A major feature addition in Nexa's Trading platform
In 2009 Desjardins commissioned Nexa to build the Conditional orders feature into their desktop DDPlus platform. Conditional orders were a major addition to our feature set.
Simple orders are used for buying & selling a single stock for a specific quantity and price. Conditional orders are actually collections of simple orders which can be set up so that the behavior of one order in the group can influence another order. Some of the behaviors that can trigger actions in other orders include an order being executed, or a particular stock price reaching a threshold. Some of the resulting actions that can be configured are sending an order to the exchange or canceling other orders in the group.
This project was a multi-team effort and my contribution spanned the following areas:
- Developed the front end code in the desktop client to display conditional orders accurately in the order panel displays including reflecting real- time updates in the orders.
- Developed backend server code, also in C++, to read conditional orders from the database.
- Wrote custom serialization & deserialization code on client and server to transmit conditional orders between client and server.
- Wrote an order caching mechanism in the desktop client to allow the application to reassemble conditional orders from simple orders received from servers at different times. This was necessary because there were other servers in our system that could only transmit conditional orders as fragmented simple orders. The challenge here was presenting conditional orders as a consistent way to the user regardless if the conditional order was received in fragments or complete.
C++ Code Sample OrderCacheManager class (PDF)
3 categories of conditional orders were defined:
- Trade triggers order (TTO) is a conditional order configuration with 1 primary order and n child orders. Only when the primary executes on the exchange are the other n child orders sent to the exchange.
- Trade cancels order (TCO) is a conditional order configuration with n child orders where all of them are sent to the exchange at the same time. The execution of the first of these child orders will trigger the cancellation of the remaining child orders.
- Event triggers order (ETO) is similar to a TTO, however, instead of primary order triggering the release of the children, it's a quote event that does this job. The child orders are triggered by a stock price reaching a threshold.
For detailed definitions of Conditional orders see Investopedia.
The conditional order required us to define a new order type in our system which was a recursively nested tree structure. This structure could contain n orders nodes and where each order node could branch out to child decedents. A descendent could be an entirely new conditional order or a simple order.
The type of conditional order that best illustrates the nested structure is the Trade-triggers-order (TTO) type. At the topmost layer, a TTO has one simple order, which branches out to a Trade-cancels-order (TCO) conditional order. TCOs contain up to 3 simple orders. The strategy behind this type of conditional order is as follows: In the beginning, the topmost singleton order is sent to the market, while the TCO orders are held back until the triggering condition. The execution of primary singleton order triggers the 3 singleton orders in the TCO to go to market. When the first of those 3 singleton orders execute, the remaining 2 singletons in the TCO will be canceled.
I was responsible for creating the real-time display and the user interaction for conditional orders on the DDPlus order panels. I displayed the conditional order structure as nested folders because I needed to show these orders in a consistent manner with legacy style orders (see screenshot above). I wrote the functionality to be able to add, modify and delete child orders as context menu actions on the orders panel. I also wrote a client side order management & caching mechanism that served 3 goals:
1) To cleanly separate out the storage of order data from the orders panel view.
2) Lookup mechanism to find a conditional order or child order's data when it's being manipulated by the user on the order panel.
3) Reconstruct conditional orders when they are received as singleton orders.
