Dr. Ed Meyer said:
However, there is other information that is not factored into the line and thus can be used to beat the NBA. A wealth of such information can be found by handicapping the performance of individual players in addition to handicapping the team as a whole:
Love the SDQL and use it often. One thing I am unclear on is that since a statistically significant trend has a 5% probability of being caused by no more than chance variation, is there a way to differentiate between the signal and the noise, so to speak, in these trends?
Thank you.
[/quote]
Mr Game, Thanks for posting this important question. Please forgive me if I aim the response to the general audience, which I realize is below your level of sophistication.
If I understand your question, you are referring to the fact that a two-sigma standard deviation has a 5% probability of being caused by a random fluctuation.
In general, the "laws of statistics" do not apply professional sports handicapping. Basketball games are not like flipping coins. There is a LOT more involved. When flipping coins, the results are independent. No two basketball games are the same. The players have lives, they have emotions, they have ups and downs.
Let's look at a specific example of a trend in past performance:
The Bulls are 0-6 OU (-18.33 ppg) at home after a road loss in which Derrick Rose had at least 5 turnovers.
A statistician would say that this trend is not significant because there have only been six trials (If you flipped a coin six times and it came up heads every time, you should not conclude that it is a two-headed coin). However, I like this trend. It makes good handicapping sense and it has a terrific margin. If you look at the game listing, you will see that the only game that did not stay under by double-digits was an overtime game.
In contrast, there are a lot of, say, 10-0 trends that are useless in forecasting the future. The secret to differentiating the signal from the noise will not be found by statisticians, but by good handicappers. Sports handicapping is not finding trends, it's separating the trends that merely define the past from those that forecast future results.
So, when I evaluate trends and systems, the questions I ask in an attempt to separate the signal from the noise are:
1. Does it make good handicapping sense?
2. It is timely?
3. Is there a strong average margin?
4. Do the reasons for the trends in past performance still exist?
Further, I always look at the game listing for every trend/system I use. This makes it easy to focus on recent results.
So, there is no easy formula or rule of thumb for separating the wheat from the chaff -- just experience and hard work.
I can't wait to tackle Friday's NBA card!
Dr M.
[/quote]
Dr.M,
Thank you for the detailed response and you understood exactly what I was asking. Your response was very helpful and made some things clearer to me.
Thank you again.