A suitable language to build a bot for a complete beginner?

Post Reply
Speculator_3
Posts: 61
Joined: Wed Feb 20, 2019 10:01 pm

LinusP wrote:
Thu May 30, 2019 7:51 am
Speculator_3 wrote:
Thu May 30, 2019 12:25 am
spreadbetting, Shaun, pythonic et al: thanks for all your replies.

Shaun: you mentioned SQL / database connectivity. I would only need this if I wanted to test historical data, right?

Another thing. Untill now, I was under the impression that Betfair API readily provides the live scoreline info. Then I saw this thread: viewtopic.php?f=50&t=12490.

Am I correct in understanding that in fact, it is not possible to obtain a scoreline update from the API (in the first post in that thread it's mentioned "Without an actual score feed to trigger bets following a goal...") ? Is this lack of a score feed due to Betfair API itself, and as such would be universal, or is that thread only relevant for BetAngel software users?
If you go down the python route, I have added the inplayservice endpoint which is what the Betfair website uses for scores.

https://github.com/liampauling/betfair

https://github.com/liampauling/betfair/ ... service.py
Thanks Linus - will have a look.

I had a look at some commercial bots available for purchase - looks like they can't do apriori exactly what I want (maybe they can, but it's not straightforward to set them up like this).

So indeed, I will have to head down the route of trying to build this bot myself.
User avatar
firlandsfarm
Posts: 2686
Joined: Sat May 03, 2014 8:20 am

ShaunWhite wrote:
Wed May 15, 2019 2:55 pm
... it might take you 6+ months working on it full time and that's before you even start to work on finding edges …
johnsheppard wrote:
Thu May 16, 2019 4:11 am
Personally, I think 6 months is optimistic....well...yes the very basics....but realistically, a lot more to work is required to get to something that doesn't already exist out there...
Shaun/John, are you suggesting it would take one person at least 6 months, full time, to design, write and debug the code for a bot? I find that amazing. I'm not saying I disagree, I have no idea what would be involved but … wow!

I have used BA off and on over the years (currently on :) ) and have tried to compare it within a group of available bots. That is at least 6 months work and at the end you still don't know if you are going to get what you want. yes you can draw up a "does and does not" list but it's not until you use a bot and really get into it that you realise just how important those little seemingly insignificant do's are. Public bots evolve 'democratically'. They start in their originator's eye and then embark down the evolution road. The journey is largely influenced by the originator's developing needs but as the originator slowly exhausts their list of needs this gives way to what the users want tempered by commercial reality so it may never do what some users want.

To discover what you can do with a bot takes an extraordinary length of time. First you learn what it does (that's the comparatively easy part, you read a manual) but then you have to link those exclusive facilities together to find out what it can do. But at that stage you are not sufficiently familiar with what it does to immediately think "if I do A then B and lastly C it will do what I want" so you spend a lot of time thinking how to do it and you ask and beg for help on a forum and maybe, if you are lucky you work your way through it. To learn each bot is no different to learning each programming language. So for some it may be quicker to learn and build their own bot than compare off the shelf products and at the end of it not only do they (hopefully) have something that does exactly what they want (within the capabilities of the market) but maybe they have something they can sell. Of course another alternative is to find a guy beavering away in his bedroom to write the code but that has it's downsides.

BA is a very good product and I would put it in the top 2 of the ones I have looked at but I think the market is wanting a bot builder along the lines of the various web builders (e.g. Wix) and financial market builders (e.g. MetaTrader).
PeterLe
Posts: 3715
Joined: Wed Apr 15, 2009 3:19 pm

Re the Book..
3) Don't waste your time, I bought it and it's pretty much useless and outdated there's even a separate web page for it to cover all the errors/omissions in the book. There's a Betfair developers forum and snippets out there for most languages to get up and running.

[/quote]

Id agree with this; i hit some problems and the author wouldn't help (nor were the errors covered in his separate web pages) Its out of date now anyway

i already had my own bots up and running anyway which were built with C++
If you are new to coding, there are some great courses on UDEMY, for around £15 - £20
regards
Peter
User avatar
firlandsfarm
Posts: 2686
Joined: Sat May 03, 2014 8:20 am

PeterLe wrote:
Sun Jun 02, 2019 10:33 am
If you are new to coding, there are some great courses on UDEMY, for around £15 - £20
Thanks for mentioning that Peter, never come across that website before, it looks to be a huge depository of information
PeterLe
Posts: 3715
Joined: Wed Apr 15, 2009 3:19 pm

firlandsfarm wrote:
Sun Jun 02, 2019 12:04 pm
PeterLe wrote:
Sun Jun 02, 2019 10:33 am
If you are new to coding, there are some great courses on UDEMY, for around £15 - £20
Thanks for mentioning that Peter, never come across that website before, it looks to be a huge depository of information
No probs; glad it was of use
By the way; DONT pay the full price, there are loads of codes around for discounts on Udemy, you shouldn't pay more than say £20...
Thanks
Peter
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

PeterLe wrote:
Sun Jun 02, 2019 12:34 pm
firlandsfarm wrote:
Sun Jun 02, 2019 12:04 pm
PeterLe wrote:
Sun Jun 02, 2019 10:33 am
If you are new to coding, there are some great courses on UDEMY, for around £15 - £20
Thanks for mentioning that Peter, never come across that website before, it looks to be a huge depository of information
No probs; glad it was of use
By the way; DONT pay the full price, there are loads of codes around for discounts on Udemy, you shouldn't pay more than say £20...
Thanks
Peter
I'd give big thumbs up to Udemy aswell. Nice curated and quality checked. A little structured learning is so much more productive than just randomly picking up things. Do a couple of hours a day and you'll soon be doing some really quite useful stuff, I don't know why everyone doesn't do it.
User avatar
johnsheppard
Posts: 283
Joined: Mon Feb 04, 2019 6:00 am
Location: Cairns Australia

firlandsfarm wrote:
Sun Jun 02, 2019 8:05 am
Shaun/John, are you suggesting it would take one person at least 6 months, full time, to design, write and debug the code for a bot? I find that amazing. I'm not saying I disagree, I have no idea what would be involved but … wow!
Honestly, its really hard to say. That's what I guetimate....I can say with 99% certainty than anything software engineering related will 100% of the time get underestimated by inexperienced people. I can also say with probably the same degree of certainty that experienced people will underestimate also. There are just way too many unknowns... It's inevitable that you make mistakes....its really easy not to plan time for that...

I can tell you from my experience. It took me about 1 month full time to grasp 80% of the betfair API and get it up and working using C# APING code....just to place orders, and monitor markets....really really basic stuff...nothing that will make anyone any money... (I have not made any yet :))... to make that code into a robust, fault resistant useful bot....is a whole lot more.... But yeah, I dont claim to be an expert, just I have poked my finger at it and made quiet a few mistakes...that's all...maybe someone that doesnt make as many mistakes as me might be quicker :)
firlandsfarm wrote:
Sun Jun 02, 2019 8:05 am
BA is a very good product and I would put it in the top 2 of the ones I have looked at but I think the market is wanting a bot builder along the lines of the various web builders (e.g. Wix) and financial market builders (e.g. MetaTrader).
I've run out of time right now...but I will return to discuss this later...
pythonic
Posts: 65
Joined: Sun Jul 08, 2018 10:20 pm

johnsheppard wrote:
Mon Jun 03, 2019 5:33 am

Honestly, its really hard to say. That's what I guetimate....I can say with 99% certainty than anything software engineering related will 100% of the time get underestimated by inexperienced people. I can also say with probably the same degree of certainty that experienced people will underestimate also. There are just way too many unknowns... It's inevitable that you make mistakes....its really easy not to plan time for that...
That is a good point indeed.
When I started programming many years ago and later started to write my first real world applications I wasted a LOT of time reinventing the wheel and that is a typical beginners error.
They also tend to underestimate the time and efford it takes to get 'small and simple' things right and bug-free.
Simple strategies like 'bet on A at time t when conditions X and Y are met' would surely be much easier to implement with BA or other software than by building your own bot from scratch, especially without any programming experience.
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

firlandsfarm wrote:
Sun Jun 02, 2019 8:05 am
Shaun/John, are you suggesting it would take one person at least 6 months, full time, to design, write and debug the code for a bot? I find that amazing. I'm not saying I disagree, I have no idea what would be involved but … wow!
More if they have to learn how to program first. And then with zero experience the design will be flawed, and code would be as buggy as hell. The simple way would also result in an exe that's pretty much just doing the one job, to create something that's a good framwork for improving easily to accomodate other strategies would require foresight and experience. If it were easy then why do quality designers and coders, who also understand the specific industry earn 5 - 600+ a day? Plus 'you' might be hopeless at it and simply not have the right type of brain for being hyper analytical.

I'd put writing a decent little system on par with making a basic gearbox, all sounds easy, 1 input, 1 output and a bunch of cogs on shafts inbetween, doable. But you need to learn a CAD system, master your CNC tooling, learn to weld, learn to make castings, re-invent what's already been discovered....and then you realise you've missed out a 3rd gear and have to start again. And that's all with stuff you can see and touch, not a bunch of electromagnetism with hidden 'features' all explained via a correspondance course (forums & yt). If you're the type of guy who likes making stuff then it's worth a try, but if your problem solving skills extend to changing a light bulb, forget it.

Incidentally, what features does this 'bot' need that can't be done either with BA, or by building an interface to BA (via the excel DDE) and letting BA do all the messy stuff like order placement, position keeping, monitoring etc etc? My first foray into my own trading stuff (after 30yrs of being in IT anyway) used BA as the transaction processing engine, and that worked reasonably well until it's limitations meant taking it out of the loop at a later stage. That said I still use it to monitor what's going on. BA's openness (data out, trades in) is a fantastic asset, don't just discount it and throw the baby out with the bathwater.
User avatar
firlandsfarm
Posts: 2686
Joined: Sat May 03, 2014 8:20 am

ShaunWhite wrote:
Mon Jun 03, 2019 3:29 pm
Incidentally, what features does this 'bot' need that can't be done either with BA, or by building an interface to BA
I have no idea but it will be something similar to what some people said what will this Wordpress thing have that you can't do with Frontpage. I was simply referring to what the future may hold. As more and more become involved in 'botting' the products will get better and better until someone says … "Hey I got it, we need to build a framework structure like that Wordpress thing where people can drop modules in and link them together to make a custom bot."

It will have database modules where you subscribe to a module that allows you to do the research you want be it sports results or price history. It will have modules for each exchange and whatever each bookmaker has evolved into to facilitate placing bets across the whole market. It will have modules to analyse your betting/trading performance and help you identify where you could improve probably with machine learning functions. The biggest thing is, like Wordpress, it will be designed specifically for plug-in modules and like Wordpress 3rd parties could design and sell their own modules … tipsters would have plug-ins that administer their selections for you and some might get a module that automatically places the bets from the tipster plug-in to the Market Bet Placement module. it won't be tomorrow, it won't be next year, it may not be in the little that's left of my lifetime but it will be the future at some point. There is lots that will be done in the future, the industry is young but when it matures …..
User avatar
johnsheppard
Posts: 283
Joined: Mon Feb 04, 2019 6:00 am
Location: Cairns Australia

firlandsfarm wrote:
Mon Jun 03, 2019 9:40 pm
It will have database modules where you subscribe to a module that allows you to do the research you want be it sports results or price history. It will have modules for each exchange and whatever each bookmaker has evolved into to facilitate placing bets across the whole market. It will have modules to analyse your betting/trading performance and help you identify where you could improve probably with machine learning functions. The biggest thing is, like Wordpress, it will be designed specifically for plug-in modules and like Wordpress 3rd parties could design and sell their own modules … tipsters would have plug-ins that administer their selections for you and some might get a module that automatically places the bets from the tipster plug-in to the Market Bet Placement module. it won't be tomorrow, it won't be next year, it may not be in the little that's left of my lifetime but it will be the future at some point. There is lots that will be done in the future, the industry is young but when it matures …..
I dont think you can commoditise bots. Once an edge is public its no longer an edge, hence unprofitable, and unsellable.
You can commoditise bot making tools. But while their application is more profitable than selling. They won't be sold. I would presume there are hundreds of people out there with the workings done bespoke.
I'm not knowledgable about the workings of markets (I spent my time at computing science instead, dang it)...but I would probably suggest one would want to learn that stuff before diving in to develop. That task in itself is complex and long. But necessary if you dont want to end up in a world of pain in 3 or 5 years (unless you have the knack for cutting balls and chain)...

I'm still learning about betangel as I just use the trader version at the moment, but seems to me the pro version and guardian and excel etc does much. The less bespoke code you can write the better...because bespoke code is expensive.... you can ask the market about that and it will tell you...(market presumably knows more than any individual)
User avatar
firlandsfarm
Posts: 2686
Joined: Sat May 03, 2014 8:20 am

johnsheppard wrote:
Mon Jun 10, 2019 12:03 am
I dont think you can commoditise bots. Once an edge is public its no longer an edge, hence unprofitable, and unsellable.
Fully agree John, I wasn't referring to individual bots but bot creators such as BA and others, tipsters who currently give out their tips via websites and emails, bookies so you didn't have to keep changing websites, tipster reviewers who are all over the Internet, odds monitors, data providers, blogs and columnists … the list of support facilities for punters goes on and on but the punter has to find them. Look at the way the financial markets have developed frameworks such as Metatrader that facilitates accessing accounts at multiple marketmakers, enables data from different sources on different securities etc. … I'm thinking along those lines. At the moment we search the Internet for data and other services. I was suggesting that in the future all of this fragmentation will be centralised inside a framework with common connectivity standards and protocols. The principle of "the Internet of things" will not just apply to fridges and car convoys. It will bring all fragmented sectors and suppliers together by way of centralised frameworks similar to the way Wordpress did for websites, eBay and Amazon have done for small online outlets and Oddschecker is trying to do for bookies and exchanges. Before Wordpress website designers had to go looking for add-ins but since Wordpress the Add-ins know the designers will be at Wordpress so they go thereto find them.
User avatar
johnsheppard
Posts: 283
Joined: Mon Feb 04, 2019 6:00 am
Location: Cairns Australia

firlandsfarm wrote:
Mon Jun 10, 2019 9:24 am
johnsheppard wrote:
Mon Jun 10, 2019 12:03 am
I dont think you can commoditise bots. Once an edge is public its no longer an edge, hence unprofitable, and unsellable.
Fully agree John, I wasn't referring to individual bots but bot creators such as BA and others, tipsters who currently give out their tips via websites and emails, bookies so you didn't have to keep changing websites, tipster reviewers who are all over the Internet, odds monitors, data providers, blogs and columnists … the list of support facilities for punters goes on and on but the punter has to find them. Look at the way the financial markets have developed frameworks such as Metatrader that facilitates accessing accounts at multiple marketmakers, enables data from different sources on different securities etc. … I'm thinking along those lines. At the moment we search the Internet for data and other services. I was suggesting that in the future all of this fragmentation will be centralised inside a framework with common connectivity standards and protocols. The principle of "the Internet of things" will not just apply to fridges and car convoys. It will bring all fragmented sectors and suppliers together by way of centralised frameworks similar to the way Wordpress did for websites, eBay and Amazon have done for small online outlets and Oddschecker is trying to do for bookies and exchanges. Before Wordpress website designers had to go looking for add-ins but since Wordpress the Add-ins know the designers will be at Wordpress so they go thereto find them.
ahh ic... Well, I can say I am not experienced enough to comment on the likelyhood of that occuring.

My experience with the CMS world (which is not extensive) is that in complex instances its cheaper to develop bespoke (a successful trading bot app that can scale is complex). Wordpress' market is low budget and no room for customization. Jam it all together bobs ur uncle sort of thing. Not really sure that works for botting...

I would be inclined to think that you would suffer the same problem with commoditising tipsters. Imagine for example there was a succesfully central tipping site with api where every tipster on earth put their tips. The edge would be quickly eroded. I also suspect successful (or perhaps professional) tipsters have a vested interest in not using centralised frameworks but I havent thought that through fully :)

As an aside, In the begining when I started out with betfair's api, I was also experimenting with scraping tipster sites etc.....In terms of scraping websites it's not all that difficult to do it's just that websites change...and crossmatching runners etc is not always easy...
User avatar
firlandsfarm
Posts: 2686
Joined: Sat May 03, 2014 8:20 am

johnsheppard wrote:
Tue Jun 11, 2019 5:13 am
ahh ic... Well, I can say I am not experienced enough to comment on the likelyhood of that occuring.
Neither can I John I was just speculating on what could be.
johnsheppard wrote:
Tue Jun 11, 2019 5:13 am
My experience with the CMS world (which is not extensive) is that in complex instances its cheaper to develop bespoke (a successful trading bot app that can scale is complex). Wordpress' market is low budget and no room for customization. Jam it all together bobs ur uncle sort of thing. Not really sure that works for botting...

I would be inclined to think that you would suffer the same problem with commoditising tipsters. Imagine for example there was a succesfully central tipping site with api where every tipster on earth put their tips. The edge would be quickly eroded. I also suspect successful (or perhaps professional) tipsters have a vested interest in not using centralised frameworks but I havent thought that through fully :)

As an aside, In the begining when I started out with betfair's api, I was also experimenting with scraping tipster sites etc.....In terms of scraping websites it's not all that difficult to do it's just that websites change...and crossmatching runners etc is not always easy...
John, I think you are still missing my point. I'm certainly not referring to the market for people like you who can build their own bot. I am looking at how the market for serious, semi computer savvy people made up of the thousands like myself and many others on the BA forum and other websites might mature to in say 10 years or so as computing cost come down and users become more accustomed to facilities and services becoming more integrated.

I'm not looking at bespoke builders. I'm not suggesting tipsters would make tips available to all and so dilute their edge. I'm not looking at traders though if they are currently trading on BA I see no reason why they shouldn't continue on what might be a BA type module. I'm looking at the current 'popular' market … I'm just saying that market could bring all it's sub-markets together within a common framework so that the various modules can interlink and cross communicate. A common marketplace, not common products.

I wasn't referring to Wordpress for quality of product, just for representation of how something might work with plug-ins. (And a debate on the Wordpress market is a debate for another day. :) )
User avatar
ShaunWhite
Posts: 9731
Joined: Sat Sep 03, 2016 3:42 am

The big cost is making it robust enough for public use, look how much people struggle with Guardian and that's pretty straight forward. And let's say it earns you 10k a month in subs, wouldn't all that time and effort be better spent finding another 10k/mo strategy rather than dealing with customers 247?

And also the issue that any system that was specifically designed to retain data would breach just about everyone's ts&cs. Millisecond horse data from BF is 200quid a month, tread on those toes at your peril.

Even BA just breaks even by all accounts, building an even more specialised complex system from the ground up (2 man yrs+) and supporting it, incl updates for all these different apis would need a huge customer base and subs before you saw a dollar in profit. Trading as been around for 20yrs, I don't know where that boom in numbers would come from. "I'm out" :)
Post Reply

Return to “Betfair Exchange API”