Hi guys. Here is the idea that I discussed with Patrick other day.
I have been using the line.com website and forum for a long time since old pickmonitor days and really love it.
For a long time I have been thinking how I can help cappers improve their predictions and service they provide to the clients. And instead of doing one more capper service we decided to go another way and develop the prediction platform for sports.
I come from the Forex background where the strategy is considered good only if it passed a long backtest (EA backtesting i MT4) and a forward test. I decided to use the same approach in sports. That's why over the last years we gathered all data for NHL, MLB, NFL and NBA since 1999 to develop a prediction model based on 80+ parameters.
To keep it short, during the last 3 years my team of programmers have been developing sports predictions engines that can help generate precise predictions and help cappers improve their results. I am talking about predictions here and finding VALUE in the line, not about one more picks service. Here is a sample of how the simple prediction look like and result from MLB 2011:
Our platform is not finished yet and in Beta stage but it already produces great results (For example we won ALL predictions except for Seahawks on NFL last week).
Here is the sample from NHL. The value was found in Toronto game for +1.5, Toronto moneyline and Over 5.5:
Or another one: Prediction was for Under 5.5, high 76% +1.5 Value on Columbus also won. We failed only on Wild yesterday.
Let me know if you are interested in my daily predictions. Currently we have NHL and NFL. While we are in Beta it would be great to share with you the predictions and discuss problems and ideas to make it better.
Hope this helps line.com cappers improve their picks. We don't have a website finished yet so no links here.
If you like the idea I will be sharing the predictions here on the forum. Let me know.
If the thing is predicting Sharks like everyone else tonight then it has the basic flaw of knowing nothing about the actual game of hockey or the players. If it's predicting the Kings somehow then stop wasting time and sign me up for this combobulation.
The main goal of predictions here is not just to guess who is going to
win but to find the value in the games. Sometimes it comes on favorite,
sometimes we go +1.5 or Underdog. Totals also may have a lot of value
(shown as stars on my examples). Here is a fresh one from NFL where
system found 5 start underdog value:
The Broncos won that one despite heavy favored Chargers playing home.
The goal is to put more money on games with more value and avoid games
with little or no value at all where the result is too overjuiced or
totally a coin flip.
The Sharks or any other team work the same way the currencies work in Forex. We find a trend and ride it, taking most of it. If we don't have a trend we scalp on the small movements. If the goal is to make money it does not matter if you watched every single game or not. If there is a trend or pattern that has a high chance of repeating itself - this is what helps to find out where the value is. This does not mean it is a replacement for a professional capper experience and feelings. It is a tool that gives you confirmation when the signal is strong or warning to avoid the trade when situation is unclear.
This could be any interesting tool - like the many Forex comp modelling tools available. BUT why not take your selections here - at Line.com based on your prediction modelling, that will provide you "real" results and returns especially when taking such short odds on selections (+1.5 selections). I note you have not made any selections here at Line.com - but have the perfect environment here to "prove" up your prediction modelling. Taking the +1.5 line in NHL requires extreme accuracy to be profitable, especially with odds of -270, -290, -300 - you will need to be accurate 75% of the time to show a very small profit on a big outlay. IF you can get the game totals in your modelling working "in real odds" this could be a potential tool.
>> BUT why not take your selections here - at Line.com based on your prediction modelling
Slowdraw, Yes, I will definetely plan to do that when the system is ready (currently it is in beta and we keep changing and improving it so long term test would have no meaning). Moreover I am waiting (like CashMagician) for line.com to introduce more flexible options such as option to bet any bet amount we want in range of 50-500 or from 10 to 100 or at least in fractions of 10 or 50. Currently I have only option to bet, say $100 for 1 star and $500 for 5 stars but that is TOO much for 10,000 bankroll with so many games on NHL and NFL it is not possible to put bets in such huge amounts on such a small bankroll. One day I tried to place the bets according to stars but I got something like "In Play: $8,610" for just 5 games which is a suicide for 10k bankroll. For 10k bankroll I would never bet on multiple bets of $500 on a real account.
Anyways, the point of the thread was to share some of the predictions which I find interesting and see what you think and how they turn out later. Last week we won nearly ALL predictions on NFL except for seahawks. NHL is more unpredictable lately.
The system generates predictions for all games but most interesting today are the following 3. Usually I place bets on all selections instead of trying to find "which is best" according to stars. I post olny three of them, interesting to see your opinions on them. If you agree or disagree with them.
Phoenix Coyotes at Chicago Blackhawks (NHL) Nov. 29th, 2011 8:30 PM EST While Phoenix is stuck with W-L pattern which sounds like they are due for a loss, Chichago is back on track. The system gives us 3.5 star "game winner" prediction for Chicago and also 40% (I like everying over 38% here) for -1.5 chance. In other words 2-4 is a possible score here.
2 Star game. Under 5.5 and Sabres to win. Also I would go Sabres-1.5 here. (42%) with a small bet. The Islanders are too weak to win 2 games in a row on road and Buffalo is coming after 5-1 Win.
Florida Panthers at Carolina Hurricanes (NHL) Nov. 29th, 2011 7:00 PM EST This game has an interesting controversy. According to the bookies the favorite based on the odds is Florida Panthers however based on ZCode calculations the real predicted game winner is Carolina Hurricanes. The system gives 3.5 Stars to Florida to win with 45% confidence on that. 1 Star for Hurricanes home underdog and pretty high chance of +1.5 for them (79%). Solid under 5.5 expecting a tight game with 2-3 or 3-2 score.
For by opinion on the above selections: they are almost exactly to what I have selected for those games - BUT I would not wager on the Panthers game - but believe that it will be u5.5. The Sabres game is as I am selecting here and wagering.....sabres -1.5 (Isles have not won away from home this season & to win two in a row is unlikely) The total I also agree on unders as a 0/1 to 3/4 game. Chicago I agree is a win BUT not confident in the -1.5 start.. My other best is the jackets-canucks game - same reason as the Isles/Sabres game (Jackets have only won once away) and I see this as a one sided result.
Predictions are helpful - BUT as with all of us - we cannot wager on every game in every sport - the whole idea is to "try" to select a limited number of the 'best' possibilities. For some this may be one selection a day or even one a week - Half the ability of long term success is "choosing" which selection to wager (many leave winning selections and have chosen our most likely only to have them unsuccessful)
So any prediction tool - must have "rules" that help produce results - ie... only wager on 5 star selections....???
no 5 star - no wager, 3 5 star - place 3 wager
You need to work out if 5 star has greater success than say 3 star selections..........
My suggestion is to trial your selections here over 30 day periods - you can reset - but without "actual" returns against actual wagers - you have no idea. As I stated above - it would have to be exceptional to make a long term profit on +1.5 selections.
I also disagree with the staking - you can trial simply with the 100 to 500 wagering available...... surely u would not wager on 1 or 2 star selections, hardly on 3 star but mainly wager on 4 & 5 star, u can assign 100 to 3 star, 200 to 4 star and 400 to 5 star selections ............ Surely on any day you would not have more than 2-6 5 star selections (assuming this is the greatest chance of success) hence say 5 wagers at $400 would give you 5 days before the bankroll is broke - IF your selections of 5 stars does not have greater than 50% success - GET a dartboard to help select So if the predictions work your bankroll would last a few months - even going poorly....
Awesome feedback, Slowdraw, I really appreciate the input. Yes, I will track of course and we are coding the option to backtest predictions back on the last games to see how they could perform depending on the stars. However the purpose of the thread is to find the discrepancies on what the human professional capper would do and system would suggest on the same game.
here is the columbus game you mentioned: I like the fact Canucks score a lot and usually win with big margins. If they win this one it is not supposed to be a tight game. Vancouver moneyline and a small -1.5 for 38% is our choice here.
About Columbus @ Vancouver and NY Islanders @ Buffalo. Both home teams have some serious losses (the importart players are out or questionable for today's matches) So, I think, it's better to play +1,5 on away teams or simply on WIN with overtime.
Perdon, I'm new in NHL, so I can be totally wrong...
No, thank you!) Very nice idea and project, wish you good luck and I will surely look after your picks. - Yes) Thank you, I'm trying to do my best to provide good picks.
Meanwhile we are developing a possibility to backtest our live predictions back say 30 days (at least) to be able to tell exactly which predictions generate best results and where we are in a long run and find hot trends. Here are the quick facts:
+$979 $ Profit by betting $100 bets with only Underdog value 4-5 star picks in NFL during the last 30 days +$1579 $ Profit by betting $100 bets with only Underdog value 4-5 star picks in NHL during the last 30 days +$655 $ Profit by betting $100 bets with only Game Winners 4-5 star picks in NFL during the last 30 days +$1211 $ Profit by betting $100 bets with only Game Winners 4-5 star picks in NHL during the last 30 days -$180 $ Profit by betting $100 bets with only Over-Under picks in 4-5 star picks in NHL during the last 30 days (yes I fully understand you might be sceptic here and think that I just faked that numbers but remember, the whole idea of this thread and public predictions is to show that our system is fully transparent so when we lose everyone can see we lose and when we win everyone can win with us, so faking anything is not the option here).
The numbers seem very promising except for OverUnder which shows we have a big room for improvement here. Later I will get more numbers for smaller star amounts as well.
However these numbers might give us wrong assumption that we should ignore all predictions lower than 4 stars which might be a mistake and I would not be surprised if those lower confidence predictions will bring more profit in a long run.
Yesterday just for fun I decided to put all the predictions on my private account here - everythign according to stars counting 1 star = $100 and 5 stars as $500. This is of course suicide because this required me putting 90% of acccount at risk with such amounts. For real account I would devide each bet by 10-20. so $5 =1 star.
So here are 29 November results. Looks like the day was profitable for our predictions despite of chicago and buffalo losses. Here are all picks that I placed yesterday based on Zcode recommendations. I put them according to stars even bets that are less than 4 stars.
Boston Bruins at Toronto Maple Leafs (NHL) Nov. 30th, 2011 7:00 PM EST Hot trends: 83% Winning Rate Predicting Last 6 Boston games Boston won 80% in favorite status in last 5 games Toronto covered the spread 100% in last last 5 games as underdog
Recommendations: Good underdog value pick (5 Stars) on Toronto Toronto +1.5 Spread line (72.00% chance to cover the spread) Very high chance 72% of the tight game that can go into overtime
Boston has 57% chance to win this one however the system found much more value in home underdog and its +1.5 bet as well expecting a tight overtime game. Boston won too much lately. Someone HAS to stop that. Will it be Toronto? What do you think?
Toronto +1.5 according to you has a 72% chance of covering ATS. Toronto +1.5 right now is priced at 1.45 thus the implied probability to break even 69% of the time. You're laying better than -200 juice to work on a maximum ROI per play of 3%. GIC's or whatever the equivalent in your country, will play better than that.
I don't have a model to tell how often a game goes to OT, nor do I really want one as I don't bet with that - but I tend to agree that this would be better than 6/10.
I tend to agree with your numbers actually, but not necessarily how you are suggesting using them.
Nice handicapping tool you have developed there. I like using multiple stat models to help make decisions and like what you have created. The Maple Leafs seems like a good play, but both teams are on fire, should be a nice game to watch. What does the Zcodesystem say for the Lightning vs. Red Wings game tonight?
I have been using the line.com website and forum for a long time
since old pickmonitor days and really love it.
For a long time I have been thinking how I can help cappers improve
their predictions and service they provide to the clients. And
instead of doing one more capper service we decided to go another way
and develop the prediction platform for sports.
I come from the Forex background where the strategy is considered good only if
it passed a long backtest (EA backtesting i MT4) and a forward test. I decided
to use the same approach in sports. That's why over the last years we gathered all
data for NHL, MLB, NFL and NBA since 1999 to develop a prediction model based on 80+
parameters.
To keep it short, during the last 3 years my team of programmers have been developing
sports predictions engines that can help generate precise predictions
and help cappers improve their results.
I am talking about predictions here and finding VALUE in the line, not about one more picks service.
Here is a sample of how the simple prediction look like and result
from MLB 2011:
Our platform is not finished yet and in Beta stage but it already produces great results
(For example we won ALL predictions except for Seahawks on NFL last week).
Here is the sample from NHL. The value was found in Toronto game for +1.5, Toronto moneyline and Over 5.5:
Or another one:
Prediction was for Under 5.5, high 76% +1.5 Value on Columbus also won. We failed only on Wild yesterday.
Let me know if you are interested in my daily predictions. Currently we have NHL and NFL. While we are in Beta it would be great to share with you the predictions and discuss problems and ideas to make it better.
Hope this helps line.com cappers improve their picks. We don't have a website finished yet so no links here.
If you like the idea I will be sharing the predictions here on the forum. Let me know.
Would be interesting to see your predications vs outcome
If the thing is predicting Sharks like everyone else tonight then it has the basic flaw of knowing nothing about the actual game of hockey or the players. If it's predicting the Kings somehow then stop wasting time and sign me up for this combobulation.
The Broncos won that one despite heavy favored Chargers playing home.
The goal is to put more money on games with more value and avoid games with little or no value at all where the result is too overjuiced or totally a coin flip.
If the goal is to make money it does not matter if you watched every single game or not. If there is a trend or pattern that has a high chance of repeating itself - this is what helps to find out where the value is. This does not mean it is a replacement for a professional capper experience and feelings. It is a tool that gives you confirmation when the signal is strong or warning to avoid the trade when situation is unclear.
I note you have not made any selections here at Line.com - but have the perfect environment here to "prove" up your prediction modelling.
Taking the +1.5 line in NHL requires extreme accuracy to be profitable, especially with odds of -270, -290, -300 - you will need to be accurate 75% of the time to show a very small profit on a big outlay.
IF you can get the game totals in your modelling working "in real odds" this could be a potential tool.
Thanked by
Slowdraw, Yes, I will definetely plan to do that when the system is ready (currently it is in beta and we keep changing and improving it so long term test would have no meaning). Moreover I am waiting (like CashMagician) for line.com to introduce more flexible options such as option to bet any bet amount we want in range of 50-500 or from 10 to 100 or at least in fractions of 10 or 50. Currently I have only option to bet, say $100 for 1 star and $500 for 5 stars but that is TOO much for 10,000 bankroll with so many games on NHL and NFL it is not possible to put bets in such huge amounts on such a small bankroll. One day I tried to place the bets according to stars but I got something like "In Play: $8,610" for just 5 games which is a suicide for 10k bankroll. For 10k bankroll I would never bet on multiple bets of $500 on a real account.
Anyways, the point of the thread was to share some of the predictions which I find interesting and see what you think and how they turn out later. Last week we won nearly ALL predictions on NFL except for seahawks. NHL is more unpredictable lately.
Phoenix Coyotes at Chicago Blackhawks (NHL) Nov. 29th, 2011 8:30 PM EST
While Phoenix is stuck with W-L pattern which sounds like they are due for a loss, Chichago is back on track. The system gives us 3.5 star "game winner" prediction for Chicago and also 40% (I like everying over 38% here) for -1.5 chance. In other words 2-4 is a possible score here.
2 Star game. Under 5.5 and Sabres to win. Also I would go Sabres-1.5 here. (42%) with a small bet. The Islanders are too weak to win 2 games in a row on road and Buffalo is coming after 5-1 Win.
Florida Panthers at Carolina Hurricanes (NHL) Nov. 29th, 2011 7:00 PM EST
This game has an interesting controversy. According to the bookies the favorite based on the odds is Florida Panthers however based on ZCode calculations the real predicted game winner is Carolina Hurricanes. The system gives 3.5 Stars to Florida to win with 45% confidence on that. 1 Star for Hurricanes home underdog and pretty high chance of +1.5 for them (79%). Solid under 5.5 expecting a tight game with 2-3 or 3-2 score.
Let me know your opinion please on these thress.
Chicago I agree is a win BUT not confident in the -1.5 start..
My other best is the jackets-canucks game - same reason as the Isles/Sabres game (Jackets have only won once away) and I see this as a one sided result.
Predictions are helpful - BUT as with all of us - we cannot wager on every game in every sport - the whole idea is to "try" to select a limited number of the 'best' possibilities. For some this may be one selection a day or even one a week - Half the ability of long term success is "choosing" which selection to wager (many leave winning selections and have chosen our most likely only to have them unsuccessful)
So any prediction tool - must have "rules" that help produce results - ie... only wager on 5 star selections....???
no 5 star - no wager, 3 5 star - place 3 wager
You need to work out if 5 star has greater success than say 3 star selections..........
My suggestion is to trial your selections here over 30 day periods - you can reset - but without "actual" returns against actual wagers - you have no idea.
As I stated above - it would have to be exceptional to make a long term profit on +1.5 selections.
I also disagree with the staking - you can trial simply with the 100 to 500 wagering available......
surely u would not wager on 1 or 2 star selections, hardly on 3 star but mainly wager on 4 & 5 star, u can assign 100 to 3 star, 200 to 4 star and 400 to 5 star selections ............ Surely on any day you would not have more than 2-6 5 star selections (assuming this is the greatest chance of success) hence say 5 wagers at $400 would give you 5 days before the bankroll is broke - IF your selections of 5 stars does not have greater than 50% success - GET a dartboard to help select
So if the predictions work your bankroll would last a few months - even going poorly....
here is the columbus game you mentioned:
I like the fact Canucks score a lot and usually win with big margins. If they win this one it is not supposed to be a tight game.
Vancouver moneyline and a small -1.5 for 38% is our choice here.
Both home teams have some serious losses (the importart players are out or questionable for today's matches)
So, I think, it's better to play +1,5 on away teams or simply on WIN with overtime.
Perdon, I'm new in NHL, so I can be totally wrong...
KeLT27 - I see you locked two picks on Sabres as well same like mine? Your NHL record looks really stable and winning. Good job!
- Yes) Thank you, I'm trying to do my best to provide good picks.
Meanwhile we are developing a possibility to backtest our live predictions back say 30 days (at least) to be able to tell exactly which predictions generate best results and where we are in a long run and find hot trends. Here are the quick facts:
+$979 $ Profit by betting $100 bets with only Underdog value 4-5 star picks in NFL during the last 30 days
+$1579 $ Profit by betting $100 bets with only Underdog value 4-5 star picks in NHL during the last 30 days
+$655 $ Profit by betting $100 bets with only Game Winners 4-5 star picks in NFL during the last 30 days
+$1211 $ Profit by betting $100 bets with only Game Winners 4-5 star picks in NHL during the last 30 days
-$180 $ Profit by betting $100 bets with only Over-Under picks in 4-5 star picks in NHL during the last 30 days
(yes I fully understand you might be sceptic here and think that I just faked that numbers but remember, the whole idea of this thread and public predictions is to show that our system is fully transparent so when we lose everyone can see we lose and when we win everyone can win with us, so faking anything is not the option here).
The numbers seem very promising except for OverUnder which shows we have a big room for improvement here. Later I will get more numbers for smaller star amounts as well.
However these numbers might give us wrong assumption that we should ignore all predictions lower than 4 stars which might be a mistake and I would not be surprised if those lower confidence predictions will bring more profit in a long run.
So here are 29 November results. Looks like the day was profitable for our predictions despite of chicago and buffalo losses. Here are all picks that I placed yesterday based on Zcode recommendations. I put them according to stars even bets that are less than 4 stars.
Boston Bruins at Toronto Maple Leafs (NHL) Nov. 30th, 2011 7:00 PM EST
Hot trends:
83% Winning Rate Predicting Last 6 Boston games
Boston won 80% in favorite status in last 5 games
Toronto covered the spread 100% in last last 5 games as underdog
Recommendations:
Good underdog value pick (5 Stars) on Toronto
Toronto +1.5 Spread line (72.00% chance to cover the spread)
Very high chance 72% of the tight game that can go into overtime
Boston has 57% chance to win this one however the system found much more value in home underdog and its +1.5 bet as well expecting a tight overtime game.
Boston won too much lately. Someone HAS to stop that. Will it be Toronto? What do you think?
Toronto +1.5 according to you has a 72% chance of covering ATS. Toronto +1.5 right now is priced at 1.45 thus the implied probability to break even 69% of the time. You're laying better than -200 juice to work on a maximum ROI per play of 3%. GIC's or whatever the equivalent in your country, will play better than that.
I don't have a model to tell how often a game goes to OT, nor do I really want one as I don't bet with that - but I tend to agree that this would be better than 6/10.
I tend to agree with your numbers actually, but not necessarily how you are suggesting using them.