We are still doing research on this subject and making interfaces.
Some ideas that may be interesting for other similar projects:
Marketing:
- 30% commission from fees of invited client or maybe up to 70% in the beginning as reward to the most active affiliates
- Find infuencers ot FaceBook, Twitter and forums
- translate on different languares
- different typ of contests
- top traders with max margin
- blog posts with interesting stats
Monetisation/Revenue model:
- trading fee 0.04%-0.1%
- widrawal fee
- listing fee
- other income like promo fees or cross sales
Order matching engine:
While doing research i've found interesting articles about order matching engines and open source projects for limit order book here some of them.
WK Selph's Blogpost:
- https://[Suspicious link removed]/KF1SRm - How to build a fast limit order book
Nice implementation on C++:
2 000 000 - 2 500 000 orders per second
https://github.com/objectcomputing/liquibookThe Buttercoin crypto exchange source code. But it's written on Scala...
https://github.com/buttercoinPort of winning
www.quantcup.org competition entry (implementing a fast stock exchange matching engine for an HFT bot) from C to Go
https://github.com/rdingwall/go-quantcupNice thread at stackoverflow aboput matching engines
https://quant.stackexchange.com/questions/3783/what-is-an-efficient-data-structure-to-model-order-bookWow very impressive! 16 million orders per second
https://github.com/charles-cooper/itch-order-bookTechnologies will use for the crypto exchange:
- programming languages: JavaScript, PHP, C/C++, GoLang
- databases Redis and MySQL/PostgreSQL
In the beginning use Golang for the trade engine and then rewrite it to C/C++
Test engine on old trading records.
Security standarts.
This is the most important part that many stock exchange forget about.
https://en.wikipedia.org/wiki/ISO/IEC_27001CryptoCurrency Security Standard
https://cryptoconsortium.github.io/CCSS/Make several level of security.
1st Level that interacts with visitors
at this level data from client will be passed to 2nd level middle/proxy servers. No direct access to trading engine from this servers.
They just filter data show nice looking interfaces and prepare data for the 2nd level servers
This servers will be replaced every 3 days with with clear new servers and checked for intrusions.
2nd level
Check the data and route it to the 3rd level servers.
At this level data will be checked for intrusions and filtered. The main goal of this layer is to alert intrusions.
This servers will be replaced every 3 days with with clear new servers and checked for intrusions.
3rd level
The most impotant servers. Access to this servers is only from private network. No connection from the world.
Simply saying split all system to small parts and monitor every activity + update and refresh the system to protect from intrusions
Few more things:
- Disable tracert and ping requests on all servers.
- Close all ports. Monitor connections, transferred data to the servers and report suspect connections.
- Add "honeypots" to the servers
Monitor suspect transactions.
- e.g. mass funds withdrawal or transfer to one account. Unusual funds transfer activities. Untill funds are inside of engine we can automatically stop suspect transactions.
- additional authentification if transaction goes in unusual way (different IP address, different browser, no cookie and etc). Collect information about usual client routes.
- duplicate all withdrawal transactions
- minfy and obfuscate all code that will be on real servers
- store funds on several wallets to split the risks
My best regards,
Valera Vlasyuk
Uexchange - Unnamed Exchange