{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T15:44:25Z","timestamp":1791992665719,"version":"4.4.0"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":24,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/www.uspto.gov\/ip-policy\/copyright-policy\/copyright-basics"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>We rebuild the displayed outright order book of NYMEX crude oil (CL) and Henry Hub natural gas (NG) futures order by order from CME market-by-order data. We compare 29 releases of the EIA Weekly Petroleum Status Report and 31 releases of the Weekly Natural Gas Storage Report (March to October 2026) with four matched control days each, under pre-registered analysis plans. After the release, depth within five ticks of the mid stays depleted beyond normal days: over the first five minutes after the price reaction the book is missing the equivalent of 33.8 extra seconds of all of its normal near-mid depth in crude and 48.3 seconds in gas. Before the release, depth at the best price is withdrawn without extra cancellation. Before crude releases, cancellations make up a smaller share of the touch withdrawal than on matched days, and missing new orders take up the slack, becoming about as large a channel as cancels (0.38 vs 0.42). About 60% of the cancel-share contrast also appears at 10:30 before the gas report. Before gas releases, more resting orders execute. Release days are Wednesdays and Thursdays and control days are not, so these effects cannot be separated from a Wednesday and Thursday 10:30 effect. We then score the queue position estimators used by open-source backtesters against order-ID ground truth on 141 crude and 125 gas sessions. Their mean absolute fill-time errors range from 0.31 to 1.07 seconds in crude and 0.68 to 1.98 seconds in gas. Their shadow-order fill-rate and P&amp;amp;L biases are small, and their sign depends on how a true fill is read (post hoc). After the gas report two estimators' fill-time errors are lower, but the difference mostly reflects which orders are scored, not better queue modelling, and in an unreplicated readout under the resting-position reading the shadow-order P&amp;amp;L bias of risk_averse and front_optimistic is larger in the same windows. Several pre-registered placebos failed; we report them beside the results that held. Every primary result was reproduced by spec-only re-implementations, run as separate AI-agent sessions set up by the author.<\/jats:p>","DOI":"10.2139\/ssrn.7609738","type":"posted-content","created":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T14:40:56Z","timestamp":1791988856000},"source":"Crossref","is-referenced-by-count":0,"title":["Where Does Liquidity Go Around EIA Releases? Order-Level Evidence From CME Energy Futures, With Implications For Backtest Queue Models"],"prefix":"10.2139","author":[{"given":"Ansh Hemang","family":"Dani","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","reference":[{"key":"ref1","author":"Will J Armstrong","year":"2019","journal-title":"Information shocks, disagreement, and drift"},{"key":"ref2","doi-asserted-by":"crossref","first-page":"916","DOI":"10.1016\/j.jfineco.2021.02.002","article-title":"Information shocks, disagreement, and drift","volume":"140","author":"Will J Armstrong","year":"2021","journal-title":"Journal of Financial Economics"},{"issue":"6","key":"ref3","doi-asserted-by":"crossref","first-page":"569","DOI":"10.1002\/fut.22577","article-title":"Real-time tracking of public announcements in the limit order book","volume":"45","author":"Mehdi Arzandeh","year":"2025","journal-title":"Journal of Futures Markets"},{"key":"ref4","author":"Ansh Dani","year":"2026","journal-title":"mbo-queue: order-id book rebuild and queue-position estimator benchmark for CME MBO data"},{"key":"ref5","author":"Databento","year":"2025","journal-title":"Getting queue position from L2 and order book data. Databento Blog"},{"key":"ref6","article-title":"CME globex MDP 3.0 (GLBX.MDP3) market-by-order and trades data","author":"Databento","year":"2026","journal-title":"Licensed historical data"},{"key":"ref7","doi-asserted-by":"crossref","first-page":"1707","DOI":"10.1093\/rfs\/hhm027","article-title":"Does anonymity matter in electronic limit order markets?","volume":"20","author":"Thierry Foucault","year":"2007","journal-title":"Review of Financial Studies"},{"key":"ref8","article-title":"Queuing and inventories in limit order markets","volume":"75","author":"Corey Garriott","journal-title":"Journal of Financial Markets"},{"issue":"2","key":"ref9","first-page":"65","article-title":"A simple sequentially rejective multiple test procedure","volume":"6","author":"Sture Holm","year":"1979","journal-title":"Scandinavian Journal of Statistics"},{"issue":"509","key":"ref10","doi-asserted-by":"crossref","first-page":"107","DOI":"10.1080\/01621459.2014.982278","article-title":"Simulating and analyzing order book data: The queue-reactive model","volume":"110","author":"Weibing Huang","year":"2015","journal-title":"Journal of the American Statistical Association"},{"issue":"3","key":"ref11","first-page":"1217","article-title":"The jackknife and the bootstrap for general stationary observations","volume":"17","author":"R Hans","year":"1989","journal-title":"Annals of Statistics"},{"key":"ref12","author":"Luca Lalor","year":"2026","journal-title":"Market simulation under adverse selection"},{"key":"ref13","article-title":"Econometric models of limit-order executions","volume":"6257","author":"Andrew W Lo","year":"1997","journal-title":"National Bureau of Economic Research"},{"key":"ref14","doi-asserted-by":"crossref","first-page":"31","DOI":"10.1016\/S0304-405X(02)00134-4","article-title":"Econometric models of limit-order executions","volume":"65","author":"Andrew W Lo","year":"2002","journal-title":"Journal of Financial Economics"},{"key":"ref15","article-title":"Liquidity withdrawal before scheduled macroeconomic announcements","volume":"268","author":"Guanchen Lv","journal-title":"Economics Letters"},{"issue":"1","key":"ref16","doi-asserted-by":"crossref","first-page":"125","DOI":"10.1111\/ajae.12479","article-title":"Is liquidity provision informative? Evidence from agricultural futures markets","volume":"107","author":"Richie R Ma","year":"2025","journal-title":"American Journal of Agricultural Economics"},{"issue":"7","key":"ref17","doi-asserted-by":"crossref","first-page":"1234","DOI":"10.1002\/fut.70106","article-title":"Intraday liquidity in international crude oil futures markets: News impacts, commonality, and spillovers","volume":"46","author":"Hong Miao","year":"2026","journal-title":"Journal of Futures Markets"},{"key":"ref18","author":"C Ciamac","year":"2016","journal-title":"A model for queue position valuation in a limit order book. Working paper, Columbia Business School"},{"key":"ref19","doi-asserted-by":"crossref","first-page":"408","DOI":"10.1016\/j.finmar.2006.04.001","article-title":"Order book characteristics and the volume-volatility relation: Empirical evidence from a limit order market","volume":"9","author":"Randi Naes","year":"2006","journal-title":"Journal of Financial Markets"},{"key":"ref20","year":"2025","journal-title":"at commit 5f3ec40b2afb"},{"key":"ref21","author":"Erik Rigtorp","year":"2013","journal-title":"Estimating order queue position. Blog post"},{"key":"ref22","doi-asserted-by":"crossref","first-page":"584","DOI":"10.1016\/j.eneco.2019.04.026","article-title":"Information content of the limit order book for crude oil futures price volatility","volume":"81","author":"Xiao Tian","year":"2019","journal-title":"Energy Economics"},{"key":"ref23","author":"Yi Wang","year":"2025","journal-title":"Order book curvature as a liquidity measure: evidence on high-frequency volatility in crude oil futures"},{"key":"ref24","author":"H Peter","year":"1993","journal-title":"seconds after the reaction on orders in that snapshot; later fills are fills in the same window on orders that arrived after the snapshot. For each, the size-weighted markout is side \u00d7 (mid t+h -price)\/tick at h \u2208 {1, 10, 60} seconds, reported together with the price jump. The +1 second value is never reported without the +10 and +60 second values. The pre-registered P3 secondary, a queue-position gradient of resting fills' markouts, was not run"}],"container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T14:45:20Z","timestamp":1791989120000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=7609738"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":24,"URL":"https:\/\/doi.org\/10.2139\/ssrn.7609738","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}