hi, hopefully im going to archive my picks and various stats here for a large sample size and give away some rice bowls!

some facts:
-im going to post the games that my model selects from all the soccer matches from pinnacle sports
-the prices however come from pinnacle as well as major higher limit asian bookies to get the best price
-im posting my plays when pinnacle lines come out, thats about 6 days prior for major leagues and 2-1 days for minor
-im not betting the first 5 rounds and last 2 rounds of the season, also in summer and winter there is only about 40% volume compared to spring and autumn
-that said i will be achieving around 15-20 bets in a week in summer/winter and 35-40 bets in a week in spring/autumn
-i model only totals and post the market closest to coinflip odds to reduce variance
-ive decided not to post picks as %s or units, so i count every pick as bet 1 for my stats overview (it wont make much difference, since my avg odds will be 1.94)
-every week i will post my overall for the week and total, i have all kinds of stats, so you can ask for smth you would like to see that isnt there

GL

23.05 Norwegian T - Viking/Stabaek under 2.75@2,06 SBO

23.05 Segunda Liga - Gimnastic/Dep. La Coruna under 3@2,08 Pinnacle

24.05 Finnish V - Honka Espoo/Kups under 2.5@1,92 188Bet

24.05 Allsvenskan - Elfsborg/BK Hacken under 2.75@1,97 Dafabet

25.05 Superettan - Degerfors/Landskrona over 2.75@2,06 188Bet

26.05 Superettan - Jönköpings/IK Brage over 2.75@1,92 188Bet

26.05 J-League - Kashiwa Reysol/Albirex Niigata over 2.25@1,83 SBO

28.05 Argentine - Colon de Santa Fe/Velez Sarsfield under 2@2,03 188Bet

is your model proprietary? or at least a black box?

not sure what you mean
i just want to reach as high level in modelling soccer totals as possible
made my algorithms for around 7 months i think, then proceeded to live-test them
the whole scraping and db thing has been running since mid-december, have 6,7k soccer games stats
now pretty much run out of ideas, trying to read market numbers, especially closing line efficiency, which has pretty weird numbers, can post some stats on that later on
this is a long grind, hoping to get better in time
im definetly +ev, but by how big margin, im too conservative to tell, because the range is too big, even with the 560 bets ive live-tested so far

can you tell us what data you scrape and use in your model? im most interested in whether injuries are accounted for

more important question. windows, mac or linux?

im scraping from several different sites, which id rather not mention
my roi is much higher betting early prices, therefore i use last games`s lineup
for me at the moment its much more profitable to focus on that, rather than wait for a few hours before the match when true lineups come out, also injuries do not happen that often for them to change a large amount of games

game-day modelling with new lineups is going to be my priority in the far future, when it is evident, that i cant get enough action down with smaller limits.

i edited my last post right as you "answered" it. i am still interested in what data you scrape NOT where you scrape it from.

also why wouldnt you bet the last 5 rounds of a season? i understand not betting the first 5 rounds since you need data on how the teams play but why not the last 5 rounds. do you think title/relegation battles affect the totals significantly?

also what language/stat package are you using? R/Python/C/HTML/Brainf*ck?

im not that computer-savvy, so im still using windows and becoming more and more concerned about everything being so slow, programming is definetly something i need to focus on in the future. ive got several large convoluted excel files, that are pretty frustrating to use. ive managed to automate everything, so pushing refresh button does the trick.

its the last 2 rounds that im not betting, and thats an arbitrary number, since i dont have alot of data on the last 2 rounds, im not taking risks betting those, but logic tells me there are definetly different goals for different teams opposed to mid-season, when everybody just wants to win

If you don't think the markets are efficient why is your roi better on early lines?

i was going to get a hedge fund to invest heavily but after reading the replies from OP i realize he is ignorant and i wouldn't advise anyone to invest in this.

at the moment i have 6,7k games since 16.12
-line movement has shifted to the winning side 52% opposed to 48%, not counting pushes
-lines moves only 0,7 cents towards winning side on average, overs move only 0,3 cents and unders move 1,1 cents
-overall lines move 0,8 cents towards over, that can be explained that a lot of people maybe use pretty basic models and using poisson results biased towards overs
-games have ended 44,1% over and 45,5% under, others push, so pinnacle lines are pretty perfectly chopping to the middle

however when im looking at steam

-10+ cent steam occurs 36,7% of the time and wins 54,2% (overs win 53,5%, unders 54,9%)
-20+ cent steam occurs 11,7% of the time and wins 56,4% (overs win 53,8%, unders 58,5%)
so betting steam is slightly -ev, however unders are definetly close to 0ev
and steam numbers tell us that the market is on overall pretty efficient

any thoughts?

i am live-testing 5 slightly different algorithms, the one im using for this thread is beating the vigged closing line on avg 4,5 cents, but one of my model beats the closing line on avg 8,0 cents (though im not using it, because it has performed very poorly over 300 samples, which has my mind boggling about the efficiency of the market)

wat

yeah I'm tutoring so distracted, but that kinda lost me too.

by 10+ cent steam i meant that closing line is 10+ cents down from opening line
so when avg odds are 1,94 usually and 51,55% needed to break-even then with 10+ cents shift that becomes <=1,84 thus minimum 54,35% to break-even, prob. around 55% in this case

just proves that blindly chasing steam is definetly not profitable

that happens every time not 36.7% of the time