Technical Research Paper
Last updated
Last updated
–––––––––––––––– 0. Abstract –––––––––––––––– Crypto users who seek diversified exposure on Solana must manually acquire and rebalance a basket of tokens, paying multiple transaction fees and gas while being exposed to execution-time risks (slippage, MEV). Existing basket protocols are either custodial, EVM-centric, or rely on long-running back-end processes that violate the execution-time constraints (≤ 60 s) of modern serverless environments. We propose HedgeVault—an architectural pattern and reference implementation that (i) automatically converts user deposits into a target allocation via Jupiter liquidity routes, (ii) stores the acquired assets in a program-controlled vault, and (iii) issues a single SPL token (hereafter HVPT) that represents fractional ownership of the underlying portfolio. Our main contribution is a three-stage mint pipeline—create-mint, mint, and metadata—that decomposes the heavy “swap + mint” workload into idempotent, edge-compatible subroutines. We show that the new design reduces worst-case API execution time from > 180 s to < 35 s without sacrificing atomicity or UX, thereby enabling frictionless portfolio tokenization directly from browser wallets. –––––––––––––––– 1. Introduction –––––––––––––––– Portfolio theory (Markowitz, 1952) and its crypto derivatives (index tokens, rebalancing funds) emphasise diversification for risk reduction. Yet retail users on Solana face four systemic frictions: 1) Multistep Acquisition: Acquiring n n tokens costs n n transactions and manual slippage configuration. 2) Custodial Risk: Centralised index issuers hold assets off-chain; users inherit counter-party risk. 3) Composability: Holding five separate tokens complicates collateralisation, on-chain lending, and cross-chain bridging. Our research asks:
Can we design a non-custodial, composable portfolio token on Solana whose back-end operates entirely within serverless limits? The remainder of the paper answers affirmatively and details the design, analysis, and empirical results of HedgeVault v1.0
–––––––––––––––– 2. Related Work –––––––––––––––– Balancer Smart-Pools (2020)—EVM pools whose shares represent dynamic baskets; require monolithic smart-contracts and L2 gas fees. Index Coop / DPI (2021)—ERC-20 index tokens issued via full-node back-ends (> 120 s issuance latency). mSOL, jitoSOL (2022)—Stake derivative tokens; single-asset, not diversified. Socean Basket (draft 2023)—Solana basket, but controller program uses WebSocket confirmations and assumes persistent back-end pods. HedgeVault diverges by (i) targeting edge functions (no WebSockets), (ii) employing REST-only Jupiter V6 routes, and (iii) isolating mint metadata creation from share-mint logic to guarantee sub-60 s path completion.
–––––––––––––––– 3. Motivation & Requirements –––––––––––––––– R1. Serverless Compatibility: Each API call must finish in < 45 s. R2. Non-Custodial Trust Model: Users should rely solely on Solana program rules. R3. Composable ERC-20/SPL Representation: Resulting token must be fungible and recognised by Phantom & Jupiter. R4. Parallel Swap Efficiency: Swap SOL into n assets with aggregate slippage ≤ 1 %. R5. Idempotent & Retryable: If any step fails or times out, the user/UI must safely re-invoke without risk of double-spending. –––––––––––––––– 4. System Overview –––––––––––––––– 4.1 Actors U User with browser wallet (Phantom) E Edge API (Next.js Route Handler) V VaultManager keypair (off-chain signer) J Jupiter REST S Solana RPC (Alchemy / Light-client) 4.2 High-Level Flow Diagram
Each arrow corresponds to an HTTP request whose internal processing is guaranteed ≤ 45 s. –––––––––––––––– 5. Detailed Architecture –––––––––––––––– 5.1 Database Schema (PostgreSQL) portfolios(id PK, owner_address, allocations JSONB, deposit_amount NUMERIC, token TEXT, status ENUM, created_at TIMESTAMPTZ)
assets(portfolio_id FK, mint_address, weight NUMERIC, amount NUMERIC)
portfolio_investor_transactions(id PK, portfolio_id FK, wallet_address, amount, tx_type ENUM, tx_hash, mint, token_amount, created_at)
5.2 VaultManager Class
Maintains a single keypair vaultAccount; private key lives in Edge runtime secret.
Utility functions: swapSolForToken, getOrCreateAta, createPortfolioSwaps, createVaultDepositTransaction.
Swap algorithm quote = GET /v6/quote?inputMint=wSOL&outputMint=target&amount swapTx = POST /v6/swap (wrapAndUnwrapSol=true) vault signs swapTx (VersionedTransaction)
5.3 Three-Stage Mint Pipeline Step-A create-mint 1. createMint (SPL) with 9 decimals 2. store shareMint in portfolios.token Duration: ≈ 5–7 s Step-B mint 1. getAssociatedTokenAddress—create ATA if absent 2. createMintToInstruction (mintAmount = deposit × 1e9) 3. sendWithFresh poll < 30 s Duration: 8–15 s Step-C metadata 1. Import @metaplex-foundation/mpl-token-metadata (lazy) 2. Derive PDA, call createCreateMetadataAccountV3Instruction 3. Sign & poll < 30 s Duration: 12–25 s Total wall-clock worst-case ≈ 47 s, but fractionated across 3 HTTP calls. –––––––––––––––– 6. Implementation Nuances –––––––––––––––– Fresh Blockhash Retry: sendWithFresh loops up to 5 times, each ≤ 30 s, catching “block height exceeded” errors without WebSockets. Parallel Swaps: Promise.all() with concurrency limit 4 to keep compute budget per RPC within 100 000 units. Edge-Safe Polling: Edge workers cannot maintain WS to RPC; polling every 3 s avoids idle CPU penalties. Dynamic Import: Metaplex library (~300 KB) is imported only in /metadata, reducing cold-start. VAULT SOL Reserve: 0.0021 SOL held for every wSOL ATA rent. –––––––––––––––– 7. Evaluation –––––––––––––––– 7.1 Experimental Setup RPC: Alchemy Solana Mainnet (priority tier). Browser: Chrome v122, Phantom wallet. Edge Platform: Vercel US-East-1. 7.2 Metrics | Metric | Monolithic (/swap-mint) | HedgeVault Split | Δ | |––|––|––|––| Average API wall-time | 92 s | 27 s/14 s/18 s | −57 % 95p API wall-time | 138 s | 33 s/19 s/24 s | −65 % HTTP 504 rate | 34 % | 0 % | – User-reported failures | 12/100 ops | 1/100 ops | –91 % 7.3 Discussion The gain derives from concurrency and the fact that block confirmation for MintTo is orthogonal to metadata creation. Latency outliers (> 24 s) in /metadata stem from occasional cache misses at Metaplex program BPF load. –––––––––––––––– 8. Security & Risk Analysis –––––––––––––––– Key Custody: While vault key resides in Edge secret, compromise risk exists; roadmap includes migrating authority to a multi-sig (Squads v3). Swap Failure: Jupiter may return null quotes for illiquid pairs; allocation weight recorded as 0 to keep accounting invariant. Re-entrancy: Each route checks portfolio status; double-spending is prevented by SQL WHERE status='DEPOSIT_CONFIRMED' locks. Price Oracle: NAV pricing currently naïve; future integration with Pyth for accurate per-token USD value. –––––––––––––––– 9. Economic Model ––––––––––––––––
––––––––––––––– 10. Societal & Practical Impact ––––––––––––––– Lowers barrier for emerging-market investors entering crypto via a single click. Enables DAO treasuries to issue transparent, on-chain Treasury-Backed Tokens for governance or fundraising. Contributes to the trend of edge-native DeFi, demonstrating that heavy blockchain tasks can be decomposed to fit stateless compute limits. ––––––––––––––– 11. Future Work ––––––––––––––– Redeem Flow Burn HVPT → proportional asset withdraw; requires Solana v1.17 token-2022 features. On-chain Rebalance Bot Using Cronos or Clockwork for autonomous swaps. Cross-Chain Representation Wormhole wrapped HVPT-w bridging to EVM, Fuel, and Cosmos. zk-Proof‐of-Reserves SNARK verifying vault balances vs on-chain supply for trust minimisation. DAO Governance Module Share-mint authority transferred to a SPL governance realm. –––––––––––––– 12. Conclusion –––––––––––––– HedgeVault validates the hypothesis that a diversified, composable portfolio token can be created entirely within serverless execution budgets by decomposing the mint workflow into micro-steps. The protocol merges classical portfolio construction theory with modern edge computing constraints, providing a blueprint for future DeFi primitives that must operate under tight runtime caps. –––––––––––––– 13. References –––––––––––––– [1] Markowitz, H. (1952). Portfolio Selection. Journal of Finance. [2] Jupiter Aggregator. https://docs.jup.ag [3] Metaplex Token Metadata Program. https://docs.metaplex.com [4] PSReadLine Issue #332. GitHub, 2024. [5] Vercel Edge Runtime Limits. Vercel Docs, 2023.
