Basic Architecture for Algo trading

batbark
Hello Traders,
I am new to the AlgoTrading world and willing to spend all the time I have to learn it.
I have a mechanical engg bg. My current knowledge with programming is restricted to very basic Python, SQL.
I am looking for a basic architecture(stack) with the flow that the algo traders use to create their own automated trading system.

My needs:
- Intraday trading
- To go through all the nifty 200 stocks live
- Implement the strategy live
- custom screener for nifty 200 stocks for manual placing
- Semi-automate or automate the trades.
- Avg. number of trades/day = 20 to 30

currently going through pandas, quandl, zipline.

I am looking for a basic end to end architecture(stack) with loose coupling(live data stream source can be from kite or other) to build upon with
case 1: using the local machine (MacBook pro with 10gb ram) as a server (with pros and cons)
case 2: using AWS service (not there yet but will be there someday as I progress)
example info to be included :
1) should I store the tick data with sql in a tabular format and implement the strategy?
2) what happens if I loose the internet connection, will I be able to retrieve the minute tick data once the connection resets?

Also please share your experience while building your system and the time it took.

Thank you in advance. :)
  • Vivek
    Hi, Welcome to Kite Connect developers community
    1) should I store the tick data with sql in a tabular format and implement the strategy?
    Its always good to store persistent data in database and load the required data in memory or memory based key value storage db like redis while processing the data. This will reduce your IO and make things faster.
    2) what happens if I loose the internet connection, will I be able to retrieve the minute tick data once the connection resets?
    You can user our historical data API for retrieving the missing data.
  • satveersm
    Do i need any separate subscription for historical data, i already have kite connect subscription.
  • Vivek
    @satveersm Yeah historical is an add on you can subscribe for your app and it also costs same price as app subscription.
  • Apoorv
    @batbark interesting to know that you are exploring pandas, quandl & zipline for algo trading. I am at the same stage right now. I am using Datacamp to learn Pandas and trying to learn zipline on my own. I was thinking that I should learn zipline first, backtest a few strategies in the Indian scenario and then come to the api part of Zerodha, but maybe Zerodha api can be learnt in parallel. I am also from a non coding background.
    How about we get in touch to share knowledge and increase our learning speed?
  • batbark
    @Apoorv that would be awesome. reach out to me at [email protected]
    cheers!
  • VenkateshV
    VenkateshV edited February 2018
    Hi @batbark @Apoorv

    its quite interestting i happened to land on this thread.
    I'm too quite interested in learning a foolproof Algo system development using Python and im too faar away from the python coding part. However i am good in Excel and reasonably ok in VBA.

    Could you also include me in this learning journey. i would be glad to be the part of the team.
    i can be reached at [email protected].

    Pls drop me your details so that we can have a breif chat.

    Regards
    Venkatesh
Sign In or Register to comment.