[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-85272":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":12,"openIssues":13,"contributorsCount":13,"subscribersCount":13,"size":13,"stars1d":14,"stars7d":14,"stars30d":14,"stars90d":13,"forks30d":13,"starsTrendScore":15,"compositeScore":16,"rankGlobal":9,"rankLanguage":9,"license":9,"archived":17,"fork":17,"defaultBranch":18,"hasWiki":17,"hasPages":17,"topics":19,"createdAt":9,"pushedAt":9,"updatedAt":20,"readmeContent":21,"aiSummary":9,"trendingCount":13,"starSnapshotCount":13,"syncStatus":22,"lastSyncTime":23,"discoverSource":24},85272,"dex-tradeability-study","DaruFinance\u002Fdex-tradeability-study","DaruFinance","Empirical study of tradeability in DEX-only cryptocurrencies (analysis engines + paper source). Working paper.",null,"Python",121,1,0,6,18,49.5,false,"main",[],"2026-06-17 15:04:39","# No Edge Without Information\n\n### An empirical study of tradeability in decentralized-exchange-only cryptocurrencies\n\n**Daniel Gatto**\n\nWorking paper, under review · [SSRN 6858778](https:\u002F\u002Fpapers.ssrn.com\u002Fsol3\u002Fpapers.cfm?abstract_id=6858778) · [Interactive write-up](https:\u002F\u002Fwww.daru.finance\u002Fresearch\u002Fdex-only)\n\n## Overview\n\nThis repository contains the analysis engines for an empirical study of\n*tradeability* in DEX-only cryptocurrencies: tokens that trade on decentralized\nexchanges and have never been listed on a centralized venue. The study assembles\na survivorship-aware corpus of 4,990 trading pairs across 27 chains and asks a\nnarrow, falsifiable question: net of realistic execution costs, is there any\nexploitable edge available to a price-based timing strategy in these assets?\n\nThe methodology is deliberately conservative. Every backtest is **long-only**\n(spot tokens cannot be shorted without a borrow facility), carries a **per-fill\ncost model** (swap fee plus constant-product slippage plus gas, applied on each\nentry and exit), and is evaluated under **walk-forward** out-of-sample testing\nwith in-sample parameter selection and intrabar take-profit\u002Fstop on the bar high\nand low. Crucially, every reported result is paired with a **mandatory\nbar-shuffled null control**: the identical pipeline is re-run on surrogate series\nthat destroy temporal order while keeping each coin's return distribution, so any\napparent edge can be measured against what pure noise produces under the same\nselection procedure. Results are reported as **full distributions** rather than\ncherry-picked top performers, and the corpus is **survivorship-aware** — sourced\nfrom currently-live pools, so every cross-sectional and holding metric is read as\nan upper bound.\n\nThe central finding is that price-based timing carries **no edge** in this\nuniverse once costs and the null control are accounted for: the live\ndistributions are statistically indistinguishable from, or worse than, the\nshuffled-bar nulls. A large battery of extended studies (order-flow features,\ncross-sectional selection, machine-learning ranking, liquidity-provision and\nmarket-making proxies, launch-cohort and dispersion effects) is reported in the\nsame disciplined frame. The one qualified positive is a **leakage-free\ncross-sectional machine-learning ranker** (point-in-time features only): its\nout-of-sample rank-IC is ~0.06–0.08 at the 7- and 14-day horizons\n(label-permutation p ~ 0.005), and insignificant at 30 days. It does not\ngenerate alpha; its skill is **crash-avoidance** (every predicted decile's\nmedian forward return is negative, so it flags which coins fall, not which rise),\nand a long-only AMM trader cannot reach the left tail. We therefore report it as\na lead, not an edge. See `ml_ranker_pit.py` for the leakage-free re-run (the\nearlier `ml_ranker.py` used end-of-sample snapshot features and is superseded).\n\n## Repository layout\n\n- `engines\u002F` holds the core analysis and data-collection scripts: phased pipeline\n  (`phase1`…`phase10`), GeckoTerminal universe and OHLCV ingestion (`gt_*`),\n  the centralized-listing exclusion filter (`cex_filter*`), and the on-chain\n  flow collectors (`bitquery_flow.py`, `tokenapi_flow.py`). These cover universe\n  construction, baseline backtests, walk-forward optimization, flow-signal\n  selection, and the null controls.\n- `chainscope\u002F` holds the supporting Python package: the per-fill cost model\n  (`costs.py`), data storage and caching, chain definitions, provider clients,\n  and shared utilities.\n- `gap-studies\u002F` holds fifteen self-contained extended-study engines, one per\n  subdirectory (beta-hedged baskets, BSC order flow, BTC-regime timing,\n  cost\u002Fdepth sensitivity, intrabar TP\u002FSL bracketing, in-sample noise fitting,\n  launch cohorts, LP market-making, cross-sectional and meta selection,\n  ML cross-sectional ranking, multipool dispersion, and a portfolio of marginal\n  streams).\n- `web-demos\u002F` holds the React\u002FTypeScript sources for the interactive figures\n  in the companion write-up.\n\n## Reproducing\n\nThe underlying market data is **not** included in this repository (it is large\nand licensed from third-party providers). Scripts expect a local `.\u002Fdata`\ndirectory and read all provider credentials from **environment variables**\n(e.g. `COINGECKO_API_KEY`, `BITQUERY_TOKEN`, `THEGRAPH_TOKEN_API_KEY`,\n`BIRDEYE_API_KEY`, `MORALIS_API_KEY`, `HELIUS_API_KEY`). Set these in your\nenvironment before running any collector. With data and keys in place, the\nphased engines in `engines\u002F` reproduce the panel, the backtests, the\nwalk-forward results, and the null controls in sequence.\n\nCompanion write-up with interactive figures:\n[daru.finance\u002Fresearch\u002Fdex-only](https:\u002F\u002Fwww.daru.finance\u002Fresearch\u002Fdex-only)\n\n## Data\n\nThis study was made possible by on-chain and market data from\n[Bitquery](https:\u002F\u002Fbitquery.io) and [CoinGecko](https:\u002F\u002Fwww.coingecko.com). A\nsubstantially expanded version 2, with broader coverage and deeper analysis\nthanks to their support, is in progress.\n\n## Cite\n\n> Gatto, D. V. (2026). *No Edge Without Information: An Empirical Study of\n> Tradeability in Decentralized-Exchange-Only Cryptocurrencies.* SSRN Working\n> Paper 6858778. https:\u002F\u002Fpapers.ssrn.com\u002Fsol3\u002Fpapers.cfm?abstract_id=6858778\n\n```bibtex\n@techreport{gatto2026dex,\n  author = {Gatto, Daniel V.},\n  title  = {No Edge Without Information: An Empirical Study of\n            Tradeability in Decentralized-Exchange-Only Cryptocurrencies},\n  year   = {2026},\n  type   = {SSRN Working Paper},\n  number = {6858778},\n  url    = {https:\u002F\u002Fpapers.ssrn.com\u002Fsol3\u002Fpapers.cfm?abstract_id=6858778}\n}\n```\n\n## Status\n\nWorking paper, under review. Posted on SSRN:\n[papers.ssrn.com\u002Fabstract=6858778](https:\u002F\u002Fpapers.ssrn.com\u002Fsol3\u002Fpapers.cfm?abstract_id=6858778)\n",2,"2026-06-17 04:12:51","CREATED_QUERY"]