United States Member #43923 July 27, 2006 5 Posts Offline

Posted: July 28, 2006, 2:42 pm - IP Logged

I have found that DSUM of 10 days ago for a pick 3 will contain most of the next 9 days of results. That trend seems to work up to day 1. (i.e. day nine contains 8,7,6,5...)

What if you took the DSUM for all 10 days and found the common numbers betwen all of them? I don't have a utility to find common numbers. I would be very interested to see the results.

United States Member #28945 December 25, 2005 1540 Posts Offline

Posted: July 29, 2006, 12:44 am - IP Logged

Quote: Originally posted by SkyHawg on July 28, 2006

I have found that DSUM of 10 days ago for a pick 3 will contain most of the next 9 days of results. That trend seems to work up to day 1. (i.e. day nine contains 8,7,6,5...)

What if you took the DSUM for all 10 days and found the common numbers betwen all of them? I don't have a utility to find common numbers. I would be very interested to see the results.

Hi SkyHawg. Interesting observation you made. Worth checking out. Thanks.

Switching between Fairfax, VA and Belgium Belgium Member #19287 July 29, 2005 2254 Posts Offline

Posted: July 29, 2006, 5:18 am - IP Logged

Quote: Originally posted by SkyHawg on July 28, 2006

I have found that DSUM of 10 days ago for a pick 3 will contain most of the next 9 days of results. That trend seems to work up to day 1. (i.e. day nine contains 8,7,6,5...)

What if you took the DSUM for all 10 days and found the common numbers betwen all of them? I don't have a utility to find common numbers. I would be very interested to see the results.

Hi,

you asked pretty much the same question in this thread: http://www.lotterypost.com/thread/139104

There is an excellent utility here at LP to check for common numbers, the Deflate 3 number reduction system. You can paste the numbers there, press "Deflate numbers"-button and it will tell you what numbers;

I tested GA mid, draw 740 is last draw:

Results: 180 (that means 180 unique numbers, boxed)

Switching between Fairfax, VA and Belgium Belgium Member #19287 July 29, 2005 2254 Posts Offline

Posted: July 29, 2006, 10:03 am - IP Logged

I did a small test on GA mid:

Used: 10 draws

Last draw

Next draw

Ranking of boxed combo

Rank straight number

827

166

171 (17 times)

674 (5 times)

166

592

60 ( 33 times)

273 (7 times)

592

331

198 (15 times)

647 (5 times)

331

628

102 (28 times)

729 (4 times)

628

442

202 (14 times)

460 (6 times)

442

647

84 (30 times)

707 (4 times)

647

235

79 (33 times)

812 (4 times)

235

294

59 (35 times)

645 (5 times)

294

978

78 (34 times)

711 (4 times)

This is ofcourse just 1 example, and with 10 draws used. It is quite some work to achieve lol

But, there is something about the straights.... playing those that have 4 or 5 occurancies, at $1 for the example, costs from $300 to $400. On these 10 tests it would have cost between $3000 and $4000 and payback would have been $7200.

More testing would be needed for this.

Somehow you would need to cut down the plays, using only the combo's with at least 1 digit from the last draw, or from the currently dominant 3-digit serie (see pick3-forum, threads by Win D)... Surely that would sometimes throw out the winning number...

Switching between Fairfax, VA and Belgium Belgium Member #19287 July 29, 2005 2254 Posts Offline

Posted: July 29, 2006, 10:40 am - IP Logged

Okay, i redid the test

Used: 10 draws: created only combo's that had at least 1 digit from the previous draw

Last draw

Next draw

Ranking of boxed combo

Rank straight number

827

166

No hit

No hit

166

592

No hit

No hit

592

331

No hit

No hit

331

628

No hit

No hit

628

442

128 (14 times)

299 (6 times)

442

647

45 (30 times)

333 (4 times) --> 142 combos with 4 or 5 occurancies

647

235

No hit

No hit

235

294

36 (35 times)

421 (5 times) --> 237 combos with 4 or 5 occurancies

294

978

57 (34 times)

477 (4 times) --> 224 combos with 4 or 5 occurancies

More testing is needed here, but this looks quite promising!! Even when considered that 5 out of the 10 draws in this example were without a repeating digit.

Now, with the dominant 3-digit series. I used not the longest running serie, but the serie with the highest Digit-Hit-percentage. If such a serie ends, then you lose for sure.

Used: 10 draws: using 3 digits from serie

Last draw

Next draw

Ranking of boxed combo

Rank straight number

827

166 : serie 1, 3, 7

113 (17 times)

446 (5 times) : 238 combos

166

592 : serie 2, 3, 6

40 ( 33 times)

158 (7 times) : 241 combos

592

331: serie 1, 2, 9

126 (15 times)

400 (5 times) : 230 combos

331

628 : serie 2, 3, 6

70 (28 times)

467 (4 times) : 184 combos

628

442 : serie 2, 3, 6

129 (14 times)

284 (6 times) : 193 combos

442

647 : serie 2, 3, 6

59 (30 times)

443 (4 times) : 183 combos

647

235 : serie 2, 3, 4

55 (33 times)

525 (4 times) : 212 combos

235

294 : serie 2, 3, 4

39 (35 times)

417 (5 times) : 231 combos

294

978 : serie 2, 3, 4

230 combos

Okay, what does all this mean: i only "played" the numbers, straght, that occur 4 or 5 times in the list created with DSUM, using 10 draws. The amount of combos this are is listed under "Rank straight number" Total draws played: 10 Total cost: $1942 Total hits: 6 Payout: $5400 Result: $3458

United States Member #43923 July 27, 2006 5 Posts Offline

Posted: July 31, 2006, 4:17 am - IP Logged

I tested three methods.

1. I produced DSUMs for 10 numbers. Then I took the DSUM from 10 deflate 3 to 9. took out the numbers that only occured in one but not another. Then 9 to 8, 8 to 7... 2 to 1. Everytime it produced about 200 results.

2. I took DSUM 10 and sampled several days and compared them to day 10.

3. I took all 10 days put them in Deflate 3 and compared.

What I found out is that method 1 produced the best option to produce future numbers and contained most of the numbers from the past.

Method 2 produced unreliable results.

Method 3 didn't sort out hardly any numbers because a number is 10 might be in 3 and so it wouldn't be sorted.