I've had the misfortune to need to use Python recently after decades of VBs then C#/SQL. I can see why it's popular for new starters but I just find it's a bit slow executing, although the maths libraries blaze through. It also seems to be unstable with it relying on so many external libs. Nothing that can't be fixed if that's your thing but it's not for me and VS doesn't support it as well as it supports C# either. I've kept all my trading comms and db access in C# and use a named pipe to python just for the model inference, best of both worlds imo.
Libre Office
-
ShaunWhite
- Posts: 10646
- Joined: Sat Sep 03, 2016 3:42 am
Hi Shaun,
Yes those external libraries!!!! .
You seem to have a good setup there and that's a good idea to pipe through to Python .
One good thing I used Python for was webscraping .It has a few quite stable dedicated libraries for scraping that were quite useful . I have never felt happy with it's database stuff though.
Have you tried using AI to write some code ? .The initial idea that all programmers would have to join the dole queue seemed to have diminished a bit .
A while back I fed a load of greyhound stuff into Perplexity AI ,basically a meetings full of cards for the dogs at Monmore . The last 5 formlines were included .It managed 5 wins out if the 12 races (two odds on) but it also found a 5/1 and 7/1 .Next day I tried the same at Romford..12 races two wins 3/1 and 7/2..loads of 2nds though .
Kinders
Tico
Yes those external libraries!!!! .
You seem to have a good setup there and that's a good idea to pipe through to Python .
One good thing I used Python for was webscraping .It has a few quite stable dedicated libraries for scraping that were quite useful . I have never felt happy with it's database stuff though.
Have you tried using AI to write some code ? .The initial idea that all programmers would have to join the dole queue seemed to have diminished a bit .
A while back I fed a load of greyhound stuff into Perplexity AI ,basically a meetings full of cards for the dogs at Monmore . The last 5 formlines were included .It managed 5 wins out if the 12 races (two odds on) but it also found a 5/1 and 7/1 .Next day I tried the same at Romford..12 races two wins 3/1 and 7/2..loads of 2nds though .
Kinders
Tico
-
ShaunWhite
- Posts: 10646
- Joined: Sat Sep 03, 2016 3:42 am
With mixed success, esp with being new to python it's been useful for generating blocks of code, and the interop but you couldn't write a system with it. The 'text book' approach it has, the flimsy grip on the bigger picture and sudden amnesia about what it's supposed to be doing all make it a long frustrating job. I find it best to spend the time doing 'teach me how to....' rather than trying to teach it what I want. That said I wouldn't be half as far along without it, some issues it's identified like clashing library versions would have taken me weeks to find, if ever. I know I lean on it too much still but at 60+ my personal RAM is full so I need to delegate
.I've never done any scraping though I probably should, but linking back to BF market and selection IDs and having so much data without it already, means it's never been a priority. Same story integrating TPD data, but that's been shelved too due to the work this year learning/adding the real-time model functionality. Then the real work starts on trying to turn the sweat into cash.