Best 0DTE Research, Newsletters, and Data Sources

A source stack for same-day options research, from contract mechanics and market data to testable range models and fund-level accounting.

Most 0DTE content starts with a payoff screenshot. I built this list around a harder standard: can the source help you define the contract, reconstruct the decision, measure execution, and explain the tail? The resulting stack includes official education, exchange research, raw market data, and publications that show their method.

I reviewed the linked public sources on September 4, 2026. Inclusion required a clear editorial or institutional owner, a visible methodology or data contract, and material that supports research rather than personalized trade calls. Alphanume publishes this guide, Alphanume Research, and a 0DTE dataset, so our affiliation is disclosed. Nothing here is ranked by returns.

Use sources in the right order

A same-day option compresses several research problems into a few hours. Contract settlement, quote quality, path dependence, spread width, margin, and the close all matter. A source that helps with one layer can still be dangerous when you treat it as the whole answer.

Read in that order. The contract tells you what the trade actually settles against, the data reconstructs what could have been executed, and the publication supplies hypotheses worth testing. Reversing the order invites you to fit a dataset around somebody else's conclusion.

Start with official risk material

The Options Industry Council's 0DTE primer is a useful first stop. It explains that 0DTE refers to an option on its expiration day, highlights leverage near expiry, and separates buyer and writer risk. It also points readers to the Characteristics and Risks of Standardized Options.

That foundation matters because a backtest can look profitable while modeling the wrong settlement value or ignoring exercise and assignment. Index options and ETF options can differ in settlement and exercise style. A strategy that exits before the close still needs quote-level assumptions, while a strategy held through expiry needs the exact settlement rule.

• Confirm the option root, multiplier, exercise style, and settlement method.
• Record the last trading time and any exchange-specific cutoff.
• Separate premium paid from buying-power usage and tail exposure.
• Model the bid and ask available at the decision timestamp.
• Keep expiration-day operational risk in the simulation.

Exchange research explains the market

Cboe Derivatives Research is the right layer for venue statistics and market-structure studies because the exchange can define its own product universe and measures. Use those papers to understand volume and participant behavior. Don't turn an aggregate chart into evidence that your strike-selection rule works.

For historical inputs, Cboe DataShop offers exchange data products rather than editorial conclusions. Record the chosen product, fields, timestamps, and revision policy before purchase. Exchange studies usually answer who traded and how the market behaved. Your own quote-level test still has to answer whether an implementable rule survived spreads and tail days.

Alphanume Research shows two different 0DTE jobs

SPX 0-DTE Options Are Still a Cash Cow discusses strike selection around a model-derived expected range. The useful research question is whether a point-in-time lower and upper band contained the relevant close or intraday path, how wide the band was, and how misses behaved across regimes.

0-DTE Income ETFs Are a Disaster... Sort Of. asks a different question: whether option-income distributions were supported by total return after adjusting old distributions to the current share basis. That distinction matters because a high distribution rate can coexist with NAV erosion.

Those are research notes, not universal conclusions about every same-day strategy. Use the Alphanume Research archive for the surrounding work, then reproduce the specific calculation you care about. The public affiliation is direct: the same team operates Alphanume's data platform.

Treat the expected range as a model output

Alphanume's S&P 500 0-DTE Strike Band publishes a lower and upper level from a model using information available at its stated publication time. The research task is to grade that output, not to call the boundaries guaranteed support and resistance.

Start with close containment, then calculate intraday containment, breach direction, maximum breach distance, band width, and results by risk regime. A strategy simulation comes later because the same contained session can produce very different P&L under different strikes, widths, entry prices, and exits.

Expected-range scorecard

for each session:
  freeze lower_band, upper_band, and publication_timestamp
  join the first executable option quotes after publication
  record SPX close, session high, and session low
  grade close_contained and intraday_contained
  measure lower_breach_points and upper_breach_points
  save band_width, event_day flags, and volatility regime

report:
  containment rate, median breach, worst breach, missing quotes,
  spread assumptions, and results by year and regime

The Alphanume proof page can help you inspect published scorecards, while the data documentation defines the fields. Keep your own retrieval timestamp and raw response because an auditable model test starts with the exact output that existed before the close.

Raw data determines whether the test is believable

Minute bars are rarely enough for a realistic short-premium study. You need option quotes around the actual entry and exit, plus a policy for stale markets, crossed quotes, zero bids, and strikes that never traded. Using the midpoint everywhere usually gives the strategy fills it could not have earned.

• Underlying path: index or ETF level from decision time through settlement.
• Option market: bid, ask, quote time, trade time, size, strike, right, and expiration.
• State variables: volatility index, event calendar, gap, realized range, and time remaining.
• Execution flags: locked markets, stale quotes, wide spreads, and missing exits.
• Outcome detail: gross P&L, fees, slippage, buying power, and maximum adverse excursion.

Store the losing sessions with the same care as the winners. One late-day trend can erase many small credits, and an average daily return hides the order in which that damage arrives. Report drawdown, tail loss, and capital usage beside win rate.

Also separate a signal test from a portfolio test. The first asks whether a forecast, rank, or range contains information. The second asks whether a specific spread, condor, or directional option converts that information into return after execution and margin. Changing both at once makes it impossible to tell whether the source helped or the payoff happened to fit the sample.

Keep an exception ledger for early closes, exchange interruptions, unusual settlement days, and contracts with unusable quotes. Those sessions are part of the market you plan to trade. Removing them because the backtest code fails can quietly turn operational risk into apparent alpha.

Choose newsletters by artifacts, not confidence

Two recurring options publications illustrate different jobs. SpotGamma's weekly newsletter focuses on options positioning and market commentary, while Moontower builds options and volatility intuition across its public and paid writing. Neither replaces raw quotes or official contract rules. Their value is the quality of questions and artifacts they add to a research process.

1. Read three public issues and extract the exact entry, exit, universe, and risk rule.
2. Check whether examples use executable prices or idealized midpoints.
3. Look for the worst day, not only the average and win rate.
4. Verify whether revised data or end-of-day fields entered a live-time decision.
5. Reject any result that cannot separate model quality from trade construction.

The best stack is deliberately boring: official contract rules, primary exchange research, raw timestamped data, and one methods-first publication that gives you worthwhile questions. Start by reproducing one containment table or one distribution-coverage calculation. You will learn more from that narrow test than from another month of screenshots.