Product roadmap

Currently, multi-leg products (spreads, strategies, combos) on CME/ICE are hard to use because our instrument definitions do not provide metadata about each underlying leg. The user has to infer the legs from the symbol. This is a form of lossy normalization, since CME/ICE does provide this in their security definitions in a repeating group, but our fixed instrument definition schemas are forced to discard this—they only provide the the instrument_id of the first underlying instrument through underlying_id. In the meantime, our recommendation to users is to either infer this from the symbol OR download the raw security/instrument definitions from the exchange (e.g. CME's is free on their FTP) OR get a pcap subscription from us. If you need historical secdefs copied from CME (since their FTP site only gives 1 day history), we can provide a courtesy backfill of these for a fixed cost.

Historical and live market news.

Currently, indices are indirectly supported through tradable index instruments on CME futures, ETFs, etc. and we don't provide the index values (non-tradable) themselves. This may be sourced from a feed like the Cboe Global Indices Feed or NYSE Global Index Feed.

This feature would allow the user to request trading calendar information (such as trading session start/end times) via our API. This is especially useful when considering trading sessions that can span multiple UTC dates (and hence the possibility of having multiple trading sessions within a single day). Keywords: Market calendar, trading holidays.

Original request from Juan Linares: "Great product but please add dark mode." There are two separate parts to this: Dark mode for the portal and main website (databento.com, databento.com/portal) Dark mode for the docs We can consider this only after Q1 2025 since we're doing a major rebranding of our website which is expected to finish by early April 2025. The new colors will make it easier for us to implement a dark mode.

The legacy live usage-based plans allowed users to access L2/L3 data. However L2/L3 were pulled from the Standard plan. One possibility to increase the value of the Standard plan is to offer limited access to L2/L3, perhaps gated by a quota on symbol subscriptions per account, etc.

Only regular trading hours are supported currently.

It would be great if Japan Exchange Group (Nikkei 225, etc) could be supported. CME Nikkei 225 volume is too low.

This request is to capture all requests for AI skills and tools (i.e Claude skills) related to Databento's APIs.

As described in this issue ticket (https://issues.databento.com/b/6vrl98vl/feature-ideas/incorrect-scaling-of-strike-price-in-definition-schema-for-some-cme-options), CME has inconsistent rules on strike price scaling that varies with product group. There is no systematic way to correct for this and adjustments must be made by hand - this is a faulty issue with CME's own security definition data. Most options trading firms maintain a table of these adjustment/display factors by hand. We decided to make this adjustment under the hood to known products affected by this, so that most of our customers who do not have their own internal adjustment table or are unaware of this underlying CME behavior will not be impacted by this issue. As a downside however, it can be counterproductive for more sophisticated firms that do have their own adjustment/display factors and need to know which strike prices have been adjusted by Databento. At the moment, it appears that the best possible follow-on solution is to provide a separate table of the adjustment/display factors applied by Databento, so firms can revert the data back to the original data.

Currently, we only have BBO data on intervals. It would be great if we could also get interval based snapshots of deeper levels in the orderbook.

Over the next 3 months, Databento will be releasing an expansion of our US equities coverage, new flat-rate tiers across all datasets for our mid-market users (internally referred to as a "Pro" tier) , and reduced fees for some of our Enterprise tier plans. We'll also be making it possible for users to subscribe to live and historical Enterprise plans separately. We believe these changes will make data more accessible to our median spending user and provide most users with more ways to manage variable pricing uncertainty.

I subscribed to wrong plan by accident. Need the Databento US Equities

Summary Add time-interval-subsampled variants of the mbp-10 schema — e.g. mbp-10-1s and mbp-10-1m — that emit a snapshot of the top-10 price levels (bid/ask price, size, and order count per level) at fixed interval boundaries, rather than on every book event. This is the natural extension of your existing bbo-1s / bbo-1m subsampled BBO schemas from 1 level to 10. Motivation / use case A large class of consumers needs the order-book state at a regular cadence (per second or per bar) — for feature engineering, model inputs, backtests, and analytics keyed to OHLCV bars — but does not need every intermediate book update. Today the only way to get periodic 10-level depth is to subscribe to the full event-driven mbp-10 stream and resample it client-side. Motivation / use case A large class of consumers needs the order-book state at a regular cadence (per second or per bar) — for feature engineering, model inputs, backtests, and analytics keyed to OHLCV bars — but does not need every intermediate book update. Today the only way to get periodic 10-level depth is to subscribe to the full event-driven mbp-10 stream and resample it client-side. For liquid names that means ingesting millions of events/day (in our archive, raw mbp-10 runs ~3–5 GB per symbol-set per day) purely to derive, say, 390 one-minute book snapshots per RTH session. The full stream is downloaded, decoded, and discarded — wasteful on egress, storage, and compute for both sides. Precedent Databento already subsamples top-of-book with bbo-1s and bbo-1m (book state as of each interval boundary, timestamped at the boundary). This request simply applies the same interval-sampling semantics to the existing 10-level mbp-10 record. The wire format can reuse the Mbp10Msg struct unchanged. Proposed behavior Schemas: mbp-10-1s, mbp-10-1m (and mbp-10-1h / mbp-10-1d if cheap to include).Record: identical to mbp-10 (10 levels × {bid_px, ask_px, bid_sz, ask_sz, bid_ct, ask_ct}); reuse Mbp10Msg.Sampling: snapshot the prevailing book state as of the end of each interval, aligned to the same UTC boundaries as the ohlcv- / bbo- schemas, so records join 1:1 to bars.Availability: Historical first (server-side resample of your archive is the high-value piece); Live periodic snapshots a plus. Edge cases to define Intervals with no book change → carry forward last known state, or omit? (Matching bbo-1s/bbo-1m behavior is ideal.)Pre-open / halted / crossed books and partially-populated depth (

My account has been banned.