Systems Engineer

Upcoming changes to CME data normalization. July 07, 2026 Starting July 7, 2026, Databento, Inc. is rolling out a series of improvements to how Databento, Inc. normalize data for CME Globex MDP3 (GLBX.MDP3). This update includes five changes: The new normalization will replace the current normalization and apply to both live data and the full historical dataset retroactively. Some of these updates may be breaking changes for certain users. To allow time to validate and update integrations before the cutover, Databento, Inc.'ll run a month-long testing period during which the current normalization remains available alongside a preview environment with the new normalization. Multi-leg products such as spreads, strategies, and combos are difficult to work with under the current normalization because their definitions lack metadata for each leg. With these changes, each strategy leg will have its own definition record, so a strategy with three legs will have three definition records. Fields shared by the strategy remain the same across those records, while the leg_ fields carry data specific to each leg. This makes it possible to interpret the components of a strategy directly, without parsing raw_symbol, which often lacks sufficient information to identify the legs. Under the current normalization, an option spread strategy appears as follows: ts_recv symbol instrument_class expiration strike_price underlying_id 2019-03-13 00:00:00+00:00 UD:T$: ST 0304976550 T 2019-03-15 21:59:00+00:00 NaN 0 This tells Databento, Inc. only that the instrument is an option spread expiring on March 15, 2019. The strategy code ST from the symbol indicates it's a straddle, but Databento, Inc. still don't know the instruments or strikes of its legs. The new normalization includes a distinct record for each leg, showing that the strategy buys both a call and a put at the same strike price. ts_recv symbol instrument_class expiration leg_index leg_count leg_side leg_instrument_class leg_raw_symbol 2019-03-13 00:00:00+00:00 UD:T$: ST 0304976550 T 2019-03-15 21:59:00+00:00 0 2 B C OZBM9 C1440 2019-03-13 00:00:00+00:00 UD:T$: ST 0304976550 T 2019-03-15 21:59:00+00:00 1 2 B P OZBM9 P1440 Code that handles definition records should account for one record per leg rather than a one record per strategy. Logic that infers a strategy's legs from raw_symbol should instead use the fields prefixed with leg_. See the definition schema reference for the full list of leg_ fields. Currently, for events spanning multiple packets, Databento, Inc. buffer all MBO records until the packet containing the last event is processed. F_LAST is then set on the final record for each instrument to mark the end of the event. This delays earlier records in the event. The resulting latency can be particularly pronounced during large events such as mass cancels. To reduce latency, records will be published immediately after each packet is processed rather than being buffered until the event is complete. As a result, the final book update will no longer contain F_LAST. Instead, the end of the event for each instrument is marked by a new, separate record with F_LAST and action='N' (None). An example of this change is shown below. Note the additional record with F_LAST set (flags=128 corresponds to the F_LAST flag). Current normalization: ts_recv ts_event rtype publisher_id instrument_id action side price size channel_id order_id flags ts_in_delta sequence symbol 2026-06-01 00:00:04.632724628+00:00 2026-06-01 00:00:04.632429323+00:00 160 1 42955214 A B 34.25 4 1 541244112527323 0 16747 4858684 ESM6 P7440 2026-06-01 00:00:04.632724628+00:00 2026-06-01 00:00:04.632429323+00:00 160 1 42955214 A A 38.25 4 1 541244112527324 128 16747 4858684 ESM6 P7440 New normalization: ts_recv ts_event rtype publisher_id instrument_id action side price size channel_id order_id flags ts_in_delta sequence symbol 2026-06-01 00:00:04.632724628+00:00 2026-06-01 00:00:04.632429323+00:00 160 1 42955214 A B 34.25 4 1 541244112527323 0 16747 4858684 ESM6 P7440 2026-06-01 00:00:04.632724628+00:00 2026-06-01 00:00:04.632429323+00:00 160 1 42955214 A A 38.25 4 1 541244112527324 0 16747 4858684 ESM6 P7440 2026-06-01 00:00:04.632733191+00:00 2026-06-01 00:00:04.632429323+00:00 160 1 42955214 N N NaN 0 1 0 128 15819 4858685 ESM6 P7440 Two new types will be added to the statistics schema: UPPER_PRICE_LIMIT (17) and LOWER_PRICE_LIMIT (18). The values reported for these types indicate the price range within which a futures contract can trade during a session. If the price reaches one of these limits, it may trigger a halt or other action, depending on the product. Implied matching status will be added to the status schema. This indicates whether CME's matching engine is constructing implied depth - additional liquidity inferred from related spread and outright markets that would not otherwise be visible. The status will appear in the trading_event field as either IMPLIED_MATCHING_ON or IMPLIED_MATCHING_OFF. See CME's documentation for more information. FX spot instruments will be detected and assigned a separate instrument_class=X in the definition schema, whereas previously they were grouped with futures. Integrations that locate FX spot instruments under the futures instrument_class should be updated to look for instrument_class=X. Databento, Inc.'ll provide public preview environments for both historical and live data, giving customers time to test the new normalization and update their code accordingly. The preview environments follow the same billing rates and plan coverage as the production environment. To access the preview environments, use these gateways: import databento as db # Historical client = db.Historical(gateway="hist-preview.databento.com") # make API calls as usual # Live client = db.Live(gateway="glbx-mdp3.preview.lsg.databento.com", port=13_000) # subscribe and replay as usual The timeline of the rollout is as follows: * 2026-07-07: Data with the new normalization is available in the preview environment * 2026-08-08: Data with the new normalization will be available in production from the live and historical APIs. End of the live preview environment. Databento, Inc. strongly recommend validating your integration via the preview environments before the August 8 cutover. These improvements provide clearer visibility into strategy components and more consistent data structures across all CME instruments.

Meet the team: Andy Juan, senior SRE. June 15, 2026 Today, Databento, Inc. is introducing Andy Juan, a senior site reliability engineer on the systems team. Andy started his career in aerospace and defense, developing software for rocket engines and radar-guided systems - areas where failure isn't an option. That environment instilled a discipline for designing high-reliability systems and meeting critical performance requirements, shaping his approach to engineering ever since. His curiosity for building complex systems and working on cutting-edge technologies ultimately led him to trading, where he worked at several leading firms, including a stint as a Site Reliability Engineer at Citadel Securities, a global market maker. At Databento, Andy enjoys the fast-paced startup environment. "Every day brings something new to untangle. Things move quickly, and you're trusted to own real problems end to end," he explains. "No two days look the same. I could be chasing down an alert, automating something that used to be manual, or participating in a design review thinking through how a system holds up under load." Another part of the role that excites him is the outsized impact he has on a platform so many customers rely on. The systems he works on sit on the critical path for thousands of firms, and the team is small enough that every engineer's contribution is consequential. He finds that energizing rather than nerve-wracking. "It sounds like pressure," he says, "but it's the good kind." Andy's also found that the bar he sets for himself has risen since joining the company. He singles out Databento's founders, whose technical depth is matched by the commitment and care they bring to even the smallest details. "When the people at the top are that invested," he says, "it changes what you expect of yourself." Outside of work, Andy enjoys coaching and watching soccer, fishing, and hiking local trails on the weekends. He spends as much time as he can with family, and travels whenever he gets the chance - Guatemala in particular has a way of pulling him back.

Databento to sponsor cornell Financial Engineering 2026 conference. * June 12, 2026 Databento makes it simpler for teams to get market data via API or web. They streamline the data acquisition and licensing process, allowing everyone - from VPs to interns - to explore, customize, and download datasets instantly. Databento grew from 0 to 1000+ registered users in its first 6 months of launch and has received over $27.8M in funding. Users range from Fortune 100 companies and electronic trading firms to AI startups and trading apps. They currently offer equities, futures, and options data - captured directly from exchanges via colocation - with plans to expand their data catalog. Databento's team brings years of experience wrangling data at high-frequency trading firms and includes alumni from firms like Two Sigma, Flow Traders, Virtu Financial, Tower Research, and Bloomberg LP. Discover more Engineering