Finally got around to the remake

[LinuxGam]

Tipster Competitions were huge once... and once had a load of traffic too... But with the way things have gone especially with AI... if you can't beat them join them. Also I LOVE AI.

It doesnt have everything it will have in the next few months, but id rarther get a MVP out there cos of Google etc... takes time.

Let me know your thoughts.

https://tipstercompetition.com

[allfreechips]

Super clean and fun looking.. not a lot of content but gets to the point nicely.

[LinuxGam]

Agreed there is no static content and as we all know that's terrible for SEO... However the site already had 1000s of ppl a day from old site, so we decided to get the core out and build on it. I hope to add some really useful content out over the next few days/weeks.

[newcustomeroffer]

I like it, looks modern - will be interesting to see how you develop it from here, best of luck with it!

[universal4]

Not a fan of mostly black sites but that is a personal preference with me and many like the darker sites.

Not sure how dynamic the top tips are, but if that doesn't change much, I would consider getting at least one of them to show any other bookie then bet365.

Good presentation for a tipster site though.

Rick
Universal4

[TheGooner]

Hey - I liked the look of it. Smooth running - nicely laid out - cutesy graphics.

----------------------

But - and this is the kicker - I doubt that I'd bookmark it or ever return to check it again.
The "tips" lack credibility - lack details, reasoning - or any real reason to back them.
The real chance and confidence values seem almost random numbers - but I assume that they're calculated on last 5 form or something trite.

800 "tips" so far in September - it's just too much volume - too little difference.
I assume that they're all simple algo picks? Not real people?

As an example :
I see there are 5 tips on the Bundesliga - all the same game - 3 overs (at various random win chance), 1 away win, 1 BTTS.
I expect at least one tipster will get a winning angle there and claim that they "knew it".

There is a similar thing in the Premier League - 5 tips - all the same game.
3 x under goals, and 2 x BTTS NO.

Quick analysis shows both these team midtable - both averaging 2.5 goals per game - any result is likely.
A human would say that this is a no bet game - but the bots pump out five "meh" tips.

What is the point?

-------------------

However - that's just me old, grumpy, an experienced punter that sees knows candy floss when he sees it.

Maybe the tweenie Gen Alpha with a cellphone and no money will love it.
It's certainly well constructed and has the "easy to consume" layout required to work on mobile.

10-20 years ago I saw lots of these sort of websites - usually with a forum - asian punters /readers loved it if you offered a monthly prize.
It gave bragging rights and a bit of banter - albeit very limited range due to the reader.
I understand why they died.

-----

But it is a very tidy implementation for a data-lite idea that could just run with very little input.
So well done on the work - but what would people keep visiting? Just why?

[kid]

At least have some decency and inform your visitors that all figures shown are random numbers, generated by a model whose only objective is to predict the next token, that will mathematically lead to losses.

[betprotips]

It looks really nice, especially your prediction cards - I would probably focus more on quality and less on quantity when it comes to all those match previews and predictions. Great job!

[LinuxGam]

@kid @TheGonner

Over 70k words of stats go into every preview before its written and each preview before it writes the end preview generates about 12 to 16 pages of real maths and calculations. Sure, ist never gonna be perfect nor are humans. What it does lack is upto date team news like managers and players out, which is a goal for the future. As regards your comment @kid explains your name

[LinuxGam]

On the tip rate, most of them actually do no bet, ranging from 30% bets to about 80%.. i def should lower that i think, but its kinda a slightly fun platform where they have personalities. If you check and read them there is a lot of NO BETS but I leave them there for SEO and also the facts and data can help people who might know better.

[LinuxGam]

This makes me laugh actually, obv im giving tipsters diff objectives, so they wont all bet the same... however.. this is like the most compatible bet ever... "There is a similar thing in the Premier League - 5 tips - all the same game.
3 x under goals, and 2 x BTTS NO."

IF you thought was an under game, you'd be looking at those two options.. That said.. being as I ask some to go for long shots, some favourites etc... they of course will differ...

Thats why we dont say follow this tipster and win as with the bookies markup that's really hard! So is more about creating a platform you can follow and make your own decisions. Ive never been one to post bullshit like 100% win cant lose **** i see ppl do

[Sherlock]

@kid explains your name

I do not think so. But AI tips are not worse or better than human tips. It is the same random crap. AI is often giving me random data even when it really should not do it, so AI is ideal to produce a white noise.

What puzzles me is how bet365 can approve it and ban other sites (no offense to you, it is nice you pushed it through).

I do not like the website, looks like cheap AI design as well (for example the texts are not aligned, which is often a puzzle with AI vibe coding). On the right side pops an uncloseable sidebar banner (at least with my screen resolution).

[LinuxGam]

Please let me know as ive been a coder and employed in awarding winning companies for over 30 years. as a developer and arcihtect. Im sure there are huge amounts of bug tho, so if can post here or PM would help, as often browser cache working ocde. TY

[LinuxGam]

Im not trying to be a dick as likely there are screen and mobile and code issues 100%.. its very new. if you could let me know would help me.

[LinuxGam]

Vibe coder .....

[LinuxGam]

I honestly appreciate your opinion and also you have been very direct here for 10+ years I've seen and always respected that. Could you send me your webiste(s) I'd love to take a look

[iso2009]

Site looks nice for an AI powered website. But it seems to get no traffic from search engines.

Do you actually check the yield or ROI for the AI placed bets? I see on the results page a lot of won/lost, but maybe you can group them by market, bot, league with totals, ROI etc.

I have the same data as you, I always postponed to feed the AI with it as see the results. I've developed for tennis my own system, not perfect but gets it right in most of the very balanced games. Still, far from perfect but it doesn't predict the next token.

What keeps me from using AI for predictions is a single fact: LLMs are dumb.

"Over 70k words of stats go into every preview before its written and each preview before it writes the end preview generates about 12 to 16 pages of real maths and calculations." - that's a lot of tokens. Are you sure it's worth it? I mean, I assume you feed the data to a LLM.

Your AI generated previews won't have much success in Google, but if they do that would be awesome.

I am curious, you said there's traffic on the site, do you get any conversions with bet365?

Edit: if you're a coder, maybe you can employ machine learning (no great success, but it's a starting point), your own system of predicting an outcome, and let the AI do the reasoning in natural language. That would be a good use of a LLM.

[LinuxGam]

Site looks nice for an AI powered website. But it seems to get no traffic from search engines.

Do you actually check the yield or ROI for the AI placed bets? I see on the results page a lot of won/lost, but maybe you can group them by market, bot, league with totals, ROI etc.

I have the same data as you, I always postponed to feed the AI with it as see the results. I've developed for tennis my own system, not perfect but gets it right in most of the very balanced games. Still, far from perfect but it doesn't predict the next token.

What keeps me from using AI for predictions is a single fact: LLMs are dumb.

"Over 70k words of stats go into every preview before its written and each preview before it writes the end preview generates about 12 to 16 pages of real maths and calculations." - that's a lot of tokens. Are you sure it's worth it? I mean, I assume you feed the data to a LLM.

Your AI generated previews won't have much success in Google, but if they do that would be awesome.

I am curious, you said there's traffic on the site, do you get any conversions with bet365?

Edit: if you're a coder, maybe you can employ machine learning (no great success, but it's a starting point), your own system of predicting an outcome, and let the AI do the reasoning in natural language. That would be a good use of a LLM.

Thanks for the intelligent and well-thought-out reply. I'll try to answer some of them as this is a very interesting discussion.

"Do you actually check the yield or ROI for the AI placed bets? I see on the results page a lot of won/lost, but maybe you can group them by market, bot, league with totals, ROI etc."

Yes there is a league table and full tracking, the top tipster last month made 17.5% ROI over 150 bets++ It only ran I think for last two weeks though.

"I have the same data as you, I always postponed to feed the AI with it as see the results. I've developed for tennis my own system, not perfect but gets it right in most of the very balanced games. Still, far from perfect but it doesn't predict the next token."

This is not changing by the year or decade but literally month by month... I used to use Llama 3 70b a year back and thouhg it was impressive whats happening now is far far far more impressive and impoves monthly.

" if you're a coder, maybe you can employ machine learning (no great success, but it's a starting point), your own system of predicting an outcome, and let the AI do the reasoning in natural language. That would be a good use of a LLM"

5 years ago before LLMs were even really a thing i trained many model;s from scratch in python, its so so much they can't do it... The issue is the amount of data you can get from football versus the results rarely correlate enough... So the training hits aritical peaks or troughs... The end result was better than pure chancew though, but you still have to beat the bookkies 4.5 to 10% overround...


"I am curious, you said there's traffic on the site, do you get any conversions with bet365?" Aslo 5++++ years ago the site when tipster competitons were a thing was quite busy and was many FTD a day... Then i kinda left it in limbo.... So I guess we will see.

As others have rightly pointedf out Google hates AI gneerated content... Its partially because its easily abused, but its mostly because it threatens there business model and yet they use AI generated answers themselvews, mostly made from other peoples work...

Like it or not, unless you work in a physical line of work and that includes professionals devs like me... our jobs are basically ****** from a corp / cost standpoint.. I tyhink graphic desingers and content sellers were the first to be hit, but honestly even hi end devs are not really needed... You def need archetects and managers... so i guess im ok for a bit...

"Your AI generated previews won't have much success in Google, but if they do that would be awesome." Its been three days... the last time i changed my sites "template" the word press guys will get the joke I guess... Google kicked me out for 3 months.. let alone indexed it.

[LinuxGam]

"Quick analysis shows both these team midtable - both averaging 2.5 goals per game - any result is likely.
A human would say that this is a no bet game - but the bots pump out five "meh" tips."

I can't see the exact tip so you may be correct, however to clarify, tips are not based on how likely they are to win, they are based on how likely they are to win versus the odds.

So there will be some really "dumb" look tips if you the average kind of bettor... To be honest if i can flip a coin 1mill times and get 52% chance, im ok with that...

If you back every sure bet in football you would not like to see your bank at the end...

The results are actually artificially low as I dont take the best odds or exchange prices... I have my own reasons for that for now.... But it is likely going to be a "feature" where it looks at over 50 bookies odds...

Did you see Bet365 updates to their terms and conditions... Gonna be a lot of sad people in 6 to 12 months.

[LinuxGam]

I do not think so. But AI tips are not worse or better than human tips. It is the same random crap. AI is often giving me random data even when it really should not do it, so AI is ideal to produce a white noise.

What puzzles me is how bet365 can approve it and ban other sites (no offense to you, it is nice you pushed it through).

I do not like the website, looks like cheap AI design as well (for example the texts are not aligned, which is often a puzzle with AI vibe coding). On the right side pops an uncloseable sidebar banner (at least with my screen resolution).

Could you send me the links, im trying to find them and can't, but looking at your own site, you def do miss obvious stuff, hazard of the job. I also tend to test mobile view in Google Chrome, rather than real mobiles, so there is def a large issue it may not look right on all mobiles.