Forums  > Trading  > Research to production pipeline  
     
Page 1 of 1
Display using:  

Jefferspin


Total Posts: 11
Joined: Jun 2020
 
Posted: 2020-12-17 16:07
Is this a real thing, somebody actually seen it or it's just an urban legend, a Santa Claus story for adults?
Are there places where people research and design trading strategies and some of those strategies are deployed in production and run live and make money?

If it's real, what's the pipeline between design stage and live stage?

Here's an accurate picture of my mental image of this pipeline:



prikolno


Total Posts: 78
Joined: Jul 2018
 
Posted: 2020-12-17 16:37
I'm having problem parsing this post. I can't tell anymore if this is GPT-3 or a human with all these spam threads.

nikol


Total Posts: 1276
Joined: Jun 2005
 
Posted: 2020-12-17 17:04
prikolno:
Same here. After all history of Jefferspin's 9 previous posts the question about "Phase-2" leaves me puzzled.

... What is a man
If his chief good and market of his time
Be but to sleep and feed? (c)

Jefferspin


Total Posts: 11
Joined: Jun 2020
 
Posted: 2020-12-17 17:49
I intend to start releasing a software that eventually will provide the full pipeline.

1) Import and display of historical market data. You need data to base your research on, and the data I have access to (relatively cheaply) has a lot of errors in it. So first feature of the software I developed is to curate and validate the data. This I plan to release not just for free but fully open source.

2) Define trading strategies and testing them against historical market data. It takes a lot of effort and failed attempts to get a strategy that's consistently profitable and not just by random accident. Eventually though, any such strategy that I produced myself (and will support in the software) is fully-algorithmic and can be run on historical data to ensure it's consistent with reality. There are lots of gurus selling trading strategies for retail based solely on "take my word for granted", oftentimes deliberately adding some "cannot be fully described in software coze trader intuition etc". If you could test them you would immediately see they aren't working.

3) Run live tests on a virtual account connected to Interactive Brokers (since it's the only decent API available at retail level). Next level after validating a strategy on backtests is ensuring it gives similar numbers in a live setup. Also check as close to production as possible that it correctly handles market events: filling of partial orders, disconnect, not triggering accidentally in a suicidal loop, market going crazy etc.

4) By this time if I'm gonna get an investor sponsoring this stage, offer the possibility to run strategies that passed #2 and #3 as managed accounts on either Interactive Brokers or another broker (would have to adapt the API).

So refining the question: do steps 1-4 reflect to some degree of accuracy the "?" sign in the gnome presentation picture?

nikol


Total Posts: 1276
Joined: Jun 2005
 
Posted: 2020-12-17 20:28
You named it all, but in weird order

... What is a man
If his chief good and market of his time
Be but to sleep and feed? (c)
Previous Thread :: Next Thread 
Page 1 of 1