Gaps in futures rollovers

Below are the identified bugs and logic oversights in your code that are likely to cause the substantial data gaps around rollovers, as described in your issue:

1. The algorithm does not handle symbol rollover events (such as when a futures contract expires and maps to a new underlying contract). After a rollover, self.symbol no longer reflects the active contract, so consolidators continue to reference and aggregate data from an inactive symbol.

2. The OnData method only checks self.symbol within data.Ticks. If a rollover occurs and a new symbol becomes active, the code will ignore incoming tick data for the new/rolled symbol. This results in missing tick data and incomplete or empty bars.

3. The minute_bars, hour_bars, and daily_bars dictionaries key bar data by the original symbol. If a rollover changes the Symbol object, then new symbol bar data will never be stored or retrieved, breaking access to consolidated bars after a rollover.

4. All TickConsolidator attachments are made only for self.symbol at initialization. When a rollover occurs, no new consolidators are attached to the updated symbol, so minute/hour/daily bars will not be generated for the new contract.

5. There is no transition logic to switch from the old symbol to the new active one, such as responding to a SymbolChangedEvent. Consequently, dictionaries, consolidators, and internal tracking are never updated or re-initialized. This leads to both missed data and disabled logic after the original symbol stops updating.

These are the core oversights that explain persistent data aggregation gaps during and after rollovers. Addressing them should meaningfully improve the reliability of your tick-to-bar processing and feature extraction across contract changes.

Sources: Handling Data - QuantConnect.com, Time Period Consolidators - QuantConnect.com, Consolidate Regression Algorithm, Short Term Reversal With Futures, Multiple Symbol Consolidation Algorithm