BGO, some of the lines show the books feel that teams are <due> to perform one way or another after a recent string of performances. I'm no expert on regression to the mean, I once explained my idea on LVSC <due factors> but got booed off the stage, lol. They have regression factors based on decades of typical team performance indicators as compared to public betting patterns in similar scenarios. A combination of team x will most likely do 'this' against 'the volume of money will be 'this' on a line of 'that'. Very complicated and easier shown with an example.
Hi all,
there's still no requirement for heads to catch up.Try 2,000,000 evenly split tosses after the initial 70/30 split.The respective outcomes are 50.001% & 49.999%.The outcomes are approaching the true odds without the need to play catch up...they may do,or even overtake the previous leader,but they don't have to.
The initial 70/30 tosses are just becoming more & more insignificant to the overall number of tosses.
I like AXT's point that games aren't coin flips,they just seem like they might be because the ats stats tend to come in at around the 50% mark.
A game has an inbuilt probability that one team will win at the start.Unfortunately you still don't know if your assessment of this probability was correct,even after the fact.
The best you can do is use certain criteria to predict a game probability,group all the games over a period of time that you think one team has a say 60% chance of winning & see if over time this tallies with reality.
Even then you've got to deal with random noise(good/bad luck) which might be quite significant in a medium scoring,technically complex game like gridiron.
Bottom line is that the ats numbers merely determine if you bet & on which side.If a team's covered 7 times in a row & you think they'll cover an eigth because of the number that you've come up with,then you bet it.The streak should be irrelevent.
Not a great fan of streaks or trends.If you stick enough rules in(dogs on the road,scored 20+ points last week,have a winning record against conference rivals who wear blue uniforms etc)you're bound to find trends that are profitable.
What you don't know is if your rules are causative or exhaustive.
Also many trends use the initial trend to prove that the trend is continuing,when at the very least you should only count games played after you "discovered" the so called trend.
The really interesting point would be to know how betting money rides the trends.When do they start playing them,for how long & when do they get off & start betting the otherway.Of course if your fundamental handicapping of a game is any good then your numbers should be able to tell you this.
P.
there's still no requirement for heads to catch up.Try 2,000,000 evenly split tosses after the initial 70/30 split.The respective outcomes are 50.001% & 49.999%.The outcomes are approaching the true odds without the need to play catch up...they may do,or even overtake the previous leader,but they don't have to.
The initial 70/30 tosses are just becoming more & more insignificant to the overall number of tosses.
I like AXT's point that games aren't coin flips,they just seem like they might be because the ats stats tend to come in at around the 50% mark.
A game has an inbuilt probability that one team will win at the start.Unfortunately you still don't know if your assessment of this probability was correct,even after the fact.
The best you can do is use certain criteria to predict a game probability,group all the games over a period of time that you think one team has a say 60% chance of winning & see if over time this tallies with reality.
Even then you've got to deal with random noise(good/bad luck) which might be quite significant in a medium scoring,technically complex game like gridiron.
Bottom line is that the ats numbers merely determine if you bet & on which side.If a team's covered 7 times in a row & you think they'll cover an eigth because of the number that you've come up with,then you bet it.The streak should be irrelevent.
Not a great fan of streaks or trends.If you stick enough rules in(dogs on the road,scored 20+ points last week,have a winning record against conference rivals who wear blue uniforms etc)you're bound to find trends that are profitable.
What you don't know is if your rules are causative or exhaustive.
Also many trends use the initial trend to prove that the trend is continuing,when at the very least you should only count games played after you "discovered" the so called trend.
The really interesting point would be to know how betting money rides the trends.When do they start playing them,for how long & when do they get off & start betting the otherway.Of course if your fundamental handicapping of a game is any good then your numbers should be able to tell you this.
P.
Potter:
Very good stuff. I believe what you are saying is that if the original 100 flips comes up 30 for heads and 70 for tails, then in the long run it becomes a matter of rounding to the 1,000th decimal point (or whatever large sample size you use). I respect that opinion as being one coming from an intelligent point of view.
At the risk of sounding argumentative, I would like to add one thing. If the original 100 set of flips comes up 70/30, then the overall set of 1,000,000 flips could come up 700,000/300,000 but it never does, even though the same mechanism causing the 70/30 in 100 flips is used for the 1,000,000 flips.
This however is not practical toward our original question of ATS trends. As I mentioned before, this is more esoteric than should be further discussed here. If you would wish to add a contradictory viewpoint here I would welcome it but I no longer intend to go back and forth on this subject. I'll let you get the last word in. Sorry if I've muddied this up for anyone.
Very good stuff. I believe what you are saying is that if the original 100 flips comes up 30 for heads and 70 for tails, then in the long run it becomes a matter of rounding to the 1,000th decimal point (or whatever large sample size you use). I respect that opinion as being one coming from an intelligent point of view.
At the risk of sounding argumentative, I would like to add one thing. If the original 100 set of flips comes up 70/30, then the overall set of 1,000,000 flips could come up 700,000/300,000 but it never does, even though the same mechanism causing the 70/30 in 100 flips is used for the 1,000,000 flips.
This however is not practical toward our original question of ATS trends. As I mentioned before, this is more esoteric than should be further discussed here. If you would wish to add a contradictory viewpoint here I would welcome it but I no longer intend to go back and forth on this subject. I'll let you get the last word in. Sorry if I've muddied this up for anyone.
Hi BGO,
here's my take on how our little conundrum might shed a bit of light on ats betting.
Even with a perfectly fair coin,the longer you toss it the more likely it gets that you'll see long runs of seemimgly unusual results.For example dozens of consecutive heads,I think its called the law of large numbers.
Now if you deliberately search out these "unusual" runs,you can convince yourself that heads(or tails)is appearing more often than it should.But as we've seen these runs are normal & in the longterm don't prove an problem to the law of averages or the idea that one coin toss is independent of the next & so on.
You carn't infer anything about the near future from the near past.
Now taking a similar thought process into supposedly profitable ats betting trends.
First I fully accept ATX's(correct order this time
) point that gridiron games aren't independent of the last.Previous performance definitely affects future performance...But the ats numbers account for this & because the odds are generally around 10/11 the aim of the books is generally to get around a 50/50 split.
Now what so called profitable ats strategies are saying is that given certain rules a set of selections will "win" more than 52.4% of the time(this being the breakeven point at 10/11).
But imo you've got to be very careful that you're not selectively choosing your timescale so that you're keying into one of these segments that are comparable with a run of consecutive heads.
You're in effect "data mining" & coming up with profitable but spurious seams.
(By "you" I don't mean you,btw)
I'd be very wary if a profitable selection criteria started by saying something like"taking home dogs since 1999".What happend pre 1999?
Similarly if your selection process goes back to the year dot,but only selets 12 bets,even if 11 won,then you're into the problem that you mentioned of small sample size.
I'm not saying ats trends or trends in general are to be disregarded,just that they're fraught with danger.
An "out of sample" sample to test the theory is as I mentioned essential.
Given the choice I'd prefer to handicap the game first & only then(if at all) look at trends to confirm my judgement.
In general,streaks or trends help to provide profitable bets by sometimes forcing the books to skew the numbers that they set,but on their own I don't think trends provide a stand alone strategy.
Great disussion,it's helped me clarify alot of points...feel free to have the last word.
P.
here's my take on how our little conundrum might shed a bit of light on ats betting.
Even with a perfectly fair coin,the longer you toss it the more likely it gets that you'll see long runs of seemimgly unusual results.For example dozens of consecutive heads,I think its called the law of large numbers.
Now if you deliberately search out these "unusual" runs,you can convince yourself that heads(or tails)is appearing more often than it should.But as we've seen these runs are normal & in the longterm don't prove an problem to the law of averages or the idea that one coin toss is independent of the next & so on.
You carn't infer anything about the near future from the near past.
Now taking a similar thought process into supposedly profitable ats betting trends.
First I fully accept ATX's(correct order this time
) point that gridiron games aren't independent of the last.Previous performance definitely affects future performance...But the ats numbers account for this & because the odds are generally around 10/11 the aim of the books is generally to get around a 50/50 split.Now what so called profitable ats strategies are saying is that given certain rules a set of selections will "win" more than 52.4% of the time(this being the breakeven point at 10/11).
But imo you've got to be very careful that you're not selectively choosing your timescale so that you're keying into one of these segments that are comparable with a run of consecutive heads.
You're in effect "data mining" & coming up with profitable but spurious seams.
(By "you" I don't mean you,btw
)I'd be very wary if a profitable selection criteria started by saying something like"taking home dogs since 1999".What happend pre 1999?
Similarly if your selection process goes back to the year dot,but only selets 12 bets,even if 11 won,then you're into the problem that you mentioned of small sample size.
I'm not saying ats trends or trends in general are to be disregarded,just that they're fraught with danger.
An "out of sample" sample to test the theory is as I mentioned essential.
Given the choice I'd prefer to handicap the game first & only then(if at all) look at trends to confirm my judgement.
In general,streaks or trends help to provide profitable bets by sometimes forcing the books to skew the numbers that they set,but on their own I don't think trends provide a stand alone strategy.
Great disussion,it's helped me clarify alot of points...feel free to have the last word
.P.
potter, the last paragraph says it all........using trends ferret out inlfated lines seems to be the most useful application of these stats..........the "Law Of Large Numbers" is the concept at play here, unfort. it's REAL large #'s..............so , for all intents purposes, we're just jumping in midstream w/ a small sample to select from...........i do think the psychlogy at work, as far as BM'S shading lines based on public perception re: trends, is worth more consideration than blindly betting a trend .......fwiw
Hi. I use a system that bets pro football. When a team has won 3 against the spread I start betting against them. Lat year it hit around 58%. This year week 4 was 1-3 this week is 2-1 and we have Tampa tonight vs Indy.Next week we will bet against Dallas,stl,and minn and Miami.Week 7 will have Cinn to bet against and depending on indy tonight for next week.If Gbay wins next week they will be a bet against the following week nobody else has won 2 in a row lately. I stop after week 14.
I dont think it would work in anyother sport either 2 many games baseball and hoops or 2 spadradic football college. bobk
I dont think it would work in anyother sport either 2 many games baseball and hoops or 2 spadradic football college. bobk
IMO, systems like BobK's have merit. They are simple and are based on value. Generally, the more ATS wins a team has in a row the higher the price will climb. Now the factor which makes this system work in Pro Football is the nature of the game itself. It is a brutal sport with every game having infinite meaning (I am over-emphasizing the fact that every game counts). As the season wears on teams are more and more aware of other teams' strategies and tendencies and adapt. The only "trends" that I feel have much use are ones based on actual value--situations that re-occur year after year b/c the money always goes to the other side. There are quite a few of these, and it is easiest to explain with a common public example: NCAA basketball--unranked home team FAVORED over a top 25 opponent. This works almost every year b/c the public takes the ranked dog the vast majority of the time.
IMO, most trends are based on backfitting to past results. These would work if the things that happened in the past kept repeating, but unfortunately we have ever-changing scenarios. Plus, for almost every trend there is an opposite trend that conflicts if you look deep enough. Also, if a similar scenario has gone 20-0 over the last few years or months or weeks, I generally think that the opposite is <due> to occur, for lack of a better term. My line of thinking is that it is more probable for something to end up 20-1 than 21-0 (if the ML isnt -109328234092 to win 100 of course). I rarely bet against something just b/c a Feist trend should regress back to the mean, but it is something that I take into consideration.
IMO, most trends are based on backfitting to past results. These would work if the things that happened in the past kept repeating, but unfortunately we have ever-changing scenarios. Plus, for almost every trend there is an opposite trend that conflicts if you look deep enough. Also, if a similar scenario has gone 20-0 over the last few years or months or weeks, I generally think that the opposite is <due> to occur, for lack of a better term. My line of thinking is that it is more probable for something to end up 20-1 than 21-0 (if the ML isnt -109328234092 to win 100 of course). I rarely bet against something just b/c a Feist trend should regress back to the mean, but it is something that I take into consideration.
In my case I have a database of some 800 football teams. If one starts looking at how they perform against the Asian handicaps, one just knows that certain teams will perform much better than others. Simply mathematical flux. I look at it occasionally just out of curiosity. As far as I do believe that there is a certain public bias towards a team like the Dutch football team being overestimated, it shows in my predictions anyhow, so I really do not find any value in it. It is nice for guys that do have no fundamental knowledge of the game, but do like to have/lay a few bets to know which teams the public overestimates or underestimates. There are not many though!
For those that are into the NBA: Are the Lakers being overestimated?
For those that are into the NBA: Are the Lakers being overestimated?
ATX:
The trick in betting against a trend is to know when to start betting against it. Do you start after 10 games, after 15 games? In your example you mention a trend going 20-0. If you started betting against that trend at 10-0 you would have lost 10 times.
I'm going to have to buy Wong's book to find out what constitutes a trend. I'll report back here if I find anything of value.
The trick in betting against a trend is to know when to start betting against it. Do you start after 10 games, after 15 games? In your example you mention a trend going 20-0. If you started betting against that trend at 10-0 you would have lost 10 times.
I'm going to have to buy Wong's book to find out what constitutes a trend. I'll report back here if I find anything of value.
acw:
The Lakers won 3 championships in a row so no, they are not being overestimated. I do believe they became complacent last year which led to their losing.
This year coming up will show how well the Lakers can handle Kobe's legal problems. How much will it affect Kobe and the team? I would hope that this mess would be resolved quickly.
The Lakers won 3 championships in a row so no, they are not being overestimated. I do believe they became complacent last year which led to their losing.
This year coming up will show how well the Lakers can handle Kobe's legal problems. How much will it affect Kobe and the team? I would hope that this mess would be resolved quickly.
BGO
I think my main point is that I dont use trends. I look at statistics and a host of other things. Many times I try to figure out why the line is off from the dat and then I see that team x has beaten team y several times in a row. BINGO!! That is where the extra value is coming from. And many times I go larger on the wager not just as fading the trend but because there is extra VALUE present because of the trend. It's easier for me to give an example, and one off the top of my head is when Ponoson with BAL faced the NYY a while back. I'm not positive but I think that NYY had beaten BAL several times in a row. Ponson had better current numbers than who he was facing (maybe Pettite?) and BAL had a better batting avg currently and at the time NYY were not dominating on the road, BAL was playing .500 or better on the road. BAL was close to +200, I was saying to myself not only is this line way off, but I would take BAL at +100. I imagine the trend gave NYY at least 20 extra cents, maybe more, a lot of lines are way off until after the All Star break and that is one example where I can remember being really confused until I saw the trend. I think the final score was BAL 5 NYY 3 and the alt rl paid over +300 I think. This is besides the point but I played the vast majority of games small this year as I dont have a whole lot of baseball experience, but on my wagers of 1% or more I have been going over my logs and almost all of them are big dogs and the win % is sick, absolutely sick. Beginner's luck? I sure hope not.
I think my main point is that I dont use trends. I look at statistics and a host of other things. Many times I try to figure out why the line is off from the dat and then I see that team x has beaten team y several times in a row. BINGO!! That is where the extra value is coming from. And many times I go larger on the wager not just as fading the trend but because there is extra VALUE present because of the trend. It's easier for me to give an example, and one off the top of my head is when Ponoson with BAL faced the NYY a while back. I'm not positive but I think that NYY had beaten BAL several times in a row. Ponson had better current numbers than who he was facing (maybe Pettite?) and BAL had a better batting avg currently and at the time NYY were not dominating on the road, BAL was playing .500 or better on the road. BAL was close to +200, I was saying to myself not only is this line way off, but I would take BAL at +100. I imagine the trend gave NYY at least 20 extra cents, maybe more, a lot of lines are way off until after the All Star break and that is one example where I can remember being really confused until I saw the trend. I think the final score was BAL 5 NYY 3 and the alt rl paid over +300 I think. This is besides the point but I played the vast majority of games small this year as I dont have a whole lot of baseball experience, but on my wagers of 1% or more I have been going over my logs and almost all of them are big dogs and the win % is sick, absolutely sick. Beginner's luck? I sure hope not.
