For this overview we'll be looking at the last digit of the Minnesota Daily Pick 3 in the date range of 2012-01-01 to 2012-07-06; a total of 188 draws.

Most of us have had that feeling there must be some pattern or patterns in the chaos of the lottery numbers we analyze.

Just by looking at the chart of the data we can see something of a pattern there.

There are a wide variety of methods that have been created to find these obvious but elusive patterns.

This will touch on a method that seeks to quantify both the order and chaos in the random selection.

From the chart below we can see the last digit of the Minnesota Daily Pick 3 for a sample of 188 draws.




In the chart we can see there appears to be gaps and clusters of data which would tend to suggest some kind of pattern, order.

It also shows the placement of these gaps and clusters are not regularly oriented in the chart, chaos.

We know there's something of a pattern, but how do we find the pattern for the chaos and use it to project into the next draw.

This is where the Wave Matrix comes in.

The Wave Matrix has three parts: a Regression, a Set of Waves, and a Remainder.

To create the Wave Matrix, we first find the regression component.

In this case, it's a mostly horizontal line that is at about the level of the average ball number of 4.5.




Other data sets will have different regressions like: linear, power, exponential, polynomial, etc.

The basic idea behind the regression is to setup a data set that oscillates about 0 on the x-axis.

Sometimes positive, sometimes negative, but usually clustered about the x-axis.

We get the oscillation set by subtracting the linear regression from the original data set, shown below.




Using the oscillation set and a process know as an Iterative Bidirectional Mean Averaging, we can find a set of waves within the oscillation set.

The Bidirectional Mean Averaging is a function that is similar to a moving average, but has a weighted natural log exponent (e^x) or degree of weighting and is run forward in time and backward in time to arrive at an average present point value.

The Iterative part the process is achieved by feeding the data out of the BMA function back into the function itself over a number of iterations, in this case it was iterated 8 times.

The Iterative process analyzes the data out after 8 times and refines the degree of weighting to optimize the data out.

Refinement is based on comparison of the frequency of the data out, the frequency of the differential of the data out and the frequency of the second differential of the data out; in addition to a proportionate value of the Root Square Mean of the data in and the data out.

The first wave is derived using the Iterative BMA process and then is subtracted from the oscillation set to get a new oscillation set.

The second wave is derived from this new oscillation set using the Iterative BMA and then subtracted for another new oscillation set.

This process is repeated recursively until we arrive at the desired sets of waves, in this case 8 waves, in the following chart.




If we add up the linear data and waves data, we get a plot that follows the data in a fairly general pattern.




It does not match the data exactly because though each iterative process there is always a remainder.




When we add in the remainder, we get the original data set.




It's reasonable to assume, if we have the linear, waves and remainder, we should be able to project the linear, waves and remainder into the next draw.

The linear is no problem, because it's based on a simple equation of y = a + b · x; where x can be the next draw.

The waves are a little bit more tricky.

How do you project a wave for just the next draw based on the current data of the wave itself?

Below we can see the first wave for the last 50 draws in our sample set.




From this we can visualize a set of possible projections.




We might get something reasonable, but how can we quantify a result that is a little more useful and standardized for all waves?

This is where the Ultimate Wave Predictor comes in.

It quantifies a result based on the wave itself and comes up with a reasonable projection of the wave into just the next draw, seen here.




When we apply the Ultimate Wave Predictor to the remaining waves and add in the linear value of projection we get the approximate value of the next draw.




This leave the Remainder and what to do about it.

Looking at the chart we can see it looks like it would fit very nicely with some kind of Gaussian Distribution.




We could work up a way of generating a random value from that and use it to complete the Wave Matrix Projection.

We could try to simulate the distribution of the remainder to mimic the lottery's randomness.

But, we realized the randomness was sitting right in front of us the whole time.

Why try to mimic the randomness of the Remainder when we can use the values of the Remainder itself by simply randomly selecting one of the Remainder values.

By doing this we achieve two things: one, it's a random value for completing the Wave Matrix Projection and two, it constrains the distribution of random selection to the actual distribution of the remainder.

The remainder is the actual chaos that is generated by the lottery itself.

After all, if the lottery says it's virtually the greatest thing since sliced bread, we can use that to our advantage and make one heell of a winning sandwich.