A system providing candle charts information.
Luminous was developed as part of an application process at a financial institution. The provided service has now also been fully re-implemented and co-exists within this repository for simplicity reasons. A few assumptions had to be made, given that the protocol streams financial data points over multiple websocket connections without further correlation. Please have a look at the corresponding document about that service and protocol.
The commits haven't changed much, besides some repackaging (mostly naming) or feature branch transplantation of the newly implemented service.
A candle aggregates quote information over a period, like 1m, 10m or 1h and so forth. It's also worth mentioning that the close and open price information of adjacent periods could be different, as a quote reflects the most recent price that a buyer and seller agreed upon. It usually composes the following attributes:
| Property | Type | Description |
|---|---|---|
| open | numerical | current price of the first quote related to the underlying period |
| opened | instant | timestamp of the first quote related to the underlying period |
| close | numerical | current price of the last quote related to the underlying period |
| closed | instant | timestamp of the last quote related to the underlying period |
| high | numerical | highest price of all quotes related to the underlying period |
| low | numerical | lowest price of all quotes related to the underlying period |
A candle chart is therefore a fixed-length bucketed stream of quote information, where every bucket is transformed into a candle with the named information, and can be displayed in a linear way. It should be considered that buckets could also be empty if quotes do not change over time. This case is handled using a replication strategy, taking the nearest (time-based) known bucket or candle, respectively. It's fair to also represent the most recent candle with the present information for convenience reasons, even if it's actually not finished or closed.
A time-based bucket uses a half closed interval and is usually aligned with our natural feeling of time, meaning that an hourly chart wouldn't start at 12:14, but rather compose quotes from [12:00, 13:00) and likewise other periods. However, the latter is not a hard requirement.
Numerical price information could be pinned using fix point decimals or treated more leniently using floating values. A technical chart analysis may not require 100% accurate data, but it may also not hurt if it meets the business requirements.
A few REST endpoints are provided to show the raw information about quotes and instruments, further backed by an endpoint displaying the actual chart information. A chart will already be conveniently back-filled. A streaming approach using a "packed" or "compact" protocol would be a better solution, also from a UX point of view. It's only a small portion of the chart that is changing frequently.
Please have a look at the corresponding postman collection below our manual.
Please execute the following steps. The first step needs to be executed whenever the partner service changes or hasn't built at all.
docker-compose buildto build a dockerized version of the partner service. This image is required during the testing phase and can only be obtained through a local build../gradlew buildto build the quarkus application from a shell, which also bundles a docker container.
Please start the system. It's also possible to run the application from Intellij IDEA
docker-compose -f docker-compose.yml -f docker-compose-self.yml upto start the partner service with a default configuration along with luminous.- A postman collection can be found in the manual directory of this project.