search

bmahakud / WorkReport-Shradha

0 stars 0 forks source link

Finalising the scope of the additional work that done for the AnyQuant.- 9th june #76

Open ss-sahoo opened 1 month ago

ss-sahoo commented 1 month ago

1)Automation Data upload 2)Trade exicution 3)Estimate computing costs for a single strategy optimisation 4)Multiply that for multiple users (e.g., 10 users doing 3 optimisations/day) 5)Compare costs across the 3 scenarios i)Entirely Web-Based (Cloud Hosted) ii)Web App + Local/Virtual Computation (Hybrid) iii)Fully Desktop-Based App

bmahakud commented 2 weeks ago

Resolve this ticket by 23th June 2025.

ss-sahoo commented 2 weeks ago

1) Automation Data Upload Discussion is ongoing. This mainly depends on which trading platform we choose to fetch the data from. The platform should provide the correct values required to fetch accurate and complete data.

2) Trade Execution Same as above — it depends on the trading platform chosen, as different platforms handle execution differently. We will discuss this with Bharath, and after that, we will estimate the time required.

3) Estimate Computing Costs for a Single Strategy Optimisation On a machine with 8 GB RAM and 4 physical cores, it takes approximately 5–6 hours to run a standard strategy optimisation.

4) Multiply That for Multiple Users We need to estimate the load and compute cost if, for example, 10 users each run 3 optimisations per day.

5) Compare Costs Across 3 Scenarios i) Entirely Web-Based (Cloud Hosted) ii) Web App + Local/Virtual Computation (Hybrid) iii) Fully Desktop-Based App

We have selected option ii) Web App + Local/Virtual Computation (Hybrid). In this approach, backtests can run on the server, but optimisation must run locally because it takes more time and computing resources.

ss-sahoo commented 2 weeks ago

Study on RAM and CPU Usage During Optimization (with Other Tabs Open like Chrome): I ran an optimization script (optimise_run.py) on my system while other applications like Chrome were also running in the background. Here's what I observed:

System Setup: Total RAM: 15.5 GB CPU Cores: 12 (100% utilized during optimization) Number of Processes: 13 (spawned by the main script)

I ran an optimization script while Chrome and other apps were open. The script used about 7.6 GB RAM and 100% of the CPU, while Chrome used around 1.2 GB RAM. The total time taken to finish the optimization was 73 minutes and 22 seconds. Closing Chrome may slightly reduce the time, but since the CPU was fully used, the impact would be minimal.

Time Taken to Complete Optimization: Total Time: 1 h 13 minutes