How to Trade AMD Stock 24/7 with Free Trading Bots on Hyperliquid

Disclosure: fomoed may earn a small commission if you open an account through the exchange links in this article.

Advanced Micro Devices (AMD) is the second name every AI portfolio has to take a position on. NVIDIA may own training, but inference is the half of the market growing fastest from here, and Lisa Su has spent the last three years quietly building the only credible alternative GPU stack in the world. The MI300X, MI325X, and MI350 series, paired with a maturing ROCm software layer and headline customer wins from Meta and Microsoft, have turned AMD from a curiosity into the obvious hedge against a single-vendor AI compute future. The catch: the stock is volatile, narrative-driven, and — like every U.S. equity — caged inside a 6.5-hour Nasdaq session that closes precisely when the next hyperscaler tweet, TSMC update, or competitor benchmark is most likely to land.

Hyperliquid now lists an AMD perpetual contract that trades around the clock. Long or short, with leverage, no broker, no PDT rule, no after-hours liquidity cliff. Combined with fomoed's free DCA, grid, and custom strategy bots, retail traders finally have an automated, 24/7, no-KYC path to the second-most-important equity in the AI buildout — and to the pair trade against its larger sibling that institutional desks have run for years.

Why AMD Trading Hours Are Broken for Retail

Like every U.S.-listed name, AMD is locked inside the same 6.5-hour Nasdaq cage we covered in detail in our companion piece on trading NVDA 24/7 on Hyperliquid. Regular hours run 9:30am–4pm Eastern. Pre-market and after-hours sessions exist on most retail brokers (4am–9:30am and 4pm–8pm ET), but liquidity is thin, spreads widen aggressively, and many brokers restrict stop-losses or block all but limit orders during those windows. Outside of those 16 hours of partial liquidity, AMD simply does not trade for retail.

AMD specifically suffers from this more than most stocks because its catalysts cluster in non-U.S. timezones. TSMC, which fabricates every meaningful AMD product, reports in Taiwan. The hyperscalers that drive the bull case for MI-series chips — Microsoft Azure, Meta, Oracle — announce capex commitments in keynotes scheduled for European or Asian audiences. Competitor news — a Google TPU benchmark, an Amazon Trainium2 launch, an NVIDIA Blackwell shipping update — lands at any hour, and AMD reacts because the market reads any NVIDIA-positive news as AMD-negative until proven otherwise.

The Hyperliquid AMD perpetual gives you a continuously priced, continuously tradable contract. The price is anchored by an oracle pulling from regular-hours equity feeds and extends through nights, weekends, and holidays via funding-rate equilibrium between longs and shorts. You will not get tick-perfect Nasdaq pricing at 2am on a Sunday, but you will get a tradable, executable price that responds to news the moment it hits the wire.

AMD on Hyperliquid: The Basic Mechanics

The AMD contract is a USDC-margined perpetual swap with no expiration. You deposit USDC into your Hyperliquid wallet, choose your leverage (1x to 10x is typical for stocks; lower than the crypto-native perps which go up to 50x), and place either a market or limit order. Funding is paid hourly and floats based on the basis between perp price and the underlying AMD reference price.

• Funding rates can run hot. AMD funding is more volatile than NVDA funding because the stock itself is more bipolar — it rallies hard on positive AI news and sells off hard on competitive concerns. Always check funding history before sizing a position you intend to hold for more than a day.
• The contract is tradable 24/7 but the underlying is not. Outside regular Nasdaq hours, the AMD perp price is set by Hyperliquid traders alone. It can wander a few percent from the most recent equity print before the next session opens, especially around earnings or major NVIDIA news. Funding is the price-discovery mechanism in extended hours.
• No physical delivery, no dividends. AMD does not currently pay a dividend, so there is no dividend-adjustment quirk. Perps cash-settle continuously via funding. You never own the underlying stock — you own a derivative that tracks it.
• No KYC at the exchange level. Hyperliquid is a self-custodial DEX. You connect a wallet (MetaMask, Phantom, Rabby, etc.), deposit USDC, and trade. There is no broker account, no W-9, no Know-Your-Customer step. Tax reporting is your responsibility.

The NVDA-AMD Pair Trade: Where the Real Alpha Lives

This is the section to read twice. AMD and NVIDIA are not just competitors — they are the two-name basket the entire institutional world uses to express views on AI compute. NVIDIA owns training with roughly 90%+ data-center GPU share. AMD is the single credible challenger in inference, with MI300/MI325X/MI350 silicon competing directly with H100/H200/Blackwell on memory-bound workloads where AMD's HBM advantage matters. Every macro hedge fund running an AI book is long one and short the other, in some ratio, at all times.

The thesis is simple. If you believe NVIDIA's training monopoly is durable and AMD's MI-series will struggle to break single-digit data-center GPU share, go long NVDA and short AMD. If you believe AI workloads are shifting from training to inference (where AMD competes more directly) and ROCm is finally closing the gap to CUDA, go long AMD and short NVDA. Either way, systemic risk — a tech selloff, a Fed surprise, a Taiwan headline — nets out, leaving only the relative-performance bet between the two names.

The math is illustrative: if NVDA outperforms AMD by 5% on a thesis-confirming day (a Blackwell production beat, a Sovereign AI deal naming NVDA exclusively), and you sized the legs dollar-neutral, your P&L is roughly 5% of one leg's notional regardless of whether the broader market was up 2% or down 2%. On an AMD-positive day — a Microsoft MI300 expansion, a ROCm 7 milestone, an MI400 leak — the neutralization works in reverse.

fomoed's custom strategy bots can run this pair systematically: one bot long AMD when its 20-day relative strength versus NVDA exceeds a threshold, one bot short NVDA in matched dollar size as the hedge. Rebalance on a fixed cadence. The trade needs roughly 2x the capital of a single-leg position because both legs need margin, but the drawdown profile is dramatically calmer — and 24/7 Hyperliquid access lets you manage the legs around overnight catalysts, something institutional desks have done for decades and retail has been priced out of until now.

AMD-Specific Catalysts to Watch

AMD's catalyst calendar is denser than most large-caps because the company straddles four different market narratives — data center AI, client CPU, gaming, and embedded (Xilinx). The catalysts that move the stock most:

Earnings. AMD reports quarterly, typically four to six weeks after NVIDIA. The market reads AMD's data-center segment guidance as the read-through on whether MI-series adoption is accelerating or stalling. Quarterly moves of 8–15% on earnings are routine, often in opposite directions to NVDA's reaction the same season — a feature pair traders exploit explicitly.

MI-series launches and roadmap updates. The MI300 launched in late 2023, MI325X in 2024, MI350 series rolling through 2025–2026, with MI400 expected to follow. Roadmap updates from Lisa Su's keynotes — Computex, AMD's own Advancing AI events, CES — move the stock measurably. Bots that read AMD's IR feed can position pre-keynote and exit on confirmation.

Hyperscaler customer wins. Public commitments from Meta (the most vocal MI300 customer to date), Microsoft (Azure deployments at scale), Oracle, and the occasional sovereign-AI win materially re-rate the stock. Every announced AMD deployment is one less data center being filled with NVIDIA hardware. Watch for headlines that quantify dollar value or chip count, not just "using AMD" without specifics.

ROCm milestones. ROCm is AMD's CUDA equivalent, and software — not silicon — has been the bottleneck for AMD's GPU adoption for a decade. Major ROCm support announcements from frontier labs (vLLM compatibility, Hugging Face native support, MLPerf submissions) move the stock because they de-risk the software story. Competitor news matters too: NVIDIA Blackwell shipment delays are AMD-positive; Google TPU and Amazon Trainium expansions are AMD-negative because they cannibalize the same merchant-GPU TAM AMD is fighting for.

Bot Strategies for AMD

AMD's price action splits into roughly three regimes, and the same three bot archetypes that work on NVDA work here — with different tunings.

Trending regime (post-catalyst). AMD trends hard after positive AI news but gives back gains faster than NVDA when the narrative cools. A custom strategy bot with a 20-day EMA filter (faster than NVDA's 50-day because regime switches happen quicker), an RSI confirmation (long >55, short <45), and a tighter trailing stop (1.2–1.5 ATR) captures the meat of these moves before mean reversion takes back the profit. Trend-following Sharpe is lower than NVDA's due to more whipsaws — but the hit rate on big moves is similar.

Range-bound regime (between catalysts). Between earnings and product events, AMD often consolidates in a 7–12% range — wider than NVDA's, because AMD's higher narrative beta produces more chop. A grid bot shines here: 10–15 orders distributed across the range, mean reversion harvesting small profits each cycle. The 24/7 perp helps grids enormously — no weekend gaps, every level fills cleanly, and elevated volatility means more grid cycles per week than on a sleepier name.

Accumulation regime (long-term thesis). If you believe in the AMD inference thesis but don't want to time entries, a DCA bot is the disciplined way to express it. Buy a fixed USDC amount every Monday or every dip below a moving average; let the average cost compound. Funding is a slow drag in uptrends but in flat-to-down regimes you may actually collect funding. Compared to the missed compounding from sitting in cash, the funding cost on a winning DCA is usually trivial.

The AI Buildout From AMD's Seat: Four Businesses, One Narrative

AMD is not one company — it is four, and the AI narrative dominates the stock disproportionately to the actual revenue mix. Data center (server CPUs and GPUs) is what the market cares about most. EPYC server CPUs continue to take share from Intel; Instinct MI-series GPUs are the AI moonshot. Together they account for the lion's share of growth and probably 70%+ of the bull-case valuation, even though they are not yet 70% of revenue.

The other three legs are the diversification story most casual traders ignore. Client (Ryzen CPUs) is cyclical, tied to PC refresh cycles and Windows upgrade cadences. Gaming (semi-custom for PlayStation and Xbox, plus discrete Radeon) is a slowly declining segment dragged by the long console cycle and Radeon's struggles in consumer GPU. Embedded (Xilinx FPGAs, acquired in 2022) is a quietly profitable, slow-growing leg serving automotive, aerospace, telecom, and industrial customers — unsexy but margin-accretive.

Why this matters for bot trading: AMD's earnings often see the stock move primarily on data-center commentary, but quarter-to-quarter softness in client or gaming can drag the headline and create selloffs mispriced relative to the AI thesis. A custom strategy can encode "buy AMD on a post-earnings selloff if data-center segment grew >15% YoY," capturing the dislocation between headline weakness and core thesis strength — impossible to execute manually during the post-earnings window without an automated bot.

AMD vs the Semis Index and Its Peers

AMD has roughly an 8–10% weight in the iShares Semiconductor ETF (SOXX) and a similar weight in SMH — well below NVDA's 30% but still meaningful. AMD's beta to SOXX has averaged 1.2–1.4 over the past two years, meaning a 5% AMD move typically translates to a 4–4.5% SOXX move. That tighter correlation creates a different pair-trade opportunity than the NVDA-AMD relative-strength trade: long AMD / short SOXX or SMH isolates AMD-specific alpha while netting out the broader semis cycle.

Among individual peers, the pair trades worth thinking about: AMD vs Intel (INTC) on the data-center CPU share-shift narrative; AMD vs Broadcom (AVGO) on the custom-silicon-versus-merchant-GPU question, since AVGO is the leading ASIC partner for hyperscalers building proprietary chips; AMD vs Micron (MU) on the HBM supply-chain trade, since every MI300 needs HBM3e and Micron is one of three suppliers (alongside SK hynix and Samsung); and AMD vs Qualcomm (QCOM) on edge-AI versus data-center-AI exposure. fomoed's custom strategy bots can run any of these as paired-leg strategies with the same dollar-neutral logic that powers the NVDA-AMD trade.

Setting Up Your AMD Bot

Here is the practical setup walkthrough. Assume you already have a wallet funded with USDC on Hyperliquid (if not, the Hyperliquid bridge from Arbitrum takes about 60 seconds).

1. Sign up at fomoed.com. Email + password, or sign in with Google. No KYC, no payment. The free tier covers DCA, grid, custom strategy, and webhook bots.
2. Connect your Hyperliquid wallet. The setup wizard walks you through a one-time builder-fee approval — a small (0.01%) routing fee that pays for the platform. You sign once with your wallet; that is the only on-chain action you need to take.
3. Pick the strategy. Trending: custom strategy with EMA + RSI filters. Range: grid bot, range-percent mode, 10–15 levels. Accumulation: DCA bot, weekly cadence, fixed USD per buy. Pair trade: two custom-strategy bots, one long AMD and one short NVDA, dollar-matched.
4. Select AMD as the pair. In the wizard's pair-search step, type "AMD". Pick "AMD/USDC:USDC" (the perpetual contract). Set leverage between 1x and 5x for stocks unless you have a strong directional view; AMD volatility is enough that 10x positions get liquidated on routine 10% earnings moves.
5. Set position size. Never risk more than 2% of your account on any single AMD trade, and never have more than 25% of your account exposed to AMD plus NVDA combined. They are correlated enough on systemic moves that they should be treated as one budget for risk-sizing — even when you are running them as a pair trade.
6. Configure stops and take-profits. Trending: 2.5% stop with a trailing stop after the first 1.5% profit (slightly wider than NVDA's to accommodate AMD's higher noise). Grids: no global stop, but a hard kill-switch if price exits the grid range by >25%. DCA: no stops.
7. Test in paper mode first. Every fomoed bot has a paper-trading mode using real Hyperliquid prices and simulated fills. Run your strategy in paper for 1–2 weeks before going live — AMD's volatility on a beta-adjusted basis is the worst possible asset to learn on with real money.

Risk Notes Specific to AMD

The NVDA dominance overhang. AMD's bull case requires market-share progress in data-center GPU. If a quarter goes by without a meaningful new hyperscaler win, or if NVIDIA's next-generation product (Blackwell, Rubin, etc.) arrives on schedule and crushes benchmarks, AMD can sell off 15%+ in a single session even without any AMD-specific bad news. This is the fundamental risk the NVDA-AMD pair trade exists to manage.

Custom-silicon competition. Google TPUs, Amazon Trainium, Microsoft Maia, and Meta MTIA are all attempts by hyperscalers to reduce dependency on merchant GPUs. AMD is more exposed to this than NVDA in percentage terms because AMD's MI-series is precisely targeting the inference workloads that custom chips compete for hardest. If custom silicon takes meaningful share faster than expected, the entire merchant-GPU thesis weakens.

PC cycle and TSMC dependency. Unlike NVIDIA, AMD has a meaningful consumer-CPU business that swings with the PC refresh cycle. A weak Windows refresh, a soft holiday quarter for laptops, or share losses to ARM-based Qualcomm or Apple Silicon equivalents in the Windows ecosystem all hit AMD's client segment directly. Separately, every meaningful AMD product is fabricated at TSMC — geopolitical tension, an earthquake, or capacity allocation favoring NVIDIA's larger orders all hit AMD disproportionately. AMD does not have an Intel-style fab fallback.

Funding compounding and tax treatment. Holding a long AMD perp for months compounds funding payments — AMD's funding tends to run similar magnitudes to NVDA's during bullish phases. Tax treatment also differs: in most jurisdictions, perpetual gains are short-term capital gains or ordinary income, not the long-term treatment available on stock held over a year. A common pattern is to use the perp for active trading and pair trades, and hold the underlying through a regulated broker for buy-and-hold.

Oracle, DEX, and self-custody risk. Hyperliquid is a young protocol; decentralized derivatives venues are less battle-tested than NYSE-listed broker stacks. Self-custody means you bear the responsibility of key management — losing your private key loses your account. The flip side is no Robinhood-style trading halts, no broker locking your account during a major catalyst, no PFOF question. Some jurisdictions restrict access; the product is not available to U.S. persons in many cases. Check your local regulations before starting.

The 24/7 Advantage: A Real AMD Example

Consider a hypothetical Wednesday: NVIDIA reports earnings at 4:20pm ET. Numbers beat, but Blackwell guidance is below the whisper. NVDA dips 4% in extended hours. AMD, weak on competitive concerns into the print, rallies 3% in sympathy because the market reads softer Blackwell as a window for MI-series penetration. By 8pm ET, retail brokerage extended-hours sessions close. Over the next 12 hours — through Asian and European open — a Korean memory supplier confirms HBM3e capacity expansion specifically for AMD orders. By Thursday's NYSE open, AMD opens 6% higher than Wednesday's close.

A traditional retail trader has no way to capture this. By the time the regular session opens, the gap is already in. Meanwhile, an automated bot on Hyperliquid running a cross-asset momentum rule — long AMD on NVDA-negative earnings if AMD's overnight relative strength exceeds a threshold — entered at 4:25pm Wednesday, scaled out half at 6pm on the initial pop, and rode the rest through Asian hours on the HBM news. By Thursday's open the bot is up 5% and partially out.

This is not guaranteed — plenty of NVDA-down nights see AMD trade lower in sympathy, and the bot has to handle both. But the optionality of being able to react is what 24/7 trading on Hyperliquid combined with fomoed's automation provides for free. For the NVDA side of this same workflow, see our companion piece on trading NVDA 24/7 on Hyperliquid.

Final Thoughts: AMD Is the Hedge, Hyperliquid Is the Venue

AMD will not replace NVIDIA as the AI compute leader in the next two to three years. That is not the trade. The trade is that AMD has the only credible second-source GPU stack in the world, that hyperscalers are paying real money to validate it, and that inference — the half of the AI market AMD is best positioned for — is the half growing fastest from here. Whether that plays out as a multi-year compounder, a series of trade-able rallies and corrections, or a violent rotation against NVDA on the next narrative shift, the right way to express it is with leverage, both directions available, and 24/7 access.

Hyperliquid closes the venue gap by offering AMD as a continuously tradable perpetual, alongside NVDA, SOXX, and the rest of the AI complex. fomoed closes the automation gap with free DCA, grid, and custom strategy bots that turn 24/7 access into actual executed trades — including the NVDA-AMD pair trade institutional desks have run for years. The toolchain finally exists. It is free, it is non-custodial, and it works.

Related Resources

• How to Trade Marvell (MRVL) Stock 24/7 on Hyperliquid
• How to Trade NVDA Stock 24/7 on Hyperliquid
• Hyperliquid Trading Bot Setup Guide
• Trade SPX 24/7 with Free Bots on Hyperliquid
• AI Crypto Trading Bots Explained
• Free Grid Trading Bot on Hyperliquid
• How to Set Up a Free DCA Bot
• Free Grid Trading Bot
• Custom Strategy Bot
• Free DCA Trading Bot
• Start Trading Free