Profiling Smart Money Wallets with Dune Analytics

By DeFi Intel Research · Reviewed 2026-07-02 · 8 min read · intermediate

Smart money wallet profiling is the systematic identification of wallets that consistently outperform the market — traders, funds, or early movers with repeatable edge. Using Dune Analytics, you can surface these wallets by combining pre-existing on-chain labels with custom behavioral filters, then analyze their strategies by examining historical transaction patterns. This guide walks through the datasets, queries, and dashboards that turn raw blockchain data into actionable intelligence.

Whether you are building a copy-trading strategy, conducting competitive research, or validating your own performance, understanding what constitutes smart money on-chain is essential. Dune provides the raw material (tables like labels.wallets, dex.trades, nft.trades) and the SQL environment to layer your own logic. By the end of this guide, you will know how to create a reproducible profiling workflow that adapts to any market condition.

What Is Smart Money Wallet Profiling?

Smart money wallet profiling refers to the practice of using on-chain data to identify wallets that belong to successful, often institutional or sophisticated participants. These wallets are characterized by:

• Consistent profitability – e.g., a wallet that sells tokens after price spikes or buys before announcements.
• Low emotional trading – long holding periods, minimal panic selling.
• Early positioning – participating in new protocol launches, initial DEX offerings, or NFT mints before the crowd.

On-chain profiling moves beyond simple whale tracking (large holders) and focuses on skill. For example, a wallet that repeatedly arbitrages Uniswap and Curve pools with high win rates is smart money, even if its total value is modest. Dune’s community-contributed labels in the labels.wallets table include categories like smart_money, arbitrageur, and early_investor, giving you a starting point. The real power comes from layering your own behavioral rules on top of these labels.

Key On-Chain Signals for Identifying Smart Money

To profile wallets with confidence, you need measurable behavioral signals. The most reliable include:

• Trade win rate – percentage of token swaps that end in profit within a defined window (e.g., 30 days). A win rate above 60% over 50+ trades suggests skill over luck.
• Holding time – median time between buy and sell. Smart money often holds for weeks to months, not minutes.
• Gas price premium – wallets that pay higher gas to front-run transactions often indicate aggressive alpha seekers. Use tx.gas_price compared to block average.
• Interaction with new protocols – wallets that are among the first 100 to use a new DeFi contract or NFT collection.
• Portfolio diversification – number of unique tokens held; smart money typically holds 5-20 assets, not thousands.

Dune tables like ethereum.transactions, dex.trades, and nft.trades contain all the raw data needed to compute these signals. Community-created dashboards on Dune often pre-compute some of these metrics; you can fork and modify them.

Essential Dune Datasets and Labels for Wallet Profiling

Dune’s ecosystem includes several datasets that directly support smart money profiling:

• Dune Labels (labels.wallets) – a community-curated table with columns: address, label_type, label_subtype, name, and source. Filter on label_type = 'smart_money' to get an initial set.
• Ethereum Transactions (ethereum.transactions) – every tx with from/to, value, gas, block time. Essential for activity frequency and gas analysis.
• DEX Trades (dex.trades) – standardised swap data across exchanges. For each trade, you get token bought, token sold, amount, price, and protocol.
• NFT Trades (nft.trades) – similar for NFT sales.
• Token Balances (erc20.balances) – current token holdings, useful for portfolio diversity.

Combine these with Dune’s built-in date_trunc and window functions to compute rolling metrics. For example, to calculate a wallet’s 30-day win rate, join dex.trades with subsequent dex.trades (or transfer events) for the same wallet to see when they sold each token.

Step-by-Step: Building a Smart Money Wallet Scanner in Dune

Prerequisites: A Dune Analytics account (free tier works for small queries; premium for heavy usage). Basic SQL knowledge.

1. Start with Dune Labels – Query: SELECT address, name FROM labels.wallets WHERE label_type = 'smart_money' LIMIT 100. This gives you a baseline set of addresses tagged by the community.
2. Fetch recent activity – For each address, pull the last 100 transactions from ethereum.transactions. Filter out failed txs (success = true).
3. Compute trade signals – Join with dex.trades to get token swaps. Calculate profit by comparing buy price to the next sell price for the same token pair (use LEAD()). Mark each trade as win/loss.
4. Aggregate metrics – Group by wallet address: count trades, win rate, average holding time, average gas price.
5. Filter for smart money – Where win rate > 0.6 AND avg_holding_days > 7 AND trade_count > 20.
6. Save as a Dune Dashboard – Visualize top wallets in a table with links to Etherscan. Add a chart of weekly new smart money wallets detected.

Example query snippet (illustrative SQL for Dune's query engine):

WITH wins AS (
  SELECT 
    trader,
    COUNT(*) FILTER (WHERE realized_pnl > 0) AS wins,
    COUNT(*) AS total
  FROM dex.swaps
  WHERE block_time > NOW() - INTERVAL '30 days'
  GROUP BY trader
)
SELECT trader, wins, total, wins::float / total AS win_rate
FROM wins
WHERE total > 20
ORDER BY win_rate DESC
LIMIT 50;

Interpreting Wallet Behavior: Case Studies

Once you have a set of candidate smart money wallets, dive into their transaction history to infer strategy. Use Dune’s labels.wallets to check for known tags like market_maker or mev_bot. Here are two illustrative examples (not real addresses):

Case 1: The Yield Farmer – Address 0x…f3a consistently deposits into new lending protocols within the first hours of launch, removes liquidity after 3-5 days, and rarely incurs losses. Their trade win rate on dex.trades is 72%. Check their erc20.balances and you see they hold governance tokens from protocols they farmed. This signals a airdrop-mining strategy.

Case 2: The Arbitrageur – Address 0x…b7e executes 20-50 flash swaps per day across Uniswap V3 and SushiSwap, often sandwiched between large trades. Their gas price is consistently 2x the block median. They rarely hold tokens for more than a block. This wallet is likely a MEV bot. Dune’s ethereum.traces can reveal the internal calls.

By cross-referencing labels and on-chain data, you can categorize smart money into archetypes: early adopters, alpha callers, arbitrageurs, and yield optimizers. Each demands a different profiling filter.

Comparison Table: On-Chain Labeling Data Sources

Dune’s own community labels are the most accessible for SQL-based profiling. They are not as comprehensive as Nansen but are transparent and auditable. For this guide, we focus on Dune Labels paired with behavioral heuristics.

Advanced Profiling: Detecting Sybil and Obfuscated Wallets

Smart money wallets often try to hide their identity using multiple addresses, mixer services (Tornado Cash), or chain-hopping. Here’s how to sharpen your profiling:

• Transaction graph analysis – In Dune, use ethereum.transactions to find wallets that share gas stations (same from address paying gas for multiple to addresses). This can reveal a cluster of wallets operated by one entity.
• Proxy contract detection – Smart money wallets may interact through a proxy. Look for contracts with delegatecall patterns. Dune’s ethereum.traces can trace internal calls.
• Cross-chain activity – Use Dune’s labels.cross_chain_bridge or bridge tables (e.g., Hop Protocol, Stargate) to see if the same Ethereum address has activity on L2s.
• Time pattern analysis – Wallets that trade only during specific hours (e.g., 9-5 UTC) are likely institutional. Use EXTRACT(HOUR FROM block_time) to detect patterns.

Be cautious: excessive filtering may exclude genuine retail smart money. Always validate with out-of-sample data.

Common Pitfalls in Smart Money Wallet Profiling

Even experienced analysts fall into these traps:

• Survivorship bias – You only see wallets that are still active. Dead or disappeared wallets (e.g., locked or lost keys) are invisible.
• Label staleness – Dune Labels can be weeks old. A wallet tagged as smart money in April might now be a bagholder. Always cross-check with recent activity.
• Confusing size with skill – A whale that bought at the top and never sold is not smart. Separate net worth from trading ability.
• Ignoring gas costs – A 70% win rate is meaningless if gas fees eat all profits. Always calculate net PnL after gas on failed txs too.
• Overfitting filters – If you set win rate > 90%, you may find only bots or tiny sample sizes. Use reasonable thresholds (e.g., 60% with >30 trades).
• Relying on a single label source – Combine Dune Labels with Etherscan tags and manual inspection of known entity lists (e.g., CoinGecko’s institutional list).

Future of On-Chain Smart Money Trackers

As DeFi and NFTs evolve, so will profiling methods. Trends to watch:

• Decentralized identifiers (DIDs) – Wallets may carry verifiable credentials (proof of skill) on-chain, replacing heuristic labels.
• MEV and order flow – Profiling private mempool transactions (via Flashbots, Eden) can reveal pre-trade alpha, though Dune currently lacks this data widely.
• Cross-Dune queries – Using Dune’s V2 engine, you can join data from multiple chains (Ethereum, Polygon, Optimism) to track smart money across ecosystems.
• AI-powered behavior clustering – Tools like 0xSift already use machine learning to label wallets; expect Dune to integrate more ML scores for behavior profiling.

Building a durable profiling system means staying updated on Dune’s dataset additions and community label curation. Join the Dune Discord and follow the #labels channel.

Step-by-step

1. 1. Create a Dune Analytics account and open the Query Editor.
2. 2. Query the Dune Labels table to get a baseline set of smart money addresses: SELECT address, name FROM labels.wallets WHERE label_type = 'smart_money'.
3. 3. For each address, pull recent transactions from ethereum.transactions, filtering for success and limiting to last 500.
4. 4. Join with dex.trades to compute per-trade profit/loss using LEAD() and token price data.
5. 5. Aggregate metrics per wallet: win rate, average holding time, trade count, average gas price.
6. 6. Apply filters (e.g., win rate > 0.6, holding days > 7, trade count > 20) to narrow down high-skill wallets.
7. 7. Create a Dune Dashboard with these results, adding a table and charts for weekly new smart money wallets.
8. 8. Validate by manually inspecting a few addresses on Etherscan for consistency with known behaviors.

Common mistakes to avoid

• Only relying on pre-existing labels without verifying recent activity.
• Confusing large balance with smart trading ability.
• Not accounting for gas costs and failed transactions when calculating win rate.
• Setting win rate thresholds too high (e.g., >90%) leading to tiny sample sizes.
• Ignoring cross-chain activity and assuming a wallet only operates on one chain.
• Failing to update profiles regularly; wallet behavior can change within days.

Frequently asked questions

Related reading

• How To Detect Fake Hardware Wallets
• How To Protect Hardware Wallet Supply Chain
• Smart Contract
• Wallet
• Basis Trade Bitcoin Futures
• Best Crypto Wallets
• Bitcoin Spot ETF Flows Analysis
• Crypto Gas Fees Explained Save Money