FlashAlpha vs OptionMetrics IvyDB 2026 - Historical Options Data

This is the comparison I get asked about most by people who actually do research, and it is the one most often described badly. IvyDB is routinely dismissed as "end-of-day only", which is wrong - OptionMetrics has shipped intraday products for years. The real distinction is subtler and more useful, and it comes down to what is stored and how it is delivered.

Full disclosure: I built FlashAlpha. OptionMetrics has been the standard in this field since 1999 and my data does not go back nearly as far. I will be precise about that.

The TL;DR

What IvyDB actually contains, accurately

Correcting the common mischaracterisation, the IvyDB family is broader than one dataset:

• IvyDB US - a complete end-of-day record of every US exchange-traded equity and index option, including options on ETFs and ADRs, from January 1996. Prices, standardised implied volatilities and option sensitivities, computed consistently across the whole history.
• IvyDB US Intraday - snapshots of option prices and their corresponding volatility calculations at 10:00, 14:00 and 15:45 ET, from January 2018.
• IvyDB Signed Volume - intraday buy / sell pressure in five- and thirty-minute snapshots plus end-of-day, from January 2016.
• IvyDB ETF - separately marketed coverage of options on US-listed ETFs.
• IvyDB Futures - historical futures option prices for US and EU futures markets.
• IvyDB Canada, IvyDB Europe, IvyDB Asia-Pacific and IvyDB Global Indices - international coverage FlashAlpha does not have at all. OptionMetrics also ships IvyDB Beta and IvyDB Implied Dividend.

Signed Volume deserves particular note, because it is the IvyDB product closest to what FlashAlpha does. It classifies trading into buy and sell pressure, which is genuinely adjacent to flow analytics, and it goes back to 2016. If your question is "was volume in this name buyer or seller initiated in 2017", IvyDB answers it and FlashAlpha's flow history does not reach that far.

The real difference: stored values versus stored inputs

Here is the distinction that actually matters, and it is not depth or resolution.

IvyDB stores inputs: prices, implied vols, per-contract greeks, signed volume. Excellent inputs, computed with a consistent and well-documented methodology, which is precisely why it is the peer-review standard. But if you want net gamma exposure by strike, a gamma flip level, a call wall, a regime classification, or a charm and vanna exposure profile, you compute those yourself from IvyDB. The dataset does not carry them.

FlashAlpha stores outputs: the aggregates themselves, already reduced under an explicit and documented dealer-sign convention, at every minute. That is the entire product.

Both positions are defensible and the trade-off is real:

• Storing inputs preserves your methodological freedom. If your dealer-positioning assumptions are your edge, IvyDB lets you express them and FlashAlpha makes you adopt mine. For a fund whose alpha is the positioning model, that is decisive, and it is the honest reason to choose IvyDB.
• Storing outputs removes a build and a class of look-ahead bugs. Recomputing 2019 analytics from 2019 inputs using 2026 code is where point-in-time integrity quietly dies. The value you backtest should be the value that existed.

Three snapshots a day versus every minute

IvyDB US Intraday is real, but it is three fixed snapshots: 10:00, 14:00 and 15:45 ET. For a great many research questions that is entirely sufficient, and for term-structure or surface work it is often all you need.

It is not sufficient for anything whose thesis is about intraday path. A 0DTE gamma study, a question about how positioning shifted through a Fed statement at 14:00, or a signal that fires on a gamma flip crossing during the session, all need the minutes between the snapshots. 15:45 is also a slightly awkward stopping point for anything concerned with the closing auction and end-of-day hedging.

The framing I would use: IvyDB samples the day, FlashAlpha traces it. Which you need is a property of your research question, not a quality ranking.

Research archive versus production feed

This is the difference people notice last and feel most.

IvyDB is delivered as bulk data - files, WRDS, or Snowflake. That is a good fit for research: you load it, you query it, you write the paper or the backtest. It is not a production feed, and it is not meant to be. There is no real-time IvyDB endpoint you point a live strategy at.

FlashAlpha is a live API where history is the same API. The endpoints that serve the current gamma flip level serve the one from 2019-08-14T14:22 with an ?at= parameter and a base-URL swap. The practical consequence is that the code you backtested is the code that trades, with no reimplementation step between research and production, and no chance of the two drifting apart.

If you research on IvyDB and trade on something else, that reimplementation gap is real work and a real source of bugs. That is not a criticism of IvyDB, it is a consequence of it being a research archive, which is what it is for.

The academic licence detail worth knowing

OptionMetrics offers academic institutions a substantially discounted IvyDB licence. Academic distributions have historically refreshed on a slower cadence than the nightly corporate feed - OptionMetrics described an annual academic refresh when it extended IvyDB Europe licensing to universities - but treat the exact cadence as something to confirm with your librarian or with OptionMetrics rather than as a current universal rule, because it varies by product and by distribution channel.

For historical research the lag is irrelevant, which is why it suits universities so well. For anything current it matters: a dataset refreshed on an annual cycle cannot support a study window that includes recent months, and cannot support anything operational. If you are at an institution with IvyDB access and wondering why your data stops well short of today, the refresh cadence is the usual explanation. It is a licence tier, not a fault.

Where OptionMetrics wins

• Thirty years of history. January 1996, every name. FlashAlpha's archive is 75 symbols, the longest-covered 14 starting 2017-01-03 and most of the rest in 2018. For anything touching the dot-com unwind, 2008, or the 2010 flash crash, IvyDB is the only one of the two that can answer.
• It is the peer-review standard. If your work will be published or shown to allocators, "we used IvyDB" is understood and accepted without further argument. That institutional acceptance is worth real money.
• Methodological freedom. Raw inputs mean your conventions, your filtering, your dealer assumptions.
• Global coverage. Canada, Europe, Asia-Pacific and global indices, plus EU futures options. FlashAlpha is US-only.
• Signed volume back to 2016, predating FlashAlpha's flow history.
• Reference-quality standardised surfaces, consistent across three decades, which is genuinely hard to do.
• Snowflake delivery, which suits firms whose research stack already lives there.

Where FlashAlpha wins

• Continuous minute resolution rather than three fixed daily snapshots.
• The aggregation is already done, with a published convention and stated limitations.
• It is a live feed. Research and production share one API and one contract.
• Point-in-time by construction, because the analytic was computed and stored at the time, not recomputed later.
• Regime, walls, flip levels, VRP z-scores and SVI parameters exist as first-class fields rather than as a project.
• You can evaluate it in five minutes on a free tier with no card and no procurement.

Pricing

OptionMetrics does not publish pricing; IvyDB is quoted per institution and varies with products, history depth and delivery method. Anyone quoting you a specific public IvyDB number is guessing, so this page will not. What is publicly documented is the structure:

Who should not use each

Do not buy FlashAlpha if your research needs pre-2017 history, non-US markets, per-contract granularity with your own conventions, or publication-standard provenance. Buy IvyDB. If your positioning methodology is itself your edge, buy IvyDB and build on it.

Do not buy IvyDB if you need a real-time feed, continuous intraday resolution, or you want the aggregates without a build. It is an outstanding research archive and a poor production dependency, because it was never meant to be one.

The combination that actually makes sense

These two compose unusually well, better than most pairs on this site:

• IvyDB for the long sample. Establish that an effect exists across three decades and several regimes, with methodology you control and provenance you can defend.
• FlashAlpha for the live implementation. Once the effect is established, trade it against a feed that computes the same aggregates every minute and replays them identically.

The reconciliation between them is worth doing on its own merits. Compute your GEX from IvyDB for an overlapping date and compare it to FlashAlpha's. Where they disagree you learn something real, either about my conventions or about yours, and two independent computations that agree is a much stronger position than one you cannot check.

Try the overlap

Pick a date you already have in IvyDB and pull the same moment from FlashAlpha:

curl "https://historical.flashalpha.com/v1/exposure/gex/SPY?at=2019-08-14T14:00:00" \
  -H "X-Api-Key: YOUR_KEY"

The at parameter is ET, so 14:00:00 lands exactly on one of IvyDB's three intraday snapshots, which makes it a clean reconciliation point. Replay is Alpha tier and served from historical.flashalpha.com. The methodology whitepaper documents the dealer-sign convention and its stated limitations, so you can see exactly which assumptions you would be adopting before you adopt any of them.

Related reading: the four professional providers compared side by side, why most options backtests are optimistic, and FlashAlpha vs ORATS.

Sources

All figures are as of August 2026. Where a vendor does not publish pricing, the figure is marked as reported rather than stated, and your negotiated number will differ.

• OptionMetrics, data products - IvyDB US from January 1996, and the full product list including ETF, Futures, Beta, Implied Dividend and the international sets.
• OptionMetrics, IvyDB US Intraday - the 10:00, 14:00 and 15:45 snapshot times and the January 2018 start.
• OptionMetrics, IvyDB Signed Volume - five- and thirty-minute intervals plus an end-of-day file, from January 2016.
• OptionMetrics, about - founded 1999; IvyDB US launched the same year.
• WRDS, OptionMetrics - academic distribution and licence terms.

IvyDB and FlashAlpha are not really competitors, they are different halves of a research programme. IvyDB gives you thirty years, global coverage, methodological freedom and provenance that survives peer review; it does not give you a live feed, continuous intraday resolution, or the aggregates without a build. FlashAlpha gives you the computed layer every minute since 2017, live and historical through one API; it does not give you 1996, non-US markets, or the freedom to substitute your own conventions. If you are choosing on depth alone you will pick IvyDB, and you may well be right. If you are choosing on whether research and production can share one code path, that is the case for the other side.