Neural Net Lottery Picker

I've been using this site's prediction section to keep track of plays (Perceplotron based since the beginning of November).  I had my 2nd Pick 3 win (on paper) yesterday.  I've also had multiple 2/5 "wins" of LD lotto but those don't show up here for some reason, in real life 2/5 would have paid $1.   62% prize ratio on all games, 96% on P3 alone, not beating the house advantage, but pretty darn interesting. 

I haven't been consistent in how I use Perceplotron's number's, wheeling or not, how many games out to play them, etc.

Today,another 3/6.

Lately,I changed my old easy method,trying get best results.

Sadly,the results were not good.

Today I started the old method again,and I won a 3/6.

(I know ,3/6 is garbage for some poster,but is important for me).

This "keeps flushed the candle of my illusion".

Regards.

How many lines do you play in each draw?

I had another P3 win last night (on paper only). P4 isn't producing results.  I've had one 3/5 and multiple 2/5 hits (again just on paper) on LD lotto. 

Illinois Pick 3 only for this month (you can check this on my stats on the prediction board - filter by state and game). 

Thinking about possibly going live, but still working out a betting strategy playing fewer lines to start off with.

Hi MillionsWanted.

Only 8 bets every day.

Regards.

Today,TWO 3/6.

Its not a big profit,but I recovered my money for two days free!

Thanks again, Stoopendaal. Your programs are amazing.

Regards.

A very good day to everyone - Ladies and Gentlemens !!!

I hope your day is going GREAT !!!

For those who are getting good results from using the 2 Perceptron Softwares

I would very much like to know what "Settings" that you are using ...

Hope to hear from you soon !!!

Warm Regards as always !!!

Dr. David Bowman.

Hello Dr.Bowman:

(I published this post some days ago,but I paste it again for you)

Settings for NNLP:

I think every game it has it´s own settings,but anyway,this is my personal 6/49 settings:

Learning:0,174848

Noise:0,176

Threshold:0,053731

Neurons:77

Training data:27

The rest of settings: Not used.

I hope this helps!!!

Best regards

I use Perceplotron64_mcee . I'm still tinkering with settings. 

I think the learning rate works best at 2x the momentum value. Seed number doesn't seem to make much difference, I leave it at 1. Neurons 2 or 1.  F at 1 usually, but it seems to be worth trying other numbers.  I change the "use latest" value around,  MCE around .08, but this needs to be changed a little when "use latest" changes.

For anyone using NNLP on Viking Lotto:

I use these settings:

Learning Rate: E9-05
Momentum: N/A
Noise: 0,24
Threshold: 0.10003
Hidden Layer Neurons: 200
Training data range: 30 - 39 draws
Skip Training Data(Latest): 0
Fixed Seeds Input Weight: N/A.

I use the settings based on ordinary backtesting, not Skip Training Data.

Hi MillionsWanted:

I have tested,(on paper),your settings with my game,and the last two days,I have had 3/6 every day.

It looks very promising.

I´ll play the next days with your settings.

I hope to win some prices and publish my results here soon.

Thanks and regards.

Today,another 3/6.

MillionsWanted settings are very good!

Thanks to everyone who share his work.

Regards.

Congrats on your win.

Perhaps the settings are similar for similar lotteries which use the same kind of lottery drawing machines.

Viking Lotto is a 6/48 lottery.

Today, two 3/6.

(With MillionsWanted settings).

Regards

To Developer;

It would be cool if you can integrate a chart that can show Gradient descent marginal error.

Also, does any one know if I can guide my self by the column of Last Error? I wonder if this column shows what the last marginal error was thus to find the output?

Thank you.