Because something I have been thinking about and finally checked into is hitting around 60% this season. Is there any easy way to check this? This has taken a long time to get to this point... How far back should I go before I decide whether or not this works? PS, I hate systems... this is somewhat of a system but has less to do with stats than most. I won't get into it now, but if I could get some answers I'd appreciate it.
I would say you would need at least 500 samples/trials to even begin to take the information into consideration. The system/method I currently use always starts to come into form around 500 plays for the month but if you asked a statstician I would assume he would go even higher than that.
here maybe this article about testing systems will help you
http://www.sportsinsights.com/sportsbettingarticles/si/statisticalsignificance.aspx
http://www.sportsinsights.com/sportsbettingarticles/si/statisticalsignificance.aspx
I totally believe in systems and systematic betting. The problem is people wouldnt be satusfied unless the system picks a high pct, give me 51.5% versus widely availbile lines, let me throw in line shopping thru hard work and I can promise you a true money maker.
Like the article said 2,000 plays is a must for true significance.
Like the article said 2,000 plays is a must for true significance.
The information you used to come up with the system can't be counted to confirm a system works. Now that you have come up with a system based on certain data (say the last two years), you have to confirm the data used to come up with the system isn't a statistical anomaly. There are 1240 games in an NBA season, are you really going to trust something that is based on less than half a season?
I would say you need at least 4-5 seasons worth of back checking
I would say you need at least 4-5 seasons worth of back checking
Any statistical sample of 600+ should get you within +/- 5 percentage points of the "most likely" outcome.
Get it up to 1500+ and your margin of error will shrink to under 3%
That's presuming each entry has a decent common thread of comparison with all the others.
Most common place most of us see this applied is in a truly acturial based poll.
600 "people randomly selected from a specific geographic neighorhood" can give you good indicators - within about 5% +/- on a given question.
But it increases in value the more common components the 600 people share - age, race, gender, education level etc.
In the case of a Sports Capping System, you'd want each entry to the "baseline" to have as many common elements as possible - ie, Home/Away, Fave/Dog, Top10Defense/Bottom10 Offense are just a few of the many possible comparative points.
Get it up to 1500+ and your margin of error will shrink to under 3%
That's presuming each entry has a decent common thread of comparison with all the others.
Most common place most of us see this applied is in a truly acturial based poll.
600 "people randomly selected from a specific geographic neighorhood" can give you good indicators - within about 5% +/- on a given question.
But it increases in value the more common components the 600 people share - age, race, gender, education level etc.
In the case of a Sports Capping System, you'd want each entry to the "baseline" to have as many common elements as possible - ie, Home/Away, Fave/Dog, Top10Defense/Bottom10 Offense are just a few of the many possible comparative points.
I agree with this post to the point of saying that yes - a sample of at least 600 is most helpful. And in a single NBA season, it may be difficult to find 600 of ANY comparative sample.
Good news is that the NBA leaguewide results and most notably - scoring averages - have been pretty consistent going back about 5 years now so you can build a variety of decent comparative samples.
I will note that this year's scoring average of 99.0 ppg is actually a teeny bit higher than past four to five years where it's sat between 98 and 98.5
The years of about 2000-2001 through about 2002-2003 had scoring averages closer to 96.0
Great stuff barman. Thanks for the numbers.
Also thanks to everyone who contributed in this thread. I'll keep tracking it and see how it does but quite frankly, it's going to be too much of a hassle to go back any further. Seeing as my normal NBA picks are down 10 units, I'm willing to try this out and hope for the best.
Also thanks to everyone who contributed in this thread. I'll keep tracking it and see how it does but quite frankly, it's going to be too much of a hassle to go back any further. Seeing as my normal NBA picks are down 10 units, I'm willing to try this out and hope for the best.
