Trading Strategy

What it is

A normalised dataset of DeFi vaults, sold by subscription, with a website on top of it and an open-source Python framework beside it. The unit of the product is a vault — a Morpho or Enzyme or Hyperliquid vault somebody else operates — and the work being sold is putting 5,440 of them into one schema: net and gross return, CAGR, three-month Sharpe and volatility, maximum drawdown, TVL against peak TVL, age, management and performance fees, deposit and withdrawal status, curator, denomination, and a technical-risk category. 117 protocols across 29 chains, updated daily.

This card sits in on-chain analytics rather than backtesting, and that is a judgement worth stating. The project began as a DEX-data-plus-backtester framework and is often still described that way. It is not how it is sold today: the site is a vault screener, the pricing page is headed "DeFi vault market data for professional research", and in the Free-versus-Pro comparison table the backtesting framework is one checkbox among six, beside "Support" and "Raw data files". What $199 buys is the data. The framework is listed as a secondary category here because it is real, free and separately useful — but a reader who arrives looking for a backtesting library should know they are being sold a dataset with one attached.

The framework half is two AGPL-3.0 Python packages: trading-strategy, the client that reads the datasets, and trade-executor, which backtests a strategy against them and then runs it on-chain against a vault. That second half is the busiest part of the project — trade-executor had a commit and a release on 2026-09-19, and its recent release notes are about Lagoon vault settlement inside a Safe, which tells you what it is currently being built for.

Pricing

Two plans, both public, read on 2026-09-19.

Free — $0. The vault analytics website with its filters and charts, plus a downloadable Ethereum sample dataset in the same schema as the paid one, requiring no sign-up. Separately, the DEX datasets are free behind a free API key.

Pro — $199 a month. Full vault metadata and history across all 5,440 vaults, raw data files, BTC- and ETH-denominated vaults, the backtesting framework, and support. The comparison table marks six rows as Pro-only; everything else on it, including DEX price data and daily updates, is on both plans.

Two details about who you are actually paying. The site is "operated by Trading Strategy DAO LLC. Majuro, Marshall Islands"; the monthly licence is "sold by Market Software Ltd", which is also the copyright holder on both GitHub repositories. The API key arrives by email after payment rather than through an account dashboard.

Data & coverage

Vault data is the live half. The public listing reported 923 stablecoin vaults out of 5,440, $28B of TVL and "updated about 4 hours ago" when read. The metric set is the product: returns net of fees as well as gross, Sharpe and volatility over a fixed three-month window, drawdown, and a technical-risk grade assigned by the vendor.

DEX data is the archive half, and its dates are the thing to check. The available datasets are substantial — 9,100 decentralised exchanges, 392,300 trading pairs, OHLCV candles from one minute to one month (the one-minute file is 1.0 billion rows and 70.6 GB), XY liquidity and TVL at the same intervals, Aave v3 supply and borrow rates at 46.4 million rows, and a daily top-movers file. But on 2026-09-19 the vendor's own "Last updated" column read two months ago for every one of them, and three years ago for the 98 GB Uniswap trades file. These are research archives, not a feed. If your question is what happened on a DEX pair last week, this is not where the answer is.

Chains, from the vendor's own list of "currently active blockchains producing trading data": Arbitrum One, Base, Ethereum, Avalanche C-chain, Polygon, BNB Smart Chain, Berachain, Unichain, ApeX, Blast, Derive, Gnosis, Grvt, Hemi, Hibachi, Hyperliquid, Katana, Lighter, Linea, Mantle, MegaETH, Mode, Monad, Optimism, Plasma, Robinhood, Soneium, Sonic and Tempo. Twenty-nine entries, all EVM or EVM-adjacent. No Solana and no Bitcoin, which rules the product out for a large share of DeFi outright.

Everything ships as JSON and Parquet with Snappy compression, and the download page warns that the one-minute candles expand to several gigabytes in memory — filter the pair universe before you load anything.

Integrations

pip install trading-strategy gets the data client; trade-executor is the framework that backtests and then executes. Both read the same datasets, which is the design's real argument: the strategy you validated against the vault history is the object that later allocates into vaults on-chain, with no rewrite between the two. Execution is on-chain and through the vendor's own integrations rather than through CCXT; centralised-exchange balances are the exception rather than the model.

The AGPL matters here more than the average licence line. It is copyleft with a network clause: if you run modified code as a service other people reach, the obligation to publish your source attaches. For a strategy you run on your own capital it changes nothing. For anything you intend to host for clients, read it before you build.

Limitations

The DEX datasets are two months stale, by the vendor's own column. Named first because it is the single fact most likely to waste a reader's week. The vault data is daily; the DEX data is not.

The vendor ranks vaults and also operates vaults. Three of its own — a multi-vault allocation strategy on Arbitrum, a Hyperliquid vault of vaults, and an options vault — take deposits through the same site that publishes the rankings, under a nav item called "Our vaults". Nothing found on either surface discloses how the two relate, whether its own products appear in the rankings, or how they would be ranked if they did. That is not an accusation; it is a conflict a reader should price in before treating the ranking as neutral.

Returns that read as measurements are computations you cannot audit. The listing shows one-month annualised figures above 9,999% on small vaults, three-month Sharpe ratios on three-month-old products, and a "Protocol Technical Risk" grade with values including "Unknown" and "Dangerous". Annualising a month of a $56,000 vault produces a number, not a forecast, and the risk grade is the vendor's judgement rather than a disclosed model. Read the methodology before you rank on any of it.

No Solana, no Bitcoin, no CEX. The universe is EVM chains and the vaults deployed on them. A DeFi picture that excludes Solana in 2026 is a partial picture, and the card's chain list is the whole of it.

Two legal entities and a DAO. The site is operated by a Marshall Islands DAO LLC; the subscription is sold by a UK-style limited company that also holds the copyright on the code. That is legible enough, but it means the entity you have a contract with is not the entity whose name is on the product.

The free API key gates the dataset table itself. Row counts, sizes and update dates for the vault datasets show as --- until a key is entered, so the free tier's actual contents cannot be sized from outside.

Alternatives

If the question is which vault to allocate to, the closest comparison in this catalogue is the free DeFi yield coverage in On-chain Analytics Platforms — the difference being that this one sells normalisation, net-of-fee returns and a risk grade rather than raw TVL and APY.

If the question is DEX price history for research, judge it on the dates above rather than on the row counts, and check whether a live indexer serves you better.

And if what you actually wanted was a backtesting engine, the rest of Backtesting & Research Libraries is the shorter path. Come here for the framework only if your strategy's unit is a vault position rather than a trade — that is the one thing nothing else in that category models at all.