Hi,
Here some screen shots of the new ANN application I'm working on. This application, 'Perceplotron' uses number statistics as input to feed the Neural Network algorithm. For each number a ranking value (probability value) is calculated (as output of the ANN tool) so you are now able to select the top x ranking numbers as output.
Here some test results of the tool.
Settings:
- Lottery game 6/45
- 990 results loaded
- Learning Rate 0.90
- Momentum 0.24
- Epochs (Learning Cycles) 1200
- Factor F 6 (used in creating the input statistics)
- Use latest 10 draw results (yo see here 15 results selected because I want to back test the latest 5 draw result (skipped results)
- As output I want the top 10 numbers (numbers with the highest probability according to the ANN )
Okay, now some back testing with these ANN settings (page back testing).
Here I back test the ANN with the latest 100 draw results of my lottery game (pick 6/45). I also want to see the performance of the ANN output with the next 3draw results (Show Max Match Next = 3 Draw Results).
When back testing is done you can visualize the results (Page 'Chart').
Here you see the top 10 ranking numbers (for each ANN prediction) gave a match of 3 or higher (15*3 and 4*4) 19 from the 100 draw results!
- Theoretical odds winning numbers matched 3 (10 numbers pick 6/45): Odds are 1 in 10.37 (ANN test results 1 in 20)
- Theoretical odds winning numbers matched 4 : Odds are 1 in 65.19 (ANN test results 1 in 25)
Results for the next 3 draw results:
30 of the 100 draw results the 10 output numbers gave a match of 3 or higher with the winning numbers.
Here test results for number position 3 of the 6/45 lottery game.
27 of the 100 draw results 1 out of the 5 top ranking numbers (see settings) had a match for position 3!
Results next 3 draw results (60 out of 100 draw results one out of the top 5 numbers was correct)
Next weekend I'll make this tool available for download.
Cheers!