One: The Complexity of Lottery Programming
(from the original post to rec.gambling.lottery)
I am truly sorry for misunderstanding. I have never had the intention to make any process difficult and time consuming! What I have said is radically different. The lottery is such a difficult task that I haven't been able to find an easy way to grasp it! I started with a strict scientific approach: if it doesn't have a formula is non-sense. At least, if I cannot find a formula for something, that thing makes no sense to me and I leave it alone. I break down the huge-odd lottery into parameters (or patterns, or filters, or eliminating conditions) with lower (more manageable) odds. Then I combine those smaller parameters into what I call "strategies". Now, it could be possible to let the computer do everything automatically, including the selection of the best strategies.
Well, there is no substitute for the human brain. I have come up with a large number of patterns or filters, as you can see in my SuperPower$ software (the WS files). It would be a huge task, unmanageable for a single programmer, to write a program that would check all the parameters, combine them in all possible strategies, and selecte the most likely to win strategies. When I started, it was the world of 16-bit DOS, with the filesize limit of 640KB. I would have needed many, many megabytes.
Now, it is 32-bit Win95/98, but it is very, very slow compared to DOS (in the case of lottery software, which is highly CPU-intensive). I have had no choice (and not much time) but let the computer show me the reports of the parameters (i.e. how the filters fared in the past); based on the past values of the parameters, I would select future values using some statistical methods. One such simple method anyone can notice is that 3 consecutive increases are immediately followed by a decrease most of the time (over 90% of the time) -- and vice-versa. Of course, 90% is not 100%, that's why it is impossible to select winning strategies
every time. I advised in my SuperPower$ package a way to play the pick- 3 game. Look at the front/back pairs parameter. It has a median of 33. It also shows values of over 100 -- it means none of the front or the back pairs of a pick-3 drawing is a repeat from the past 100 drawings. Using such a filter, would reduce the number of total possible combinations to 125 (or a probability of 1 in 8). So, one strategy would be based on the "pairs" (Pr) parameter with a value of 100. Other parameters in the same file would have also higher values. At run time, in addition to setting PR to 100) select higher but safer values for other parameters and also use the simple rule of 3 increases/decreases. Indeed, it takes some effort, but I HAVEN'T BEEN ABLE TO FIND AN EASIER WAY, ALTHOUGH I WISH I HAVE. It is not may intention to make it harder and painful. On the contrary, I have tried to make it as easy as possible. Remember, I am the most frequent user of my software and I am not masochistic. I dislike sufferring and pain. Again, I rely on my eyes and brain (that is, effort required) because I am unable to write a computer program that would do what I do mentally. Maybe in the future...
Two: Original Pick-3 Strategies
(from the original post to rec.gambling.lottery)
• Speaking-Of-Devil Department:
Soon after I posted my last message, I had a moment and I checked Pennsylvania lottery results. I ran the new Pick-3 software that I presented in my previous message. Following is the top of one of the report files. The results comprise the last drawing as well: 7-1-6, July 20, 1999.
The column Pr is of interest now, since I referred to a strategy based on it in my posts. As you can see, PR1 is over 100, therefore it satisfied the main condition of our strategy. Other filters also have higher than normal values (as specified in my strategy selection reasoning).
• Let's say the day before the drawing took place, I wanted to see if the moment was right to play a strategy based on the filter Pr1=100. I would run my program that checks any strategy. The number 5000 is a dummy: it represents a maximum value which is never reached. If I don't want to check for maximum values, the program will use that default value of 5000.
You can read the strategy as: we are only interested to see how many times Pr1 was 100 or more, regardless of any other filters.
• Our strategy occurred 53 times in the last 400 drawings (let's approximate it to one year). The Skips section shows how many times the strategy waited between hits. Zero means that the strategy hit in two consecutive drawings.
• Median Skip: 5 drawings
• Skips of 0 to 5 = 32 times (for a total of only 82 skipped drawings!)
• over 10: 10 times (for a total of 178 skipped drawings!)
The worst playing situations occur when the skips reach 10 or more drawings. That would be most costly way to play the strategy. Based on my Gambling Formula, a skip can reach even 55 drawings! That was one of my sins up until not long ago: I believed that it was best to play only strategies with high skips (over 20)...
It is better to play lower skips, around or below the median (a requirement of the Gambling Formula).
• Playing only when "Skips" shows 4 or 5 as the first number:
- playing 21 times @ $150/play = $3150
- number of wins: 14 x $500 = $7000
- net profit: $3850/year (400 drawings)
That profit comes WITHOUT applying any filter to Pr=100!
We want to play with more modest budgets, so we apply some other filters. As I stated before, when a filter such Pr is high, other filters in the same WS file have a tendency to reach higher values. We can set other filters conservatively (i.e. "safe" values). For example:
Tot1=200, Rev1=200, Vr1=1, Syn1=100, Bun1=1, Any1=100.
• Applying other filters in addition to Pr1=100:
- playing 21 times @ $50/play = $1050
- net profit: $5950/year (400 draws)
Actually, just using Pr=100 without other filters generates 130-150 combinations. Using additional filters generates from 10 to 45 combinations.
I think it is a lot more convincing now that even the random events follow rules and formulae; that they can be analyzed scientifically.
Nevertheless, I still wish you all good luck!
•• I received an insightful message from a Brazilian user. He found my strategy to be FLAWED. After further review, he was right! I rechecked the entire process. I miscalculated. Indeed, I should have counted also the times when the skip went 6 or higher. Actually, I looked at front skips which were already higher than 6. The right strategy is still based on the MEDIAN.
Briefly now, if there are more than 5 skips below the median do not play. Also, after 1 or 2 skips above the median, there will follow skips lower than the median -- play the strategy!
Or, play ALL the streaks, but only from 0 to 5. There are 53 streaks.
Playing only from 0 to 5 drawings means 6 runs. Multiplying 53 streaks by 6 runs gives 318 runs. There are 32 winning situations for streaks between 0 and 5. Total winning is $500 x 32 = $16,000. In order to make a profit, we must play no more that 50 combinations per run. In other words, if using additional filters to PR=100 generates 10 combinations to play, the cost will be 10 x 318 = $3,180. If the strategy generates 20 combinations per run, total cost would be 20 x 318 = 6,360. Just 20 combinations per run is pretty common, given the strong co-variance of the filters in the same WS file (the same layer). There is a stronger still co-variance between PR and TV, or between TOT and REV, etc.
Always use other filters in addition to the main filter (PR, in this case). I have received quite a bit of input on the subject of strategies applied to my software. Some users with computer programming skills have written their own utilities to check the values for other filters. Others use the Excel spreadsheet to do comprehensive analyses. Basically, a user opens as a text file the WS reports generated by SuperPower$. If they want to apply the PR the main filter, they will sort the spreadsheet on the PR column. Next, select all the cells where PR is 100 or greater. Paste the block in to a new spreadsheet. Do a data analysis of the new worksheet: get the minimum, maximum, average, median, mode for every column. You'll notice that PR>=100 is accompanied by higher values in other columns as well. Applying other filters in conjunction with PR will substantially reduce the combinations to play (under 10 in quite a few
cases). I am adding now these features to my strategy checking utilities. The lottery is such a deep phenomenon scientifically! I didn't know what I was getting into! I only wanted a means to sponsor my highest goal: the "humans-computing-beasts" opus.
•• Playing the lotto games with SuperPower$:- the best way is to wait for very high filter values. Theory of probability is, from a different point of view, the science of streaks. The streaks can vary from very long to very short, from very low to very high, from loose to tight. It is unrealistic to expect winning every lotto drawings. Therefore, the best strategy is to wait for
those advantageous filter values. Such values, although rare, do come out. It is a law of probability. When applied, such filter values generate very few combinations to play.
And such filters can be a gold mine, literally.
While developing the lottery software, I was met with astonishing situations. In Pennsylvania lotto 6/48 game some filters were very high from time to time, as it normally happens. In one case, however, they were so high that the software generated only ONE combination to play!
In another case, in the old Maryland 6/40 game, the filters were as such that only 8 combinations were generated. Cases like those made me stay in this lottery software business. It is possible, albeit very rarely, that a huge-odds lotto game can be won with just a handful of combinations!
Three: More On Gambling Formula
(from the original post to rec.gambling.lottery)
• Shobbolun, that's a very good book, the one from which you picked those formulas: "Lady Chance" by Warren Weaver. The author, however, totally refuses to believe that we can improve our odds at gambling, regardless how much we study and how well we understand the phenomenon. The moral issue was very strong at the time he wrote his book: gambling was really stigmatized in the U.S. (60s). My approach is different and I have had proof that chances can be improved.
I presented in this post just one of the strategies that can be used in the Pick-3 game (using PR = 100). Anybody can check it in any pick-3 lottery. I presented a report for 400 drawings, but I have also checked 1,000 and 2,000 past drawings. (There is now a better HTML report at this address: Pick-3 Lottery Method, Strategy, System, Play, Software).
I also presented on one of my Web pages (Lottery, Lotto, Software, Systems)
another strategy. I presented real data for a 6/48 lotto game.
The individual probability is p=1/8. According to the Fundamental Formula of Gambling (FFG), half of the lotto6/48 numbers (3) in every drawing will be repeats from the most recent 6 drawings. Moreover, there were drawings in which all six winning numbers had been
repeats from the past 11-12 drawings, even from the past 6-7 drawings! In the last 500 drawings of the late PA Wild Card lotto game (6 of 48), in 123 cases (25%) all 6 winning numbers were repeats from the last 12 drawings. In 25 cases (5 times a year), all 6 winners were also drawn in the last 7 drawings. More amazingly, in 7 cases, all 6 winning numbers
were repeats from the last 4-5 drawings! Again, it can be checked against any lotto database. I believe this is proof that using knowledge (real formulas) can improve player's chances in gambling.
• • There are people who simply do not believe that my previous statement can be true, ever. Gambling is random, and random means a phenomenon that cannot be forecast. Winning chances stay always the same, they can never be improved. Such people stay away from any type of gambling and they are all right with me. But there is something I don't understand.
There are players, like in this forum, who fight against any idea that gambling can be analyzed with the purpose to improve player's chances. Why do they participate here? They appear to gamble or, at least, be interested in gambling. They may hope to win. They are here because they hope to find winning strategies, or just tips. But their hope seems to take the form of a placebo. They want the strategies to just be very easy. Easy means no mathematics at all, no difficult analyses at all. Thorough analyses just show how difficult gambling can be to grasp; mathematics will just destroy the placebo effect. They only want somebody tell them:
"These are cold numbers, these are hot numbers, you should play these 10 combinations!" Nice and easy! But don't you ask them "How much did you spend and how much did you win
over the past year, or two years?" The question destroys the placebo effect.• • • And you know what? That's the most common reaction in humans. We all rely on all types of placebos our entire lives. We always consider first things that come nice and easy. Nobody likes to go immediately after difficult things, phenomena requiring extra effort.
Best of luck to you all, believers and anti-believers in strategies like mine! As of me, I will always choose bitter-sour knowledge over sweet placebos.
Four: Using Spreadsheets to Analyze The Lottery I
(from the original post to rec.gambling.lottery)
In article email@example.com,
"Jack Handy" wrote:
" I'm writing a little thing in Access and was wondering if anyone else had done this and would like to share their query's because its mostly new to me and I'm still a little fuzzy on how I'm going to do what I want to do. Like out of a pick 4 lottery with repeats, searching for the first 3 in any order. stuff like that.
SQL, databases in general, are not suited for the type of statistical analysis you intend to do. They only have basic statistical functions. You will need to do a lot of coding, writing your own functions. You would probably need to use a highly sophisticated programming tool such as Visual Basic, included in a database application such as Access.
I understand you are not into programming. In this case, your best tool of choice is Excel. The spreadsheet application has a large variety of powerful statistical functions. You need to install the add-in called "Data Analysis". Once installed, it will show up in the Tools menu.
Here is what type of analysis I just did for my lotto-5 data file. The game draws 5 numbers from a field of 39. If it is for the first time you've seen my name, I recommend you visit my lottery site first: saliu.com/LottoWin.htm.
The lotto-5 data file is a simple ASCII (text) file with EXACTLY 5 numbers per line:
1 12 20 28 35
It is also assumed that your drawings are sorted in ascending order. You shouldn't use the drawings as they appeared during the TV drawing show. In Excel, open File menu. Select Open. Choose "Files of type: All (*.*)". Select DATA-5 (or the correct name of your lotto-5 drawings file). Excel is an intelligent application: It will format your Data-5 file in the correct
spreadsheet grid. Your sheet will have 5 columns: A, B, C, D, E (A represents the first number in the drawings, etc.). Let's say your Data-5 file has 500 drawings.
Select column A. Click on Tool, then Data analysis. In the ensuing dialog box select Descriptive statistics and finally Summary statistics. Excel will create a new sheet, named Column1. There are some important numbers in the analysis sheet1, especially "Median" and Mode. The Median divides the column in two halves. The Mode represents the most frequent number in the column. Here is what kind of data I got for my data-5:
The first number (col. A): Median = 5; mode = 1
The 2nd number (col. B): Median = 12; mode = 10
The 3nd number (col. C): Median = 20; mode = 18
The 4th number (col. D): Median = 28; mode = 30
The 5th number (col. E): Median = 35; mode = 39
If you visit my site, you'll come across what I call the "Gambling Formula". The medians you see above are very close to what the Fundamental Formula of Gambling gives for a game of any probability p.
• The median can also be calculated in advance; i.e. the theoretical value. The only tool you can use is the well-known-by-now Fundamental Formula of Gambling! I call the parameter FFG median.
I give you one more tip now: Use the Median in conjunction with the Mode. How about this kind of number selection? For any position, select numbers between "Median – 4" to "Median + 4". As you can see, the Mode is right within that range. This is not a secret: It is a representation of the so-called "Normal probability rule". The most frequent numbers drawn IN THAT POSITION fall within the range we just set. In the case of my data-5, my best numbers will be within the following ranges:
The first number: 1 to 9
The 2nd number: 8 to 16
The 3rd number: 16 to 24
The 4th number: 24 to 32
The 5th number: 31 to 39
It is better to analyze larger data files. More than 500 is best, in order to get as close as possible to the gambling formula. In general, in more than 60% of the drawings, the winning numbers will fall in the ranges we set above.
While in Excel, you can sort data-5 by each column. You can see how numbers are distributed around the median. Click on the menu item Data, then Sort. Type the range $A$1:$A$500 (in the case of column A from drawing 1 to drawing 500).
Five: Using Spreadsheets to Analyze The Lottery II
(from the original post to rec.gambling.lottery)
Somebody started a thread on using SQL or Access to analyze the lottery. SQL is a database programming language. I pointed out that SQL is not very capable of analyzing the lottery, since it does not have built-in statistical functions to satisfy such a requirement. I recommended instead the use of Excel, the spreadsheet that is bundled with Access in
Microsoft Office Suite. First of all, I am not a software salesman. I only posted in the thread because I assumed the initial writer already had Excel on his PC.I showed how to use an Excel add-in: "Data Analysis". Its "Summary Statistics" generates very useful reports regarding lottery drawings files. You can read my post again. I need to mention now that my
"findings" are, in fact, a consequence of the so-called "normal probability rule". The rule is in turn derived from the "Gaussian Distribution". The normal probability rule establishes the
probability for an event in correlation with "the most probable result" in a probability distribution. A file with lottery drawings represents a probability distribution. The "median" is one of the most probable results in drawings series. Let's take as an example the median of the first numbers in the drawings. In the case of my lotto-5 drawings file, the median of the first number is 5. The standard deviation was 6. The rule of normal probability states that the probability of any first number in the drawings is 95.4% less than 2 standard deviations away from the median, and 99.7% less than 3 standard deviations away. Trying to make it more clearly. My lotto-5 drawings file has over 1000 drawings. Over 97% of the numbers drawn in the first position are between 1 and 18 (which approximates 3 standard deviations from the median).
Over 98% of the numbers drawn in the 5th position are between 21 and 39 (again, approximately within 3 standard deviations). This is valid for one number position at a time. If I want the simultaneous probability of the 1st and 5th numbers, it is .97 x .98 = .95. There is a 95% probability that the 1st AND the 5th numbers will be between 1-18 AND 21-39. The simultaneous probability for ALL five numbers was around 60% for my lotto-5 file. In over 60% of the drawings, the winning numbers were as follows:
1st number: between 1 - 18
2nd number: between 3 - 21
3rd number: between 11 - 29
4th number: between 19 - 37
5th number: between 21- 39
Total combinations of 18 taken 5 at a time is: 8,568.
My lotto-5 game uses 39 balls. Therefore, total combinations of 39 taken 5 at a time is: 575,757.
Now, important to note that Excel does not have a function to generate lottery numbers. The user needs to program such functions in Visual Basic for Applications (incorporated in every Office application).
Or, do what I do: write specialized lottery software outside Excel or Microsoft Office, using general-purpose programming languages (VisualBasic or PowerBasic). But if you do not program at all, it is obvious that it is far better to select numbers from the ranges determined by the normal probability rule. If you are to choose 5 numbers from the most frequent 5 ranges, the winning probability is 1 in 8,568. Choosing 5 numbers from a field of 39 has a
winning probability of 1 in 575,757. In 60% of the cases, you will face odds of only 8,568 to 1 if you choose numbers from the most frequent ranges.
Finally, the facts I have presented here are meaningful only if you already have Excel on your PC. The code I use in my software is entirely my business and I do not make it
available. My freeware software at FTP download site does not include the probability feature presented here. I do not accept analyzing your lottery drawings files for any fees. Keep in mind, however, that everything I have presented here represents undeniable mathematics. You may hear some nay-sayers screaming foul play. Your best reaction is to verify the facts yourself. Also, always question the motivations of any opinions. Here are my motivations. Number one, I am a soldier of the truth. I am searching for the truth, no matter how hard it is to reach it. Second, I am trying to
introduce my software and my Web site to a larger audience. This second motivation seeks both the financial profit and serving the public.
To Shobbolun and Joytsar:
Both of you are wrong. Shobbolun goes even beyond being wrong. By his nature probably he gets mad at any person who expresses something. You need a whole lot more study in mathematics and statistics. I am not here for the purpose of teaching such studies. The subsets I presented in my post are THE most representative samples in the series. They always are. Their probability is at least 60% and one can increase it by increasing the ranges (for example standard deviation+1 or +2).
If you apply the gambling formula you'll discover that an event of 60% probability only rarely skips 2 or 3 drawings in a row. You see two skips of the subset, it is a good chance the subset will hit the next drawing. But if you are not too much into theory, why don't you just check some real lottery drawings? Do you know the results of other readers of this forum? Not everybody is versed in mathematics. But everybody has the intelligence to check the facts. Get your lotto-5, or lotto-6, or lotto-7 ASCII drawings files and analyze them in Excel. Then try to be truthful and publish your results in this thread. Fair enough?
Drawings #1 and #2 have 5 winning numbers each from the subset generated by the normal probability rule. Drawing #3 has 4 winning numbers; drawing #4 has 4 winners; drawings #5 and #6 have 5 winners each; drawing #7 has 5 winners from the most representative sample.
• I just wrote the program to generate lotto-5 combinations within ranges. The program uses no filtering at all. Indeed, I was wrong regarding the total combinations to play! I mean, wrong, although not quite as wrong as some other calculations indicated.
I used first the following ranges:
1 to 18
3 to 21
11 to 29
19 to 37
21 to 39
The program generated 371,036 combinations. That represents 64.4% of all possible combinations for a lotto-5-39 game. The positive side is that the ranges eliminated 204,721 combinations. That's a very powerful filter, as far as I am concerned.
I used next the following ranges (some love to copy me; they named the ranges 'positional limits'):
1 to 9
8 to 16
16 to 24
24 to 32
31 to 39
The program generated 53,352 combinations. That represents 9.3% of all possible combinations for a lotto-5-39 game. You should expect to win 1 every 10+ drawings using those ranges. The best situation to play the ranges is after 4 to 8 skips (losses). In other words, if the ranges don't come up after 4 drawings, play them until they skipped 4 more
drawings. Evidently, the ranges by themselves are not sufficient. They need to be used in conjunction with other filters. Using other filters, I obtained from 0 (zero) to 8,000 combinations
to play. Of course, the winning frequency is lower. In my book, the best way to play the lotto games is to set filters that eliminate the largest possible number of combinations. Of course, you will have the chance to win more rarely, but a jackpot win compensates.
If you are interested in the program, I uploaded it to my FTP download site.
The name of the freeware is RANGES5.EXE. You need to download also the file RANGES.5, which consists of the ranges for the 5 numbers. The file is in ASCII format, so you can use any text editor to change it. Right now, the RANGES.5 file looks like this:
To generate ALL possible combinations of N numbers, make the five lines read:
1 N (substitute N for particular numbers). For example, to generate all the combinations in a 39-number lotto-5 game, edit RANGES.5 to read: