Python, should I invest the time?

[craignmoss]

Hi guys,

I've almost completed a data analytics and statistics course which is going to help me look at historical sports data. The next obvious step for me to analyse data efficiently is to learn Python but obviously it's going to take some time, I have zero knowledge.

Has anyone gone through this process before and if so, is it worth it? I don't mind investing the time but I need to see a noticeable ability to save time in the long run and a heightened ability to analyse data at the end of it.

Thanks,

Craig

[LinusP]

100%, do it, I did a course and then jumped straight into using the betfair API and haven’t looked back.

[craignmoss]

Thanks Linus. Would you recommend any course in particular? I’ve seen a free one through code academy but I don’t know if it covers everything I need.

[Euler]

It may be worth learning to analyse data, but not so sure about coding your own app. The time and expense isn't worth it for most people I would think. The hard work of automating it has already been done in Bet Angel and all the cost and hassle of creating a platform.

People seem to spend a lot of time looking at historical data, but data mining is statistical deja vu IMO. It's best to use data to model.

[LinusP]

Thanks Linus. Would you recommend any course in particular? I’ve seen a free one through code academy but I don’t know if it covers everything I need.

I used code academy but that was a long time ago, @peterle should jump in soon as he has learnt python recently

[craignmoss]

It may be worth learning to analyse data, but not so sure about coding your own app. The time and expense isn't worth it for most people I would think. The hard work of automating it has already been done in Bet Angel and all the cost and hassle of creating a platform.

People seem to spend a lot of time looking at historical data, but data mining is statistical deja vu IMO. It's best to use data to model.

Hey Euler,

This isn't to build my own app it's to use coding to help analyse data better. You can pull data from online portals (such as Livescore) and then work with it in Excel.

[Euler]

What sort of data are you looking for?

[craignmoss]

I’m not even sure myself yet but my interest is in football. My thoughts are that any data people can see on a website is too public, I don’t see how I could get an edge there. The harder data is to extract, the more likely I could find patterns and profit from them. It could be timings of yellow cards in the Serbian top flight for example!

[alexmr2]

I’m not even sure myself yet but my interest is in football. My thoughts are that any data people can see on a website is too public, I don’t see how I could get an edge there. The harder data is to extract, the more likely I could find patterns and profit from them. It could be timings of yellow cards in the Serbian top flight for example!

You could focus on finding an edge with data in Excel and test it manually, then if it works consider learning how to automate it rather than the other way around

[craignmoss]

That is exactly my plan! It’s getting ‘not easy to gather’ data into excel that Python will be able to help with.

[PapaShango]

I was in the same boat too. I knew I wanted data, but didn’t know what data I wanted or what I would use it for.

However, after learning Python during lockdown I now have no end of data which I now know exactly what to do with.

Some data is harder to get, and isn’t, or shouldn’t be, publicly available. However there are lots of people who you can ask privately who can help get you this data.

It’s not easy , but well worth investing the time.

[gazuty]

I've said on other occassions (viewtopic.php?p=265873#p265873) you need an idea, a plan.

You need to say to yourself, if these conditions are satisfied and I do x, then the result I can expect is y. That is - if I take a particular action there is an expected value outcome of that action, and if that expected value outcome is positive after commissions - bingo was his namo. Try to discover things that can be repeated many times. Things that can only be repeated occassionaly can be ruinous even if expected value positive because of draw downs.

Each person has their own poison, I don't like to back test on data, I like to take ideas I think will be sucessful and forward test by using real money for real outcomes. For that purpose, I just set aside money I'm willing to lose to R&D (while expecting profit).

[PeterLe]

I'd say do it. Learning anything new is good for growth - growth mindset
There is another thread on here on Python where I list the Udemy course i did. (Cost about £12- £15?)
I must admit though; I was actually expecting it to be easier that it was, so you have to be prepared to invest your time.
The udemy course was easy to follow, but then putting that into practice was harder as is anything when you are new.
If I were to sit down at a new PC now, I could probably have flumine installed and strategy running in 20 mins now I know what im doing.

There is a big plus. There is always someone on the slack group who will answer any questions you have. Truth be known; without their help i would never have achieved it.

I now have a few Flumine bot's up and running each day 3 accounts, (making money). Im able to record the data and use for backtesting too.
Its worth noting that I also use betangel. I have a few accounts on this too. I particularly like betangel for development purposes as its simple to test new ideas. Still a great product in my opinion and they make money every day.
To cap it off; I have my own stuff written in C# for the fast in play stuff (TPD etc) (Python is much easier to learn that C# by the way ).
So in summary, have a go, nothing to lose other that time. One thing I would say is that I feel I have a much better understanding of the API, matching rates etc than I did before.
One interesting thing; Ive found that you can have the same strategy running in betangel; Flumine and C#, same stakes, criteria etc and they all give different results each race/day

regards
Peter
(PS Its highly likely that you will never reach a level of understanding of python as some on the slack group. However, even if you were the very best at Python it doesn't guarantee that you will be successful on Belfair. For me, to be successful, you have to think different in you application of your strategy. Some may disagree but I think that if you can't make money with betangel, then its probably likely you wont using other languages.

[decomez6]

The harder data is to extract, the more likely I could find patterns and profit from them.

Have you truly explored all the options that excel got to offer?
Eg Power BI, AI …etc…. If what you are looking for is patterns and correlation excel got some excellent capabilities.
On the other hand I would be very careful with things like fill rates and human behaviour ,hard to predict and yet any data you collect will inherently contain this information.

In case you choose to go the python or R way , be prepared to go all the way or Not at all. Atleast to a level where you can debug your own code without falling into a situation where your own bug result to false positives , going undetected until a major loss.
You will need to have enough knowledge to build a reliable platform that you can trust to handle your scaled up bets in a live market .

I personally think it’s not worth starting something you can’t finish .I have wasted a lot of time on projects that are half finished and I think it’s a steep learning curve.
That being said , if you have the time and enthusiasm….go for it !!
Coding should in theory give you an executional edge ,allows your robust profitable trading plan develope an artificial intelligence (AI) !

all the best .

[michaelg2708]

what extra in formation is available using python through the api that you cant get by sending data to excel?