20% of equity on one trade. The signal did not change; the risk grew more than eightfold.
A useful way to set these values:
- Decide the maximum loss per trade you accept as a share of equity.
- Decide where the stop must be for the strategy's logic to be wrong, as a percentage from entry.
- Choose size and leverage so notional times stop distance equals that maximum loss.
This is a simplified fixed-fractional approach. It ignores gaps through stops and fees, so leave a margin of safety.
Leverage and liquidation
Higher leverage means less margin behind the position, which moves the liquidation price closer. Hyperliquid's margining documentation explains that maintenance margin is half the initial margin at an asset's maximum leverage, and that cross margin shares collateral across positions while isolated margin does not. Its liquidation page describes how positions are closed when equity falls below maintenance margin.
The practical rule: your stop should always sit well inside your liquidation price. If a normal-sized move can reach liquidation before your stop, leverage is too high for that stop distance. Trigr's backtests approximate this for leveraged crypto positions: a position is force-closed once the adverse move eats through its margin down to an estimated maintenance level, even if a farther stop has not been touched yet.
When should you use fixed stops, ATR trails or time stops?
Each exit type suits a different kind of strategy, and they combine well.
Fixed take-profit and stop-loss
Percentage barriers are simple and predictable. They suit mean-reversion ideas where the expected move has a rough size. Their weakness is that a fixed percentage means different things in different volatility regimes: 2% is a lot for BTC in a quiet week and very little for a small-cap altcoin in a volatile one.
In Trigr backtests, if a take-profit and stop-loss are both touched within the same bar, the stop is assumed to have hit first. That is the conservative choice, and it keeps tight-barrier strategies from looking better than they are.
ATR trailing stops
The Average True Range, introduced by J. Welles Wilder in 1978, measures the typical size of a bar's range including gaps. A trailing stop set at a multiple of ATR, for example 2.5x a 14-period ATR, follows price upward on a long and never moves back down.
ATR trails suit trend-following strategies because they give winners room in volatile phases and tighten automatically when the market calms. Smaller multiples exit sooner and more often; larger multiples tolerate deeper pullbacks. Test a few values, not dozens.
Time stops
A time stop closes the position after a maximum number of bars. It enforces the idea behind many signals: if the move has not happened within a window, the reason for the trade has probably expired.
Time stops matter more on perps because holding has a cost. Funding accrues while a position is open, and a trade that drifts sideways for weeks can lose steadily to funding. Hyperliquid's funding is settled hourly; in Trigr backtests, funding is opt-in and accrues at roughly 8-hour settlements from a historical series or a flat rate you set.
Signal flips and opposite signals
Exit on flip closes a position when the entry signal reverses. Opposite-signal handling decides what happens next: reverse into the new direction, ignore the new signal, or allow it only to reduce the existing position. A long-only strategy with reduce-only handling behaves very differently from a both-sides strategy that reverses on every flip, even with the same trigger.
How many positions should a strategy hold at once?
Max concurrent positions caps stacking. With a cap of one, each new signal is ignored until the current trade exits. Higher caps let a strategy scale into repeated signals, which increases exposure exactly when the same idea is firing repeatedly, often in a crowded move.
Combined risk is the number to watch. Three concurrent positions at 20% size and 3x leverage is roughly 180% of capital in notional exposure on one asset.
At the Agent level, Trigr adds allocation to sizing. Each slot selects an equity share; effective strategy size and dynamic multiplier determine margin inside that share. Allocation does not reserve cash. Paper, Hyperliquid, and Propr use current account equity. At $1,000 equity, 50% allocation, 20% effective size, and 3x leverage, target margin is $100 and notional is $300 before adjustments. Allocations can total more than 100%; shared free margin limits concurrent entries. An entry whose margin does not fit is skipped; a fee-only shortage can reduce its size. See multi-strategy agents for combining slots.
Small accounts face venue minimums. The builder distinguishes a rounded minimum that fits the allocation budget from one that exceeds it. Runtime can increase the first case to the minimum, subject to other checks; the second cannot enter at current equity. Missing fresh equity, constraints, or quote leaves the check unavailable. Each case allows saving and reserves no margin. Start and funding guards remain separate.
How do you check risk settings in a backtest?
Change one risk setting at a time and compare. Turn on slippage and funding first, since both are off by default and a first result is labeled gross. The guide to slippage and funding in perp backtests explains the gross-to-net gap.
Look beyond total return:
- Maximum drawdown and its duration. Could you sit through it?
- The trade log. Are exits happening for the reasons you intended, or is the time stop doing all the work?
- Deflated Sharpe Ratio. An ML optimization run reports it, and it discounts for the number of variants you tried. Testing twenty stop settings and keeping the best inflates raw Sharpe.
Studio's out-of-sample split (the last 30% by default) stops being unseen once you tune settings on it. Once settings are frozen, forward-test on a paper agent. Backtests are not guarantees; perps are leveraged and can lose more than expected.
Next steps
Take a strategy you already have, fix the entry logic, and re-test it with a smaller size and a volatility-based trail. When you are ready to run it forward, see pricing for paper-agent limits on each plan.