Lucky Games for 3/17 & 3/18

With today being the unofficial day of all things lucky, I thought it the perfect time to debut a new feature in LuckSmith's prediction engine that we'll be releasing formally to all members in the next week.

Today's Lucky Games

LuckSmith's stores the historical accuracy of its predictions going back hundreds of drawings, and even thousands in some cases, for each lottery game we track. This means we now have an effective way to determine how accurate a particular PowerSet will be for an upcoming drawing. We do this by looking historically how all PowerSets of the same size performed, and averaging those scores.

Each day, we'll analyze what games look to have the greatest potential PowerSet accuracy, based on past performance, and post those in the Members area of the LuckSmith site. Additionally, we'll post the games with PowerSets that look like they won't be as accurate, so you can know which ones to avoid.

For this weekend, here are the Lucky Games from the LuckSmith library, and the stats for each:

Game Name	Average Matches	Accuracy	PowerSet Size	Draw Date	Weight
Colorado Cash 5	5	100	20	3/17/2006	1
New Jersey Cash 5	5.13	85.41	41	3/20/2006	8
Texas Lotto	4.16	83.33	35	3/18/2006	6
Florida MegaMoney	3.33	83.33	34	3/17/2006	6
CA Super Lotto Plus	4.17	83.33	37	3/18/2006	6
Texas Cash 5	4.15	83	31	3/17/2006	20
PA Match 6	4.9	81.67	42	3/17/2006	10
Canada 6/49	4.82	80.30	37	3/18/2006	11

Past Powerball trend continues

Tonight's Powerball winning numbers were:

1-5-7-21-43 Powerball: 20 Multiplier: 5

The initial 44-number Powerset generated for tonight's drawing contained 4 out of the 5 winning numbers (as the historical data indicated it likely would). More importantly, however, the improved Powerset, (reduced in size to 38 numbers), also contained 4 of the 5 winning numbers, which further improves odds.

We'll post another update for the Powerset for Saturday's Powerball drawing, including additional numbers that the new prediction engine predicts won't be drawn. If the performance continues to be solid, (we expect it will), we'll eventually replace the current prediction engine with the new one for most of the games we track.

Good luck to all!

Smaller PowerSet for Tonight's Powerball drawing

We are testing new prediction features that reduce the PowerSet size another 10-15%. Before we release those features, however, we thougt it might be valuable to let members have a "sneak peak" of sorts at the improved PowerSet recommendations.

For the 3/8/2006 drawing (Wedensday), the existing PowerSet contains 44 numbers. Looking at the performance history for a PowerSet of this size, the average score is 4.2 out of 5, so you could expect to see 4 out of 5 numbers appear in the PowerSet tonight.

The new PowerSet prediction system created a set for tonight''s drawings that has only 38 numbers, a difference of about 15%.

The 6 numbers that the new PowerSet has removed are:
11 30 38 39 41 53

What this means is that LucKSmith''s predictions indicate these numbers should not appear in tonight''s drawing. If you received any picks with these numbers in them, you might consider choosing different numbers. (You could also play the picks you were sent, and just watch how the new smaller PowerSet performs).

Check back tomorrow for an update on how well the new prediction system did with the smaller PowerSet.

Members hit 4/5 in NY Take 5, MegaMillions

Tonight, two LuckSmith members had PowerPicks matching 4 out of the 5 winning numbers in New York's Take 5, and the MegaMillions multi-state game.

For NY Take 5, the predicted PowerSet for 3/7/2006 had 30 numbers. All 5 winning numbers were in it, and one LuckSmith member had a 4/5 winning PowerPick. 5 PowerPicks matched 3/5 numbers, and 17 matched 2/5. LuckSmith generated a total of 150 PowerPicks for the 3/7/2006 drawing.

In MegaMillions, 1 member hit 4/5 without the bonus, and another hit 3/5 without the bonus. (The PowerSet for 3/7/2006 contained all 5 winning numbers, but missed the bonus number.) 521 picks were generated for tonight's drawing. Studying the the MegaMillions number matrix, basic probability says the most expected hits from a total of 521 picks would be a 2+1 winner, or 2 matches plus matching the bonus number. We had 47 PowerPicks matching 2/5 numbers, and although this doesn't win any cash prize, it shows our system clearly bests the standard playing odds. (See LuckSmith's prediction stats for more details.) If we'd only hit the Money Ball too!

We're frustrated with the accuracy recently on MegaMillions, and are recalibrating the PowerSet engine to get back into the 70-80% range we like to see.

We're hoping our Powerset for tomorrow night's Powerball drawing does well. It contains 44 of the 55 possible numbers, which nets 4/5 on average, but we're due to match all 5 again here soon. The Bonus Powerball set contains 27 of the 42 possible numbers, which gives it a good chance of hitting that as well.

Congratulations to the lucky winners, and good luck to everyone in tomorrow's drawings!

Pick-3 and Pick-4 games almost ready

I'm working on the prediction and scoring engines to make them compatible and functional with some test Pick-3 and Pick-4 games. The plan will be to release these features as a test or Beta program, so that users can evaluate how well it works and provide feedback on what they'd like to see. I wanted to have it all ready my March 1, but there are still some kinks.

Our first games will likely be in Pennsylvania (Daily Number and Pick 4), and Texas (Pick 3). California and New York will be close behind. We've actually been capturing data for these, and many other games, for some time. It's just taken time to configure the prediction engine to work with a range of only 10 possible numbers, rather than 35+. I think it's pretty much nailed. (In my tests, I hit straights in PA and TX games on the same night!)

Watch my predictions on Lottery Post (WannaWinSoon) for some test picks.