[Journey] Predicting Sports Results with Artificial Intelligence

[Boriss]

What does U2.5 and O2.5 - mean?

 

[Deleted member 719276]

What does U2.5 and O2.5 - mean?

u2,5 You’ll win if max two goals are scored and lose if at least three goals are scored.
o2,5 You’ll win if at least three goals are scored and lose if max two goals.

Source: http://www.winnergambling.com/sports-betting-bingo/sports-betting-guide/

 

[Boriss]

Awesome @Neon - Thank you.

 

[mnunes532]

Vitoria Setubal - Maritimo U2.5 @1.44

2 portuguese teams hehe good luck

 

[midnightlee]

Thanks for all the input again, I'll answer later guys, just dropping my predictions for now

I only cared about games for today yet because I don't have any money left right now and do have to wait for the bets to finish

Silkeborg - Sonderjyske U2.5 @1.7
Pescara - Avellino O2.5 @1.47
Khabarovsk - Ufa U2.5 @1.3
Lokeren - Charleroi U2.5 @1.85
Sligo Rovers - Derry City U2.5 @1.85
Sandecja Nowy Sacz - Cracovia Krakau U2.5 @1.8
Vitoria Setubal - Maritimo U2.5 @1.44
Kerkyra - Panetolikos U2.5 @1.4
Coventry - Colchester U2.5 @1.95
Coventry - Colchester (Colchester U1.5) @1.28
Blackpool - Wigan (Wigan U1.5) @1.8
Rennes - Lille U2.5 @1.61
Levante - Getafe U2.5 @1.57
Cesena - Foggia O2.5 @1.85
Sheffield United - Reading (Reading U1.5) @1.33
Fc Zurich - Grasshoppers Zurich (Grasshoppers U1.5) @1.36
Huddersfield - Manchester United (Huddersfield U1.5) @1.12

In total I spent 19,29€ so I'll update tomorrow about my return

64.71% correct. Keep going to improve the algorhitm. I lose only Sligo Rovers - Derry City

 

[issorc]

u2,5 You’ll win if max two goals are scored and lose if at least three goals are scored.
o2,5 You’ll win if at least three goals are scored and lose if max two goals.

Source: http://www.winnergambling.com/sports-betting-bingo/sports-betting-guide/

Thanks for explaining

Vitoria Setubal - Maritimo U2.5 @1.44

2 portuguese teams hehe good luck

Thanks, unluckily it didn't work out

64.71% correct. Keep going to improve the algorhitm. I lose only Sligo Rovers - Derry City

Wouldn't they had scored in the 90' minute I'd had finished the day in profit, now I'm 1,88€ down

Todays bets:
Asteras Tripolis - Panionios (Asteras U1.5) @1.5
ZSKA - Zenit St. Petersburg U2.5 @1.61
Olympiakos - PAOK U2.5 @1.61
Nurnberg - Dresden O2.5 @1.61
Austria Vienna - Rapid Vienna BTTS @1.61
Spal - Sassuolo BTTS @1.75
Urawa Red Diamonds - Gamba Osaka BTTS @1.53
Troyes - Lyon NoBTTS @2.05
Almeria - Reus Deportiu NoBTTS @1.53
Gimanstic - Alcorcon NoBTTS @1.61
Benevento - Fiorentina NoBTTS @2.0
CD Aves - Benfica NoBTTS @1.66
Seattle Sounders - Colorado Rapids NoBTTS @1.75
Real Salt Lake - Sporting Kansas City NoBTTS @2.05

BTTS = Both teams will score at least one goal
NoBTTS = At least one team will score no goal

 

[Elon Snup]

Great. A lot of good inputs from a lot of people. I like where this is going.

I say it's time to start making an external Skype group of everyone involved in AI & want to help this topic go further.

So if you're developer, web designer, if you have ideas, anything, let's make this happen. The goal is not to post prediction on this group but rather have everyone involved in working on this in order to post more prediction here in BHW.

PM me and I'll set this up

 

[seyer]

This can be a very dangerous thread... careful with your bets, guys. Machine learning can only recommend but never predict with 100% accuracy, ever. Past data can only predict so much and can't completely disregard any random events that may happen in the future.

Nonetheless, interesting to see how you'll develop your model OP. Would be interesting to see what sort of systems thinking model/archetypes you've drawn up to create your algorithm.

 

[theRevolt]

You should work out along with your predictions a confidence score and see how that impacts the results when only betting above a certain confidence level. While the general subject is very interesting I am very skeptical that this can be profitable an scale able over a year/years but it's definitely very interesting to follow.

 

[Universe0]

Thanks for explaining

Thanks, unluckily it didn't work out

Wouldn't they had scored in the 90' minute I'd had finished the day in profit, now I'm 1,88€ down

Todays bets:
Asteras Tripolis - Panionios (Asteras U1.5) @1.5
ZSKA - Zenit St. Petersburg U2.5 @1.61
Olympiakos - PAOK U2.5 @1.61
Nurnberg - Dresden O2.5 @1.61
Austria Vienna - Rapid Vienna BTTS @1.61
Spal - Sassuolo BTTS @1.75
Urawa Red Diamonds - Gamba Osaka BTTS @1.53
Troyes - Lyon NoBTTS @2.05
Almeria - Reus Deportiu NoBTTS @1.53
Gimanstic - Alcorcon NoBTTS @1.61
Benevento - Fiorentina NoBTTS @2.0
CD Aves - Benfica NoBTTS @1.66
Seattle Sounders - Colorado Rapids NoBTTS @1.75
Real Salt Lake - Sporting Kansas City NoBTTS @2.05

BTTS = Both teams will score at least one goal
NoBTTS = At least one team will score no goal

Followed your predictions, and with 14€ bets i had 17.23€ returns. So about 23% ROI.

 

[midnightlee]

Followed your predictions, and with 14€ bets i had 17.23€ returns. So about 23% ROI.

Do you bet 1 euro per match all single bets?

Saturday i tryed this multiple bet:

Pescara - Avellino O2.5 @1.47
Lokeren - Charleroi U2.5 @1.85
Sligo Rovers - Derry City U2.5 @1.85 (lose)
Rennes - Lille U2.5 @1.61
Cesena - Foggia O2.5 @1.85
Udinese - Juventus O2.5 @2 (personal prediction)

5€ --> 132€ return

Too much matches. I want to figure out what's the best number of matches for a good return, considering the OP has a 70% of average correctness

 

[Universe0]

Do you bet 1 euro per match all single bets?

Saturday i tryed this multiple bet:

Pescara - Avellino O2.5 @1.47
Lokeren - Charleroi U2.5 @1.85
Sligo Rovers - Derry City U2.5 @1.85 (lose)
Rennes - Lille U2.5 @1.61
Cesena - Foggia O2.5 @1.85
Udinese - Juventus O2.5 @2 (personal prediction)

5€ --> 132€ return

Too much matches. I want to figure out what's the best number of matches for a good return, considering the OP has a 70% of average correctness

Yes i only bet one match, 1€ each. I never place system bets.

 

[midnightlee]

any bet for this 2 days? (Serie A will play)

 

[issorc]

any bet for this 2 days? (Serie A will play)

Perugia - Cesena O2.5 @1.85
KV Oostende - Charleroi (Oostende U1.5) @1.61
Cittadella - Venecia O2.5 @2
Ascolia - Spezia NoBTTS @1.75
Carpi - Palermo NoBTTS @1.61
Gent - Eupen NoBTTS @2.1

For today

 

[issorc]

I do have a few minutes right now to answer to all posts I've missed in the last few days I'm currently really busy at university and need to take care of other projects so that at least they continue running which slows down the development of this journey sadly.

sounds great @issorc! Keep up the good work!

@issorc keep up! i'm very interested

Sounds great... Keep Up, mate!

Thanks guys

Hi issorc.

Fortunately/unfortunately I have plenty of experience with gambling and trying to figure out systems.

Regarding getting banned you can always use Betfair Exchange and you will never be banned.

I bought a soccer stats crunching piece of software a while back, it doesn't predict anything just helps you sift through data and identify value. I found it cool, it is someone's hobby project that keeps getting bigger and bigger. Perhaps it or its databases might be of use to you.

It's here: http://www.cgmbet.com/

Betfair Exchange sadly is not available in germany. Betfair is only available as a bookmarker here and I consider using it because they do have great odds. The tool looks very interesting and similar to the stuff I'm working on right now so I'll keep an eye on it

Hard work Beats Tallent, if Tallent didnt work hard

Absolutely true

Hey, for everyone who is interested in this topic here, the search engine bing is not bad in predicting the results of an game.
You can read more about it here: qz(dot)com/233830/world-cup-germany-argentina-predictions-microsoft/




@issorc You said that you have more than 105k games in your database.
According to qz(dot)com/259556/after-crushing-the-world-cup-microsoft-is-predicting-the-nfls-whole-season/ they wrote "A full 256-game NFL season will be tough to predict accurately. While the performance in the World Cup’s knockout rounds was impressive, the model was less successful during the more frenetic and much larger group stage of the tournament." And that is right. I don't know how many leagues you are using at the moment, but keep in mind that more is not always better. You can cover up a wide range, but you need a massive amount of data to cover up the range and also get usefull results.
I don't remember if you use other variables than the game database, but you can take a lot other variables with in your database. To be honest I am doing the same like you. I also make an semi ai to predict some results and love ai, machine learning, etc.. The last weeks I talked sometimes with my father about an ai which can predict all soccer results perfectly. Yes that can never become true, but we talked about what variables influence an soccer game. There a lot of useful things you can use:

• Player statistics (google sofifa and then you can find nice stats from the fifa about the player)
• The weather (was/ is it sunny, rainy, hot, warm, ... all can influence the performance of the team a lot)
• You can use whoscored(dot)com to find many many useful informations about the games
• Search how was the percentage of the fanbases from the teams in the station (don't think that there is an online source for that, but it's worth a try)
• You can also use Twitter, Instagram, etc.. I can't find the paper at the moment, but they described in there how the predicted the outcome of an game only using Twitter. That's very interesting in my opinion .
  
• ...

I hope I can help you with these proposals.
Good luck man, you can do it

I see that leagues and group stages of tournaments are maybe harder to predict because more often the underdogs wins. In later rounds every team really needs to give its best so the favourite takes it home more often. The results of Bing's prediction algorithm are impressive anyway. Right now we're working on improving everything. I'm really thankful for your list of possible data to use because most ideas are pretty good. But before looking for more data I want to have everything else up and running

I posted a to-do list a week before and I finished some of the tasks.
Right now my scripts are able to automatically:

• Update the database with the newest results of 33 different leagues
• Get the upcoming fixtures of all these leagues
• Calculate the odds for these upcoming fixtures
• Display everything in a simple web interface using a table to list all games and sortable columns for the odds
• Delete fixtures after the games have finished

All of this happens without any input from my side. I just sit back and wait for the newest odds to be posted on my website I'm working on a script now to get the odds for the fixtures from a bookmaker to compare the odds and display the most valuable bet. The tool calculates 8 odds for every game so I can't manually compare 100+ games on a weekend day to the bookmaker odds. This will still need some time to finish but I hope that this will be done in 2 weeks max. Thanks for wishing me luck

Great. A lot of good inputs from a lot of people. I like where this is going.

I say it's time to start making an external Skype group of everyone involved in AI & want to help this topic go further.

So if you're developer, web designer, if you have ideas, anything, let's make this happen. The goal is not to post prediction on this group but rather have everyone involved in working on this in order to post more prediction here in BHW.

PM me and I'll set this up

Really great idea, you know I already hit you up, but I hope more people are following. I wanted to do this alone in the beginning because I didn't have a lot to share nor did I really need input at that time. Now I'd love to share my thoughts with some like-minded people.

This can be a very dangerous thread... careful with your bets, guys. Machine learning can only recommend but never predict with 100% accuracy, ever. Past data can only predict so much and can't completely disregard any random events that may happen in the future.

Nonetheless, interesting to see how you'll develop your model OP. Would be interesting to see what sort of systems thinking model/archetypes you've drawn up to create your algorithm.

This is completely true. That's why I've right now switched from correct score bets to betting on BTTS or Over/Under Goals. My Correct score model wasn't able to output a probability for that event to happen. This way I just could hope that some of the higher odds would work out because I couldn't compare the odds to the bookmakers and find value in it. Right now we're using an exponential declining function to weight the games over time and poisson distribution for probability of a specific amount of goals. I hope everything I say is right I actually don't know a lot about mathematical terms and equations in languages different to german yet.

You should work out along with your predictions a confidence score and see how that impacts the results when only betting above a certain confidence level. While the general subject is very interesting I am very skeptical that this can be profitable an scale able over a year/years but it's definitely very interesting to follow.

We're currently working on a formula to find the best combination of risk (or probability given by our calculated odds), the odds of the bookmakers and the amount of money to bet. Not that easy though. I understand your skepticism and I hope to be able to prove you wrong

Followed your predictions, and with 14€ bets i had 17.23€ returns. So about 23% ROI.

I'm happy that you've made some money but please don't invest more into my bets. This is all very experimental and in an early stage. Nevertheless I appreciate your trust in my project

Do you bet 1 euro per match all single bets?

Saturday i tryed this multiple bet:

Pescara - Avellino O2.5 @1.47
Lokeren - Charleroi U2.5 @1.85
Sligo Rovers - Derry City U2.5 @1.85 (lose)
Rennes - Lille U2.5 @1.61
Cesena - Foggia O2.5 @1.85
Udinese - Juventus O2.5 @2 (personal prediction)

5€ --> 132€ return

Too much matches. I want to figure out what's the best number of matches for a good return, considering the OP has a 70% of average correctness

Although this bet didn't work it out was really good actually. Sad that Sligo stopped you from winning

I just wanted to give you a quick overview of the result of the bets I posted in this thread with my newest software to keep a small record. For this record I'll estimate that for every game the same amount of money (=1 unit) will be used.

Bets from 21.10.2017:
Bets won: 10/17
Spent units: 17
Won units: 16,35
Total: -0,65 units
ROI: -3,82%

Bets like the huddersfield game and other low odds killed this day. I learned my lesson and not gonna bet on such stupid odds again, no matter how safe it seems. In the end it's still soccer and everything can happen

Bets from 22.10.2017:
Bets won: 10/14
Spent units: 14
Won units: 16,8
Total: +2,8 units
ROI: 20%

This day actually was quite good

Bets from 24.10.2017:
Bets won: 3/6
Spent units: 6
Won units: 5,95
Total: -0,05 units
ROI: -0,83%

These bets were really unlucky, once again two games that failed in the last minutes. Nevertheless nearly hit break even.

Total history so far:
Bets won: 23/37 (~62%)
Spent units: 37
Won units: 39,1
Total: +2,1 units
ROI: 5,6%

I'm really looking forward to tomorrows games and will post my bets once again then

 

[redjebo]

We're currently working on a formula to find the best combination of risk (or probability given by our calculated odds), the odds of the bookmakers and the amount of money to bet. Not that easy though.

You most probably know about this, but just in case, the optimal formula regarding ROI for each wager expressed as % of your bankroll is the Kelly criterion. It can be optimized a bit further for Risk of Ruin i.e scaled down a bit, which is better in this case since you can never be sure your calculated odds are accurate enough and you can easily overestimate your edge.
For Kelly to make sense though you need to know your edge over the bookies odds, i.e. how much higher the actual probability (your estimation of it) is, than what the bookies have calculated OR implied in their odds. You obviously need to be more accurate than the bookies in calculating that probability (which is very hard, especially in major leagues events) OR (the good news) accurate enough to just recognize the value in odds, construed on a basis of balancing the money flow between outcomes, which bookies are forced to do more or less. For example a very popular team playing a more mediocre opponent will often attract the masses to bet more confidently on the favorite than they actually should due to fan bias, thus creating a huge potential loss for the bookie if the favorite goes on and win. So the bookie will be forced to offset it's risk by lowering their odds on the fav and raising the underdog's odds to balance the bets, thus sometimes giving away value on the underdog.

 

[issorc]

No bets today, spent my time improving my tool and automating the comparison of my odds to the bookmakers odds. Here are some bets for tomorrow

Spoiler

Sporting Gijon - Almeria NoBTTS @1.66
Inverness CT - Dunfermline NoBTTS @2.25
San Martin de San Juan - Estudiantes BTTS @2.1
Heracles - VVV NoBTTS @2.37
Queretaor - Atlas Over2.5 @2.1
Venezia - Frosinone NoBTTS @2
Bari - Ascoli NoBTTS @1.83
Salernitana - Empoli Over2.5 @2.1
Reims - Nancy NoBTTS @1.66
Lorient - Tours NoBTTS @1.9
Guingamp - Amiens NoBTTS @1.72
Strasbourg - Angers Over2.5 @2.1
Sandecja Nowy Sacz - Gornik Zabrze NoBTTS @2.1
Hull - Nottingham Forest Over2.5 @1.66
Cracovia Krakow - Nieciecza NoBTTS @2.1
Cardiff - Millwall NoBTTS @1.8
Fulhalm - Bolton NoBTTS @2.1
Luton - Coventry NoBTTS @2
Charlton - AFC Wimbledon NoBTTS @1.83
Wigan - Blackburn NoBTTS @1.72
Liverpool - Huddersfield NoBTTS @1.95
Arsenal - Swansea NoBTTS @2.05

 

[SOCOM]

I think you need a sport with less winning options. Soccer, for instance, has 3 ending possibilities which is a nightmare. On the other hand, American football has 2. Your system selects the top 5 games based on confidence percentages and you place bets on those games.

 

[SEOMadHatter]

A few of my major current projects over the last two years are relevant to this and a friend of mine helped build a deep learning system which kills UK flat racing.

I've been reading into neural networking recently and damn it's interesting stuff. With a billion different projects I struggle to find the time to work on this as much as I'd like to but I've a potential JV starting in the next week or so to start training a network. We're doing things a little differently so we're not in direct competition.

Betting/trading is an obvious target for machine learning so you're up against anyone with a laptop but more importantly traders who have teams of data engineers and investments in hardware because there's money to be made. So I was struggling to believe it wasn't massively saturated but if people do this stuff manually by studying patterns then maybe a network can still be trained to be profitable. It's just about finding that niche and paper testing like mad before committing too much of a bankroll.

Also be careful with the Kickstarter product you linked. Dug into it a bit, spoke to them and a couple of math/trader guys and I don't see anyone making money with it. You can make old data fit any kind of model until it looks profitable and they're data engineers selling this not traders. I think for a service to be worth looking at you really need both.

 

[issorc]

I decided to stop or at least pause this journey at this point. The tool I developed has nearly any features I wanted and anything else is too much right now. I will continue to educate myself about deep learning and artifical intelligence because I'm really interested in this field and this is where I want to go. For now I have some stressful months in university ahead of me and will shit my focus there and back to my affiliate projects. I decided to clearly set my priorities and this was and still is fun to do but nothing really important. I'm thankful for everybody who placed a comment or just read everything and followed along. This was a fun project for me and at least I learned a lot I see no reason to keep posting my bet advice here because it's not proven that these bets are good (although I made some profit last weekend but a small loss on this weekend yet). This thread should keep you guys up to date what happens with my tool/product/service whatever and not some silly bets In case I'll return to this project I promise you guys to keep this thread up to date. And in case that I should start another project that uses AI or deep learning I'll create a new journey Until then I hope everybody enjoyed this journey at least a bit because it contains many really useful and interesting posts imo