Regression towards the mean.
I write this topic for my self and if you are interested, then you are more then welcome to contribute.
I been thinking about the STD and how you measuring things.
So one red is one event and one black is one event.
This mean i can measuring red and black using math.
I can measuring balance and imbalance.
This means i can also measuring loses and winnings in the same way.
Where one loss is one event and one win is one event.
And if we take the law of series, then one singles is one event and one series is one event.
50/50 situation as there is as many singles as there is series.
Different benchmark and different variance.
The bell curve has no limit as nothing is due to happen, but the bell curve tend to change after reaching 3.0 STD.
There is three main states to talk about and observe.
One is that the imbalance can continue to grow stronger or it can hovering around zero state or getting weaker.
The two last states is what you hope to catch after a strong imbalance, hovering state and opposite draw-downs.
But you can create your own state and benchmark with regression towards the mean.
A wave of overrepresented events can be 1.5 STD and you look for tendency of change and playing that the waves will change before reaching 3.0 STD, but only if there is present change that indicate hovering state or draw-downs.
They come in tiny, medium and large states.
You lower you set the imbalance of overrepresented events you more action you will get and in the same time higher variance.
With other words a more bumpy ride.
Mapping or clustering patterns.
Bayes have made software that match and miss-match patterns to get winning and losing sequences.
Where each winning and losing mark is one event and you can act upon that information.
Drazen once mention if i remember it correct 28 loses and 2 wins.
That is 4.74 STD.
But the beauty of the animal is that you would never place any bets during your observations as there is no tendency for present change.
Hovering state or draw-downs.
This made me thinking, so i come up with my own clustering march of the random flow where i set my own benchmark for regression towards the mean.
I pick a more bumpy ride.
The benefit of this methodology.
That is when imbalance grow stronger, you do nothing, so in that way you are never chasing for the opposite to show.
You just observe the flow grow until you get your indications of correction.
They come in tiny, medium and large states.
This is the true meaning with tendency play or you could also name it trending, but not based upon guessing.
This methodology is based upon probability and math.