How a prediction market on Katana eliminated missing transactions with Ormi's real-time blockchain data
Foresight, a prediction market built on Katana Network, faced missing transactions due to in-house indexing. By integrating a real-time blockchain indexer, they eliminated data gaps and restored trust. Here’s what every team launching a prediction market should learn.
Data infrastructure is rarely production-ready when building on a new chain. This quickly becomes a problem if the product depends on an accurate on-chain state.
Many are left with two options:
- Either build an internal indexing pipeline or
- accept gaps in the user experience.
Foresight, a prediction market built on Katana Network, experienced this firsthand. Their journey highlights a reality many Web3 teams eventually face.
Your data layer is not backend plumbing. It is the product.
Building a prediction market on Katana Network
Foresight is a social-first prediction market focused on Asia. Unlike most prediction markets concentrated in the US and Europe, Foresight deliberately chose to operate in regions that demonstrated strong trading activity and fast-growing user bases.
At launch, Katana Network was still early in its ecosystem maturity. There were no hosted blockchain indexers available on the network.
For a prediction market, indexed data is critical:
- User positions must render accurately
- Trades must appear immediately
- Market resolution must reflect the correct state
- Transaction history must be queryable
Without indexed blockchain data, none of this works reliably.
The initial approach: in-house indexing
With no indexer available on Katana, Foresight built its own indexing pipeline.
Their system:
- Listened to on-chain events
- Parsed contract logs
- Stored results in their own database
- Used that database to power frontend queries
This is a common approach on new chains. It works in development, and also works under a light load.
But it introduces hidden risks:
- Reorg handling must be implemented correctly
- Event gaps must be detected
- Crash recovery must be deterministic
- Backfills must be idempotent
- High transaction throughput must not overwhelm ingestion
These are non-trivial infrastructure problems.
The beginning of missing transactions
Once real users began trading, issues surfaced quickly.
Users reported:
- Missing positions
- Trades not appearing in UI
- Delayed market state updates
The pipeline could not consistently keep up with real-time chain activity. Under load, events were dropped or processed out of order.
For a prediction market, this was a critical problem and could impact the survival of any application involving financial value.
If users cannot verify that their trades were executed:
- They stop trading.
- They lose trust.
- Liquidity declines.
In financial applications, stale or missing data directly impacts product credibility.
Running indexing in-house is extremely hard
Manual indexing pipelines often fail for predictable reasons:
- If the chain reorganises, previously processed events must be rolled back. Many homegrown systems do not implement deterministic rollback correctly.
- Throughput spikes expose system scalability limitations. Systems that work at 5-10 transactions per second may fail at 50-100.
- Event listeners are not state systems. Listening for events is not the same as maintaining a consistent, queryable state model.
- Operational burden grows quickly as engineers end up debugging ingestion issues instead of building features.
Foresight reached this inflection point.
The solution: integrating a real-time blockchain indexer
Foresight integrated Ormi’s real-time blockchain indexing infrastructure on Katana.
Instead of relying solely on a custom listener pipeline, they deployed a production-grade indexing layer designed to:
- Stay synchronised with the chain head
- Handle reorgs automatically
- Support higher request throughput
- Serve consistent GraphQL queries
The transition was straightforward.
Within days of integration:
- Missing transaction reports stopped
- User complaints dropped to zero
- Frontend state aligned consistently with on-chain reality.
“Our in-house pipeline worked at low volume, but it wasn’t deterministic under load. With Ormi, we stopped worrying about reorgs and missed events. The indexer stayed aligned with the chain head, and that eliminated the daily transaction complaints.”
— Eason Chai | Founder & CEO
Why Foresight chose Ormi
Foresight evaluated indexing providers on three dimensions:
Sync Speed
How quickly does the system reach the current chain head? Chain-head lag directly affects UI freshness.
Request throughput
How many queries per second can the API handle? Prediction markets can generate frequent user reads, especially during volatile events.
Availability
Is the indexer consistently accessible? Downtime equals invisible transactions and degraded UX. These are the metrics that matter in real-world Web3 applications.
Preparing for scale: multi-chain and order books
Foresight is expanding to Base and BNB, adding multi-chain complexity.
They are also launching an order book system, which introduces:
- Higher transaction frequency
- More contract interactions
- Greater indexing complexity
- Increased query load
As they noted:
“With the order book, there is going to be a lot more transactions happening. Indexing complexity will increase significantly compared to our AMM infrastructure.”
Scaling transaction volume without scaling indexing reliability is a recipe for failure.
By migrating to a real-time Web3 indexer early, they avoided compounding data-layer risks as the product evolves.
Learnings from building a prediction market
Foresight’s experience on Katana illustrates a broader pattern across Web3:
If an application depends on real-time crypto data, indexing infrastructure is fundamental to the user experience.
In-house solutions require too many engineering resources and RPC alone cannot serve historical or structured queries at scale. .
For teams building prediction markets, exchanges, DeFi protocols, or any application where the financial state must remain consistent with the chain tip: Treat your data layer as production infrastructure from day one.
Once users start trading real money, missing transactions are no longer a technical bug, but an incomplete product.
Full customer story
About Foresight
ForesightAsia-focused is an Asia focused social-first prediction market that rewards users for sharing their convictions. Foresight harnesses collective intelligence by aggregating real-time insights and surfaces nuanced information through a social discovery layer.
About Ormi
Ormi is the next-generation data layer for Web3, purpose-built for real-time, high-throughput applications like DeFi, gaming, wallets, and on-chain infrastructure. Its hybrid architecture ensures sub-30ms latency and up to 4,000 RPS for live subgraph indexing, sub-10ms responses across 100+ API schemas, and a powerful SQL engine for historical and AI-ready data.
With 99.9% uptime and deployments across ecosystems representing $50B+ in TVL and $100B+ in annual transaction volume, Ormi is trusted to power the most demanding production environments without throttling or delay.