# Behind Stablecoin Yields: Some Strategies You Can Run Yourself

> Hold HYPE spot and short an equal perp on Hyperliquid to collect funding: our live account returned 12%+ annualized. We break down where the yield comes from, Portfolio Margin, costs and risks, and open-source our live Rust code and AI skill.

---

> Author: ![](/images/brand-assets/logo.svg) Barker Research

### Build a 12%+ APY yield strategy yourself on Hyperliquid [(strategy code open-sourced)](https://github.com/royrzguo/hype-funding-carry)

Most people chasing high yields on stablecoins take one of two paths: chase incentives from farm to farm, swapping and bridging between stablecoins and moving on whenever yields drop; or jump into high-yield "Ponzi" products and bet they can get out before the collapse. There is a third path: build a stablecoin yield strategy yourself.

## 1. Unpacking yield products: where does the yield come from?

Over the past months we have tracked and analyzed a large number of yield-bearing stablecoins and on-chain yield products. Broadly, the underlying yield falls into three categories:

- **Project incentives**: the most direct source, but every campaign differs in duration and in the token used for rewards
- **Transparent on-chain strategies**: on-chain lending, funding-rate arbitrage, trackable DEX trading strategies, and so on
- **Black-box off-chain strategies**: off-chain private credit, repackaged off-chain financial products, CEX trading strategies (once funds enter a CEX they enter a black box you cannot see through), and various other opaque sources of return that rely on audits and the project's own disclosures.

Yield product issuers bundle strategy, execution and day-to-day management together: you deposit, the manager allocates the funds, runs the strategy and manages risk, and you receive a yield. The bundle is convenient, but your money passes through several layers, and a single layer may spread funds across multiple platforms. **Every layer requires your trust, and a failure at any one of them can cost you all of your principal**:

![Behind the yield, who do you have to trust?](/images/research/hype-funding-carry/vault-vs-hyperliquid-self-managed.png)

Some of the simpler strategies **you can actually run from your own wallet**. No handing your money to anyone, and no management or performance fees.

## 2. The third path: build your own stablecoin yield strategy

### We ran it live ourselves

We chose a **funding-rate carry strategy** that is easy to set up: buy HYPE spot and short an equal amount of the perp, so the two legs hedge each other's price moves while you collect funding.

Running it with our own money on Hyperliquid from July 29 to October 13, 2026, about 76 days, the strategy's NAV grew **2.45%, an annualized historical APY of 12.33%**, with a maximum daily NAV drawdown of **0.17%**, annualized volatility of 1.55% and a Sharpe ratio of about **4.2** (daily NAV, 5% risk-free rate). Positions and trading activity are publicly verifiable on-chain: [our wallet account](https://hl.eco/0x54251e24e7e5dfc66c02ea02f41bcb2419380bad). On hl.eco, the Positions tab shows only the perp short, whose unrealized loss is offset by gains on spot; the HYPE spot appears under Supply in the Lending tab.

![Daily NAV of the strategy account](/images/research/hype-funding-carry/fig-account-nav.png)

### The top borrower on Hyperliquid runs the same strategy

[This wallet](https://hl.eco/0xf02d16a272a842f8bac1d9a9e773aba1933454c6), the largest USDC borrower on Hyperliquid, runs the same funding-rate carry strategy. It holds about **693,000 HYPE in spot** and is short about **648,000 HYPE in perps**, with a short notional of about **$52.71M** and about **$35.47M** in USDC borrowed. Most of the spot is hedged by the short, at roughly 1.75x leverage, for an estimated yield of around 14%.

This gives an intuitive sense of the strategy's capacity: tens of millions of dollars do not move this strategy much.

### Setting it up is simple, can be done by hand, and needs almost no management once running

Done manually, the process is simple: enable Portfolio Margin, buy spot while opening an equal short, and then mostly leave it. After the initial entry we added capital twice, and apart from testing scripts we barely touched the positions. After our first entry, HYPE rose from about $55 to as high as about $98, and we trimmed the position once along the way. The two later top-ups were at $77–80; since then HYPE has risen another 27% or so, and on September 17 spot briefly deviated from the perp by more than 500 bps. The account stayed hedged throughout. **Under normal conditions rebalancing is rare and payments settle automatically; abnormal markets need a timely response.** When entering, the main concerns are price impact and the fill spread between the two legs; while holding, the main concern is borrowing cost.

If you want lower entry costs, automatically paired execution, tighter control of the fill spread between legs, lower trading latency, and a full set of automated risk controls and rebalancing, we have also open-sourced the Rust code we run live: [open-source GitHub repository](https://github.com/royrzguo/hype-funding-carry). It comes with a companion AI skill, making it easier to deploy through an AI agent and adjust the configuration to your needs.

We want to stress that both the strategy and the scripts carry risk. Only consider running them once you fully understand how the strategy and the code work. AI agents make execution easier and faster, but AI will not share your losses. We recommend discussing the strategy and script logic thoroughly with your AI until you understand them before running anything live.

## 3. Why run a funding-rate carry strategy "now" and on "Hyperliquid"?

Funding-rate carry has existed for as long as perpetual futures have. Many versions of it have already been scaled up and productized, and a large number of stablecoins and vaults are built on it. We chose to run it on Hyperliquid now because, compared with similar setups on a CEX, Hyperliquid makes the strategy easier to set up and manage, with higher returns and lower risk.

### Portfolio Margin lets spot and short share collateral

The key change that makes this strategy easier to manage is **Portfolio Margin**. Hyperliquid [announced on June 25, 2026 that it was entering Beta with expanded limits](https://t.me/hyperliquid_announcements/563), and an account value above $10k currently meets the eligibility threshold. In this mode, HYPE or BTC spot can serve as collateral, and the system automatically borrows USDC to margin perp positions. When the price rises, the appreciation of spot covers the additional margin the short needs, removing the hassle of frequently moving funds by hand.

CEXs offer similar features: Bybit calls it the **Unified Trading Account (UTA)** and OKX calls it the **Unified Account**, with modes including multi-currency margin and portfolio margin. They likewise improve capital efficiency. What draws us to Hyperliquid is that spot collateral, automatic borrowing and the perp hedge all sit in a single on-chain account controlled by your wallet.

### A deep, reliable order book and liquidity

What a funding-rate carry strategy fears most is a sudden wick on one leg: chasing 10%+ annualized funding can cost you your principal. Illiquid small-cap tokens may carry very high funding rates, but once you build a large short, your position can become someone else's target, and volatility in thin markets is a serious risk. BTC and HYPE both have active markets on Hyperliquid: in an October 13 snapshot, 24-hour perp volume was about $3.18B and $530M respectively, and the order book is directly visible. What we value is liquidity in major markets and the depth you can actually fill against when entering and exiting. Papertrade's use of the HyperCore best bid/offer midpoint (BBO mid) to price synthetic trades also reflects trust in this order book.

### Self-custody, independent of any centralized platform

What is different about running the same strategy on a centralized exchange? Another important reason to choose Hyperliquid is **self-custody**: we control the on-chain trading account with our own wallet, and the trade history is publicly verifiable. On a CEX, you also have to hand your assets to the exchange for safekeeping, taking on an extra layer of custodial platform risk, and withdrawals depend on its withdrawal service.

The recent Bitget incident is a concrete example. On September 24, 2026, assets in some hot and warm wallets were transferred without authorization, and the platform then suspended withdrawals. According to official disclosures, deposits and trading continued during this period, and withdrawals resumed in batches, completing on October 2. For holders, a balance that is still there and trading that still works are one thing; being able to withdraw assets at any time is another, and both need watching. [Bitget incident review](https://www.bitget.com/asia/academy/bitget-security-incident-what-happened-timeline-impact-response)

Another test came during last year's "1011" crash. Binance later confirmed that some modules malfunctioned, delaying internal transfers and yield product redemptions, which kept some users from topping up margin in time. Liminal, which runs neutral strategies on Hyperliquid, noted in its post-mortem that Hyperliquid stayed up throughout with no bad debt; however, the system triggered large-scale auto-deleveraging (ADL), and some shorts sitting on unrealized profit were force-closed. Liminal promptly closed the remaining spot and moved to USDC. Of the users running its strategy who were affected by ADL, 99.94% ended up in profit, with an average return of +3.43% on that event; the funding-rate carry strategy benefited significantly from ADL. **In extreme markets, being able to act on your account in time is itself a key criterion for choosing a platform.** [Binance announcement](https://www.binance.com/en-KZ/support/announcement/detail/d9cb0d52d7c142a5be4f49732bd8760c), [Liminal post-mortem](https://paragraph.com/@liminal-2/liminal-s-breakdown-october-10-market-event)

Self-custody keeps control of the account in your hands, while trading and withdrawals still depend on Hyperliquid's chain, trading system and bridging mechanism. What we chose is a shorter dependency path that is easier to see through.

### Higher funding yields over the same period

Could the same strategy work on centralized exchanges with HYPE, BTC or ETH? We compared actual settlement data across four platforms, Hyperliquid, Binance, OKX and Bybit, over a common window of roughly the last 90 days:

![Funding rate comparison across four platforms](/images/research/hype-funding-carry/fig-funding-compare.png)

From July 16 to October 13, for the same asset HYPE, annualized funding was about **9.9%** on Hyperliquid, about **6.5%** on Binance, about **5.1%** on OKX and about **3.5%** on Bybit. BTC and ETH on the three CEXs annualized at about **4.1%–5.8%** over the same period, before covering the costs of entering, holding and exiting. Small-cap tokens may pay higher rates, but they are illiquid, prone to wicks and may be delisted, which is a different order of risk. HYPE on Hyperliquid delivered the highest yield in this comparison, with about 9.1% of hours at a negative rate. If you are more comfortable with BTC, you can build the same strategy with BTC; the mechanics are the same.

In addition, Portfolio Margin lets you use modest leverage to improve capital efficiency. HYPE's collateral ratio is 65%. Keeping the one-leg position at 1.3–1.5x principal means borrowing is about 33%–43% of spot value at entry, leaving plenty of borrowing headroom. With spot and short kept equal, management comes down to borrowing headroom and divergence between the two legs.

## 4. Today's 12%+ yield: can it last?

HYPE funding averaged 22.1% annualized over all of 2025 and about 9.2% so far in 2026. Our live historical APY is **12.33%**, with an average one-leg position of about 1.46x principal; raising the position multiple can further amplify funding income. So can this yield last?

Account return breakdown as of October 13:

| Item | Amount (USDC) | Notes |
| --- | ---: | --- |
| Funding received | +880.19 | Positive-rate portion of 1,822 settlements |
| Funding paid | -60.62 | Shorts pay when the rate is negative |
| Borrow interest | -184.08 | Estimated from borrow balance and rate accrual; average annual rate about 5.1% |
| Trading fees | -50.05 | 317 fills |
| Two-leg spread P&L | +4.96 | Entry spread and end-of-period valuation |
| Other | -1.00 | Account activation fee |
| **Net return (breakdown estimate)** | **+589.39** | Account equity about 42.9k USDC |

Principal was deposited in three tranches, and the return is calculated on NAV, so net return divided by ending equity comes out below 2.45%.

Almost all of the return comes from funding, and the largest cost is borrow interest. Fees are incurred mainly on entry and exit: for short holding periods they matter a lot, while for a position held long-term their share keeps shrinking.

### Position size and returns

Rather than hedging one unit of capital with one unit of position, we borrowed some USDC and sized the one-leg position at about 1.5x principal. The table below uses the average funding rate during the live period (10.1%) and a 5% borrow rate, before fees:

| One-leg position / principal | Borrow / principal | Annualized return (est.) |
| --- | ---: | ---: |
| 0.8x (no borrowing) | 0 | About 8.1% |
| 1.0x | 0.1x | About 9.6% |
| 1.4x | 0.54x | About 11.5% |
| 1.75x | 0.93x | About 13.1% |

The larger the position, the more funding you receive and the more you borrow, and the earlier you need to trim when price rises sharply in one direction. Also, the portion of HYPE perp notional above $20M can use at most 5x leverage, so large accounts tie up more margin and capital efficiency is actually slightly lower.

### Funding history and return expectations

![HYPE funding rate over the past year](/images/research/hype-funding-carry/fig-hype-funding-year.png)

In the nearly two years since HYPE launched, annual average funding has stayed above 9%: 22.1% for all of 2025, 9.2% so far in 2026, and 9.7% over the past year. Over the past year, the rolling 7-day annualized rate sat between 5% and 15% about 80% of the time.

**12.33% is the historical result of this live run**, and future returns will move with funding and borrowing costs:

- **Funding will decline**: the more people run this strategy, the more shorts there are, and the less funding longs have to pay. The chart shows funding has already fallen from above 20% in 2025 to around 10% now.
- **Borrowing will get more expensive**: the more people borrow USDC, the higher pool utilization goes, and the higher the rate.

Assuming funding falls another 10%–20% from its past-year level (9.7%), a rough estimate at 1.3–1.5x position and a 5% borrow rate gives about 8%–10% annualized after borrow interest. This is what we consider a more conservative expected range.

## 5. The strategy carries real risks; understand them fully before deciding if it suits you

**We do not recommend anyone start blindly without understanding how the strategy works and the risks it faces.** The strategy does not bet on direction, but it is not risk-free, and NAV will fluctuate. Each of the following can reduce your returns or even cause losses:

- **Funding turns negative**: about 9.5% of hours during the live run were negative, with a longest streak of 8 hours; the worst historical week (October 9 to 16, 2025) annualized at about -11.9% on a 7-day basis.
- **Borrow rates rise**: above 80% utilization the rate climbs quickly (5% + 4.75 × (utilization - 80%)). If borrowing costs exceed funding income for a sustained period, the strategy can lose money.
- **Brief divergence between spot and perp**: on September 17, HYPE spot fell from $80.7 to $75.1 in about two minutes while the perp barely moved. Estimated from order book mid prices, the divergence exceeded 500 bps and recovered after about 12 minutes. Our live strategy went through exactly this episode, with no fills and no liquidation. This divergence recovered while we held, and the paper volatility converged with it; if an extreme divergence persists, it can still cause real losses. The highest PMR observed over the entire live period was about 0.62, still well away from the 0.95 liquidation line.

![Spot vs. perp price divergence on September 17](/images/research/hype-funding-carry/fig-basis-0917.png)

- **Sharp rallies require trimming**: when price rises, the short's loss and the resulting increase in perp margin are covered by USDC the system borrows automatically, and the account risk metric PMR rises accordingly; above a PMR of 0.95 the account is liquidated. Assuming 10x perp margin, PMR reaches 0.95 when HYPE rises by the amounts in the table below. We recommend trimming proactively once PMR exceeds 0.8. In a normal decline, the USDC earned on the short automatically offsets the borrowing; leg divergence and ADL during a sharp drop still need prompt handling.

| One-leg position / principal | 1.0x | 1.3x | 1.5x | 1.75x |
| --- | ---: | ---: | ---: | ---: |
| PMR at 0.8 (trim recommended) | About +175% | About +112% | About +83% | About +57% |
| PMR at 0.95 (liquidation line) | About +360% | About +250% | About +210% | About +160% |

- **Auto-deleveraging (ADL)**: in extreme markets, Hyperliquid may force-close profitable shorts to cover liquidation losses. Once the short is closed, the spot is no longer hedged, and you need to close the spot or reopen the short quickly. This happened on October 11, 2025. ADL can in some cases be a profit opportunity, but the actual risk still needs to be assessed.
- **Platform risk**: all assets sit on Hyperliquid, so technical failures, rule changes or problems with USDC bridging affect you directly. Portfolio Margin is still in testing, and when the borrowing cap is fully used the account falls back to standard margin mode.
- **Your own operational risk**: losing your private key, miscalculating positions, misaligned legs, or failing to trim when you should are all on you.

## Appendix: How the strategy works and step-by-step setup

### How a funding-rate carry strategy works

Suppose you hold 100 HYPE in spot and short 100 HYPE in perps: if the price rises, spot gains and the short loses; if it falls, the reverse. With equal quantities on both legs, most price moves cancel out. When the perp funding rate is positive, longs pay shorts, and Hyperliquid settles every hour; after borrow interest and trading costs, what remains is the strategy's return. When funding is negative, shorts pay.

### Getting started

**Eligibility**: enabling Portfolio Margin on Hyperliquid requires an account value above $10k, or weighted cumulative trading volume above $5M on the main account; the current Beta also requires account value below $25M; and the account must have no open positions or orders before enabling.

### Manual checklist

Using 20k USDC of principal and a 1.5x one-leg position as an example, assuming a HYPE price of $85 and perp margin estimated at 10x leverage; actual figures depend on prices and the interface at the time:

1. **Enable Portfolio Margin**: deposit USDC, go to the Portfolio page and select Portfolio Margin (or, on the perp trading page, click the account mode button at the top right of the order panel and switch to Portfolio Margin).
2. **Size the position**: 30,000 USDC per leg, about 353 HYPE; the system automatically borrows about 13,000 USDC, roughly 43% of spot value, below the 65% cap. Using the past year's funding rate of 9.7% and a 5% borrow rate, this works out to about 11.3% annualized, before fees.
3. **Enter in tranches**: split into 5 tranches. For each, place a limit buy of about 70 HYPE on the spot page (the last tranche covers the remainder), and as soon as it fills, short the same quantity on the perp page in cross margin mode.
4. **Verify**: spot and short quantities are roughly equal, PMR is below 0.65, and the borrow amount and rate are visible. You can also enter your address on hl.eco and cross-check the Positions (short) and Lending (spot and borrowing) tabs.
5. **Hold**: funding settles automatically every hour and borrow interest accrues automatically. Under normal conditions rebalancing is rare; in abnormal markets, check positions and borrowing headroom promptly.
6. **Exit**: also in tranches, sell spot while closing the same quantity of the short; the USDC from the sale automatically repays the loan. Once everything is closed, confirm the borrow balance is zero and no positions remain, then withdraw to your wallet.

Fee reference (official base rates, October 2026): spot maker 0.040%, taker 0.070%; perp maker 0.015%, taker 0.045%. Taking on both legs, a full open-and-close round costs about 0.23% of position size.

### What to watch while holding

Below are the monitoring thresholds for this article's example:

| What to watch | Normal range | Signal to act |
| --- | --- | --- |
| Account risk metric PMR | Below 0.65 | Trim above 0.8 |
| Borrow Health Factor (Health Factor in the Lending tab) | Above 1.3 | Below 1, automatic borrowing stops and you must top up USDC yourself to cover short losses |
| Quantities on both legs | Spot and short roughly equal | Rebalance when the gap is noticeable |
| Funding income vs. borrowing cost | Income covers cost over the same period | Consider exiting if income fails to cover cost over the long run |

### Open-source code: let AI deploy it for you

Our live account runs this exact code, fully open-sourced (MIT license): [github.com/royrzguo/hype-funding-carry](https://github.com/royrzguo/hype-funding-carry)

It handles paired entry, order execution, control of the spread between the two legs, and account risk monitoring. You do not need to fully understand the code itself: open the repository in an AI tool such as Claude Code or Codex to configure it. We recommend talking it through thoroughly, understanding all of the mechanics, and testing extensively before any live deployment. Once configured, you can simply ask the AI to set the parameters you need, such as "change the position to 1.2x" or "alert me if funding is negative for 24 hours straight", and it will update the configuration and deploy it for you.

But we will stress it once more: **you do not have to understand the code, but you must understand the strategy's logic and what the code is doing on your behalf.** AI can configure the strategy and place orders for you, but it cannot bear the risk for you.

### AI skill: get your AI agent up to speed faster

The repository also includes an AI skill ([skills/hype-funding-carry](https://github.com/royrzguo/hype-funding-carry/tree/main/skills/hype-funding-carry)) that teaches your AI agent how to compile the code, configure the account and API wallet, run read-only checks (account status, positions, borrowing and PMR), and start, stop and exit the strategy. It has built-in safety rules: unless you explicitly ask in the conversation, the AI will not run any order commands, and it will not read or copy your private key.

Installation:

- Claude Code: `mkdir -p ~/.claude/skills && cp -r skills/hype-funding-carry ~/.claude/skills/`
- Codex: have `$skill-installer` install `https://github.com/royrzguo/hype-funding-carry/tree/main/skills/hype-funding-carry`

Before using it, ask your AI plenty of questions until you truly understand the strategy and what the code does for you.

### Safety reminders

- When letting a program trade, use only a Hyperliquid **API wallet**: it can place orders but cannot withdraw. Never paste your main wallet's private key into a conversation or into code.
- Run the read-only checks first (no keys loaded, no orders placed). Once you understand the output, try a small $10–50 trade, and only scale up after confirming everything is correct.
- Run it in a dedicated account: the program manages all HYPE positions and open orders in the account.
- The open-source code is provided as is; configuration and the safety of your funds are your own responsibility.

We will cover the code's implementation details in our next article.

## Data Sources

- Live wallet: [0x5425…0bad](https://hl.eco/0x54251e24e7e5dfc66c02ea02f41bcb2419380bad). Live and funding data in this article are current as of October 13, 2026.
- Largest borrower on Hyperliquid: [0xf02d…54c6](https://hl.eco/0xf02d16a272a842f8bac1d9a9e773aba1933454c6); [ASXN lending leaderboard](https://hyperscreener.asxn.xyz/lending).
- The 1011 event: [Binance announcement](https://www.binance.com/en-KZ/support/announcement/detail/d9cb0d52d7c142a5be4f49732bd8760c), [Liminal post-mortem](https://paragraph.com/@liminal-2/liminal-s-breakdown-october-10-market-event).
- Account modes: [Hyperliquid Portfolio Margin](https://hyperliquid.gitbook.io/hyperliquid-docs/trading/portfolio-margin), [Bybit UTA](https://www.bybit.com/en/help-center/article/FAQ-Unified-Trading-Account), [OKX Unified Account](https://www.okx.com/zh-hans/help/why-use-a-unified-account).
- Operations and fees: [Enablement FAQ](https://hyperliquid.gitbook.io/hyperliquid-docs/support/faq/portfolio-margin), [Funding rules](https://hyperliquid.gitbook.io/hyperliquid-docs/trading/funding), [Trading fees](https://hyperliquid.gitbook.io/hyperliquid-docs/trading/fees).

*All data in this report comes from public disclosures and public on-chain data. It is not investment advice.*