Cross-asset trading is not a basket: reading XYZ100, TSLA, NVDA, GOLD correlation with OneKey Perps

Summary: Loading XYZ100, TSLA, NVDA, and GOLD into the same perp account looks like four independent setups. In practice you are usually stacking leverage on one macro position. XYZ100 and NVDA share the same AI capex curve, TSLA tends to swing harder than the index in both directions, and GOLD is the only leg that has a real chance of bleeding green when the other three sell off together. This piece is about how to sequence funding-rate priority, pick event windows, and pool position budgets across these four names — and why running Hyperliquid perps natively inside OneKey App / Extension is the cleanest way to execute the combination.

What you are actually trading when these four names sit together

XYZ100 is a tech-heavy index. NVDA is one of its heaviest weights. TSLA is also a constituent, but its volatility and narrative premium live in a different league. GOLD looks unrelated, but it frames the macro backdrop from the other side.

Eyeball the charts and you see four different price series. Drop them into a correlation table and the real structure shows up: XYZ100 and NVDA move together most of the time, and the link tightens around NVDA earnings; TSLA's co-movement with the index is weaker but its amplitude is larger, so when the index ticks, TSLA tends to lurch; GOLD drifts with weak correlation in normal regimes but tends to flip into clear negative correlation when a real risk-off impulse hits.

So you are not trading four tickers. You are trading three layers:

• AI capex plus US tech beta — XYZ100, NVDA, TSLA all share it
• Single-name event premium — NVDA earnings, TSLA deliveries / AI Day / Musk himself
• The mirror of real rates and the dollar — GOLD

If you do not separate these layers, you fall into the most common trap: you think you diversified, then the tech tape sneezes and all four lines bleed on the same candle.

One table: drivers, common misreads, and what to check before clicking buy

The point of this table is not to tick boxes. When you catch yourself thinking "I want to go long NVDA," scan the right two columns first — are you reading the driver correctly, and have you actually looked at the variable that matters?

The event calendar matters more than the chart pattern

These four names share a property: the biggest moves cluster around specific events, not moving-average breaks.

NVDA earnings can drive most of XYZ100's weekly variance during the window — not because of where NVDA itself closes, but because the print sets the tone on whether the AI capex story still holds. So even if you have no intention of trading NVDA, if its earnings hit that week your XYZ100 position needs to be treated as an event trade, not a trend trade.

TSLA's event structure is messier: delivery announcements, earnings, AI Day / Robotaxi day, Musk's posts on X, regulatory or political headlines. The event uncertainty is usually higher than NVDA's, so at the same leverage, an outright TSLA position carries materially more overnight risk than the equivalent NVDA position.

XYZ100 has its own calendar — FOMC, CPI, PPI, NFP, Treasury auctions. These macro days push or compress all four names simultaneously. A common FOMC-day mistake is "long XYZ100 in the morning, long GOLD in the afternoon" — it looks like a hedge but you are really betting that post-FOMC real rates and equities drop together. The directions are not in conflict, but your real exposure is bigger than the position screen suggests.

GOLD's events live on the real-rate side: CPI, jobs, FOMC, central bank speeches. Geopolitical shocks are short-term thrusts; for position rhythm, the rate path tends to matter more.

In practice, I keep all four event calendars in one sheet aggregated by day. When a week stacks multiple high-impact events, I cap total leverage and total notional at the portfolio level rather than per-name.

Funding rates and depth: a perp is not free leverage

For synthetic equities and commodities on-chain, the biggest cost is rarely the trading fee. It is funding and slippage.

Funding reflects the imbalance between longs and shorts — see the Hyperliquid funding docs for the implementation. For this group of names, a few common funding patterns are worth keeping in your head:

• The week before NVDA earnings: long funding tends to print positive, meaning longs pay to carry overnight. When the tape "looks unstoppable," the market is effectively charging you rent to keep being long.
• TSLA during a major narrative beat (Robotaxi, Musk statement, price war): funding can swing in either direction depending on whether the narrative is bullish or bearish. Fading the move in real time can cost more than expected.
• XYZ100 ahead of FOMC: funding tends to stay tame, but it can reprice quickly after the event.
• GOLD in the first innings of a geopolitical flare-up: long funding tends to push up; the more it feels like a must-buy, the more expensive being long overnight gets.

For the fee structure itself, see Hyperliquid fees. Separate three things: exchange trading fee, funding rate, slippage. The OneKey wallet-layer Perps fee is 0%, but the underlying Hyperliquid trading fee, funding, and book slippage do not disappear because of which front-end you came through. They are part of market structure, not the integration path.

On depth: XYZ100 and NVDA tend to carry healthy books, so normal-size market orders carry tolerable slippage; TSLA is a notch thinner, and depth can evaporate inside event windows; GOLD has good depth during US/European overlap but the lull between Asia open and US close is where market orders punish you.

Two habits save a lot of money: break large orders into limit or iceberg, and use limit orders around events whenever you can.

Position budgeting: combine same-direction risk before you size it

This is the core of the piece, and it is the part I paid the most tuition for.

The wrong version goes like this:

"I'm bullish AI, so long NVDA at 1x; TSLA is oversold short-term, long at 1x; XYZ100 broke its moving average, long at 1x; GOLD as a haven, long at 1x. Nothing is big. I'm diversified."

Your actual exposure looks closer to this:

• Tech beta: NVDA, TSLA, and XYZ100 stacked together usually behave more like one amplified long-tech position than three separate trades.
• Haven exposure: GOLD at 1x, which only meaningfully offsets when tech sells off hard and the risk-off flow actually fires.

If NVDA prints a miss, you are not "diversified into a small drawdown." Several same-direction positions cut at once, and the GOLD leg usually absorbs only a small fraction.

The right version starts with a total risk budget, allocates it to themes, and only then maps to tickers:

1. Decide the maximum drawdown you are willing to take this week (say 3% of account).
2. Slice that budget across themes: tech beta, single-name events, haven hedge.
3. Inside a theme, multiple tickers must be sized after correlation netting, not added independently.
4. Leverage is an output, not an input — compute notional first, back out leverage.

A concrete example. Account of 10,000 USDC, willing to risk 3% — 300 USDC of book loss this week:

• Tech beta theme: 60% of the risk budget, 180 USDC. Express it through XYZ100, stop 3% from entry, notional around 6,000 USDC, roughly 0.6x. If you want NVDA or TSLA exposure, you carve it out of the XYZ100 allocation, not on top.
• Single-name event theme: 25%, 75 USDC. Only deployed in short windows around NVDA earnings or major TSLA events.
• Haven / inverse theme: 15%, 45 USDC. Long or short GOLD depending on the current real-rate direction.

The framework's job is to turn "I have a lot of ideas" into "I have a few buckets of risk." Ideas are abundant; budget is the scarce thing.

Running the combo on OneKey Perps

From framework to execution.

OneKey App and Extension natively integrate Hyperliquid perpetuals, so you do not bounce out to an external front-end, paste in contract addresses, or worry about signing on the wrong network. XYZ100, TSLA, NVDA, and GOLD all live as synthetic markets on Hyperliquid and switch one-tap inside the same interface.

For this particular group, my workflow runs roughly:

1. Build a watchlist inside OneKey: pin XYZ100, TSLA, NVDA, GOLD so you can compare current funding, 24h change, and open interest at a glance.
2. Three checks before sizing: any event collision this week, which side of funding pays whom, and how much USDC margin the target position translates to.
3. Prefer limit orders: especially for TSLA and GOLD, where market slippage tends to cost you a tick or two extra.
4. Unified margin: the four tickers share one margin account. That is a structural benefit of running through OneKey into Hyperliquid — you do not shuffle collateral between four silos — but it also means total leverage has to be controlled more deliberately.
5. Watch liquidation at the portfolio level: with cross-asset positions, the danger is not one ticker hitting its individual stop. It is tech beta selling off together while maintenance margin slips. Track portfolio maintenance ratio, not just individual stop levels.

Cost structure, once more: the OneKey wallet-layer Perps fee is 0%, but the trading fee documented in Hyperliquid fees still applies, the funding mechanism in Hyperliquid funding still settles on schedule, and book slippage does not disappear because of which front-end you used.

If you do not have the wallet installed, grab the official build from Download OneKey. App and browser extension share the same seed, which makes checking positions across devices straightforward.

A few real scenarios

Stitching the above together, here are situations you actually run into:

Scenario A — The week before NVDA earnings, sentiment leans long. Long funding is rich. XYZ100 has already run, TSLA is front-running. Going outright long with leverage here is expensive. The more reasonable expression is to use XYZ100 to play "the index keeps grinding up," leave NVDA for intraday around the print, and keep TSLA below half the usual sizing. GOLD direction follows the week's CPI — if CPI prints hot, a short GOLD position hedges the real-rate risk.

Scenario B — FOMC week, market expects a pause. Tech has beta upside, but pre-event uncertainty is high. Long XYZ100 rather than picking NVDA or TSLA captures beta without amplifying single-name event variance. Halve the GOLD position — post-FOMC real-rate reactions are sharp and direction is hard to call in advance.

Scenario C — Geopolitical shock fires unexpectedly. GOLD gaps higher at open, tech opens lower. The temptation is to chase long GOLD and short XYZ100 immediately. That is exactly when funding flips fastest, slippage thickens, and the book gets unfriendly. The steadier play is to let the first wave of emotion clear, then place limits on the pullback. If the price runs without you, skip the trade — do not buy at the most expensive moment.

Scenario D — Risk appetite returns, AI narrative repairs. NVDA leads, TSLA follows, XYZ100 grinds higher, GOLD chops or drifts. Going long NVDA outright has a reasonable hit-rate here, but the size has to come out of the tech beta bucket. Do not add NVDA long on top of an existing XYZ100 long — that is adding leverage, not rotating themes.

FAQ

Q1: XYZ100 is an index perp — isn't long-only there safer than buying NVDA or TSLA? Lower variance is not the same as safer. XYZ100 is tech beta — its daily drawdowns tend to be smaller than NVDA's, but when the AI capex narrative breaks, it falls with NVDA, and because the weights concentrate at the top, the "diversification" is less than it looks. The reliable way to reduce single-event variance is to reduce leverage and notional, not to swap names for an index.

Q2: How are TSLA and NVDA perps different from holding the actual stock? The price tracks a synthetic reference, not a share sitting in your brokerage. The upsides: 24/7, USDC collateral, easy shorting, leverage. The costs: funding, basis risk between the synthetic reference and the underlying, and a liquidation mechanism specific to perps. Treat it as its own instrument, not "an on-chain version of your brokerage account."

Q3: Is shorting / longing GOLD a reliable hedge inside a perps book? It depends. If your "hedge" is meant to offset a tech beta drawdown, GOLD provides only partial buffering most of the time — the correlation is not stable. The real hedge moments are sharp real-rate drops during risk-off impulses, and those are not frequent. Treat GOLD as a second leg under specific macro scenarios, not as guaranteed compensation if your tech longs lose money.

Q4: How do I check whether adding a position is secretly adding leverage? A simple self-check: before this new trade, what is the sum of notional across all my positions? After this new trade, what is it? If the new name correlates strongly with existing positions in normal conditions, the increase in notional is closer to a real increase in risk exposure than to diversification.

Q5: How high is "expensive" funding? There is no absolute threshold, only a relative one: annualize the current funding rate and compare it to your expected directional return. If annualized funding eats a meaningful chunk of your expected upside, you are paying rent with a sizable share of the trade's edge — and time is working against you until price catches up.

Q6: With four positions running, which metric matters most? Portfolio-level maintenance margin ratio, and aggregated tech-beta exposure. Those two numbers will tell you "you are already over-concentrated in one kind of risk" well before any single stop triggers.

Risk notice

Nothing here is investment advice. XYZ100, TSLA, NVDA, and GOLD on OneKey Perps / Hyperliquid are synthetic perpetuals and differ structurally from spot equities or physical gold. Perpetuals carry leverage, funding-rate fluctuation, and forced liquidation risk; during volatile or thin-liquidity periods, slippage and funding can be materially higher than usual. On-chain perpetual trading is restricted or prohibited in some jurisdictions — you are responsible for confirming the compliance and tax requirements in your region. Positions can produce losses beyond initial margin; size according to your own risk tolerance and prioritize position budgeting and risk controls.