You start with an idea, let’s say you have a basic strategy in futures. You would then work with the data team to gain access to the necessary data and partition the data into in-sample and out-of-sample. Next step should be to work on the development code to test out the idea on the in-sample while looking at the quality of the data and outliers. If your idea meets the in-sample performance benchmark then you would test out the idea on the out-sample data. Should the out-data meet expectations then you would work on developing trading logic and risk management using our in-house backtesting platform. If all is going well then you would deploy the strategy in paper trading mode and track each day’s performance of paper trading and compare it with the back test from the previous day. Supported by the operations team the entire way, you would thoroughly test the strategy before going live; for instance, if some inadvertent look-ahead bias is introduced in the research process that could be caught in this paper-trading phase. When all benchmarks be met then your strategy is set to actual live trading!
In general, the process to write the code over an hypothesis or theory is basically the same follow this flow below.
Watching the market movements makes you understand the cycle.
Translate your conclusions to comprehensive rules using market data and indicators.
The process where you convert the algorithm in software.
Important! Are the rules following the expected path? We do plenty of tests for that.
Time to check the efficiency and performance. Results might request improvements.
When all things are ready to go, strategies or indicators can be turn on in realtime market and live accounts.