TL;DR: A multi-strategy agent on Trigr runs up to 52 strategies, one per market. Allocation selects an equity share; effective strategy size determines target margin inside that share. Allocation does not reserve cash, and totals can exceed 100%. Entries share available collateral. The setup shows a combined backtest and a risk panel before you go live. Diversification only helps if the strategies' returns are not highly correlated, so choose slots for different behavior, not just different tickers.
Why run more than one strategy?
Any single strategy has long stretches where its edge does not show up. A trend follower bleeds in ranges; a mean-reversion strategy suffers in strong trends; a funding-rate strategy waits for crowded positioning that may not come. Combining strategies whose bad periods fall at different times can smooth the equity curve and shorten drawdowns.
That is the standard argument for diversification, and it holds in trading systems as in asset portfolios. The benefit depends almost entirely on correlation: two strategies that win and lose on the same days do not diversify each other, however different their rules look.
A second, practical reason is capital efficiency. One slow strategy may be in the market only a fraction of the time. Several strategies sharing one account can keep capital working without raising the size of any single bet.
How does a Trigr multi-strategy agent work?
An agent is a venue (Paper, Hyperliquid or Propr), one capital figure, and up to 52 strategy slots:
- Each slot holds one strategy and one asset. Slots must trade different assets; the UI blocks duplicates and the server rejects them.
- Each slot has an allocation percentage. It selects a share of current account equity on Paper, Hyperliquid, and Propr. It does not reserve cash. Totals can exceed 100%.
- Each strategy sizes inside its allocation using its RISK node: position size (1 to 100% of the allocation, committed as margin), leverage that multiplies it into notional, max concurrent positions, stops and exits.
- Slots can be your strategies or marketplace strategies you run, including pinned versions of published strategies.
The distinct-asset rule is not arbitrary. Hyperliquid nets positions per coin per account, so a long from one strategy and a short from another on the same coin would cancel into one position, and neither strategy's exits would behave as designed.
What happens when allocations add up to more than 100%?
Entries share the account's available collateral. If target margin exceeds free margin, the entry is skipped. If margin fits but opening fees do not, runtime can trim the entry to fund fees. The strategy's row shows held entries for insufficient free margin.
Admission is first come, first served. The entry whose signal bar closed first wins. Entries on the same bar go in the order the strategies are listed in the agent. The same equity and free-margin policy applies at or below 100% total allocation.
How do you read the combined backtest?
During setup, every Agent preview requests a shared-margin simulation with equity sizing, including plan-backed Propr Agents and totals at or below 100%. The result reports per-leg contribution and skipped entries. It answers useful questions:
| What to look at | What it tells you |
|---|---|
| Combined drawdown vs. each leg's drawdown | Whether the legs offset each other or pile up |
| Per-leg contribution | Whether one slot drives nearly all the return |
| Overlap of losing months | Whether the strategies share a hidden common factor |
| Total exposure over time | How much of capital is at risk when several legs are in trades |
Two cautions. First, the combined curve inherits every weakness of its legs: if a leg's backtest was overfit, the portfolio is too. Check each leg on its own terms first; how to read a backtest covers what to look for. Second, choosing which strategies to combine after looking at their combined curve is itself a form of selection. Ten legs picked from fifty candidates because they looked good together will look better historically than they will forward.
How should you choose the strategies?
Aim for different sources of return, not just different coins. Crypto assets are often strongly correlated with Bitcoin, so twenty long-only trend strategies on twenty altcoins may behave like one leveraged Bitcoin bet. Filling all 52 markets does not fix this: in a sell-off many of the strategies lose together.
Ways to add genuine variety:
- Direction. Mix long-only, short-only and both-direction strategies.
- Signal type. Combine price-based signals with ones built on funding rates, open interest, liquidations or macro data. See funding-rate and open-interest strategies for examples.
- Timeframe. A 15m mean-reversion strategy and a daily trend strategy rarely hold positions for the same reasons.
- Market. TradFi markets such as gold or the S&P 500 can add a different driver, but on Trigr they are backtest and paper only in the current beta, so a portfolio that includes them must run on a paper agent.
If you prefer a starting point, ready-made agents are multi-strategy agents you can inspect and use as a starting configuration, from six-strategy bundles to "Every Market" agents. Each shows a forward record from a paper book, or its combined backtest until a forward record exists.
How should you set allocations?
Equal allocations are a reasonable default when you have no strong reason to prefer one leg. Some traders instead scale allocations so each leg contributes a similar amount of risk: a high-volatility altcoin strategy with 3× leverage gets a smaller slice than a 1× BTC strategy. The builder offers four allocation modes: Equal, By Sharpe, Inverse vol and Custom.
Keep three constraints in mind:
- Order minimums. Runtime can increase an entry to the rounded minimum only within its allocation budget and other checks. Builder warnings allow saving; Start and funding checks remain separate. See position size vs allocation for minimum warnings.
- Concurrency. The risk panel shows max concurrent positions per strategy and agent-wide, plus per-asset exposure, and flags over-leverage before you go live.
- Headroom. Leaving part of capital unallocated gives margin a buffer when several legs are in drawdown together. Above 100% total allocation the account can fill up once allocation × size across open strategies passes 100%; after that, later entries are skipped. The builder's committed-% line shows the total and warns when it passes 100%.
A worked example
Consider an illustrative agent with $2,000 of capital, four slots and 20% left unallocated. (On Hyperliquid, an agent's capital is the connected account's balance; paper agents start with $10,000 of virtual capital, so scale accordingly.) The numbers below are arithmetic on the settings, not a performance claim.
| Slot | Allocation | Slice | Position size | Leverage | Margin per entry | Notional per entry |
|---|---|---|---|---|---|---|
| BTC daily trend, long/short | 25% | $500 | 50% | 2× | $250 | $500 |
| ETH 4H funding fade | 20% | $400 | 50% | 2× | $200 | $400 |
| SOL 1H mean reversion | 20% | $400 | 25% | 3× | $100 | $300 |
| HYPE 15m breakout | 15% | $300 | 25% | 3× | $75 | $225 |
If every slot were in a trade at once, total notional would be $1,425 against $2,000 of capital, under 1× on the account as a whole even though individual legs use 2× to 3×. Every entry is also far above the $11 notional floor. At $200 of capital the smallest slot, HYPE, would still size $22.50 of notional ($7.50 of margin at 3×); at $100 it would drop to $11.25, right at the edge.
For how size, leverage, ATR trailing stops and time stops interact inside each slice, see crypto position sizing and risk settings.
What risks does a shared account add?
On Hyperliquid, a live agent trades one account, and the agent's capital figure is a sizing and accounting limit, not a wallet boundary. All legs share the account's margin. Under cross margin, collateral is shared across positions, so heavy losses on one leg reduce the margin available to the others.
Practical safeguards:
- Keep total leverage modest across all legs, not just per leg.
- Treat the account as exclusive to your Trigr agents; external trades trigger a safety pause of every agent bound to it.
- If you qualify for Hyperliquid subaccounts (after $100,000 of volume), give the portfolio its own subaccount so it is isolated from anything else you hold.
Diversification lowers some risks but does not remove market risk. Correlations tend to rise in sharp sell-offs, exactly when you want them low, and leveraged perpetual futures can lose more than you expect.
How do you test a portfolio before going live?
Run it as a paper agent first. The combined backtest shows how the legs interacted historically; a paper agent shows how they interact now, with real signal timing and real overlap between open positions. After the frontier marker, the record is forward-only.
Watch for concentration during the forward test: one leg generating most trades, several legs entering on the same day, or aggregate exposure near the risk panel's limits. Our guide to paper trading a crypto bot covers what to compare and for how long. When the paper portfolio behaves as expected, follow the deployment guide to connect Hyperliquid with a trade-only agent key.
Next steps
Read the trading agents documentation for the slot rules and risk panel, then build a first paper portfolio from strategies you have already tested.