Welcome Guest
Log In | Register )
You last visited December 7, 2016, 5:07 pm
All times shown are
Eastern Time (GMT-5:00)

50/50 filters

Topic closed. 13 replies. Last post 9 years ago by time*treat.

Page 1 of 1
51
PrintE-mailLink
time*treat's avatar - radar

United States
Member #13130
March 30, 2005
2171 Posts
Offline
Posted: March 14, 2008, 7:30 pm - IP Logged

I see some of you like those 50/50 filters so ... here are the odds with the 1,024 possible outcomes using 10 of them.

 

                                                                                                                                                     

filter  count:10   
pass rate:50%   
     
passingfailingarrangements  ~odds
100                    1 0.098%
91                  10 0.977%
82                  45 4.395%
73                  120 11.719%
64                  210 20.508%
55                  252 24.609%
46                  210 20.508%
37                  120 11.719%
28                  45 4.395%
19                  10 0.977%
010                    1 0.098%

 Cool

In neo-conned Amerika, bank robs you.
Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.

    LANTERN's avatar - kilroy 28_173_reasonably_small.jpg
    Tx
    United States
    Member #4570
    May 4, 2004
    5180 Posts
    Offline
    Posted: March 15, 2008, 12:35 am - IP Logged

    Filters are risky and 50/50 filters are even more, either way it takes only one filter failure to ruin the prediction.

    The only good thing that I see about them is that if you guess right with one of them you cut the complete source of combos in half, that is a lot of filtration.

    As I said, it is safer to cut off 1 or 2 out of 8 or 10 than 1 out of 2.

    That is why it is so important to have so many filters that hold or have 5 or more patterns each.

    We will not be using all those many filters every single time only those few that appear to be safer to use at a particular time for a particular prediction according to the patterns of the past draws.

    Very tricky stuff as many people have already found out, not wise to over use any particular filter nor to use way too many filters at a time as those 2 things compound the rate and chance of filter(s) failure, also not all filters are equaly safe, some filters appear to be more stable than others overall in general, also at a particular time the past draws might guide as to what might be best to use at that time.

    Practice on paper first very many many times.

    I know that I am not giving any particular prediction details here on this post, just some general advice. 

      benmas's avatar - waveform
      Rhode Island
      United States
      Member #56010
      October 28, 2007
      434 Posts
      Offline
      Posted: March 15, 2008, 12:42 am - IP Logged

      can you please explain excatly what games this applies to and give some examples of these filters if you can

        time*treat's avatar - radar

        United States
        Member #13130
        March 30, 2005
        2171 Posts
        Offline
        Posted: March 15, 2008, 3:29 am - IP Logged

        @LANTERN: No worries. I'm just presenting some ideas on filters in general, not a specific game. I'm using 50/50 because it is the easiest ratio for most people to understand. The calculations are also easier for people to grasp. The thing you immediately noticed, that if one filter fails you lose, is what I'll be focusing on. 

        -----------------------------------------------------

        @benmas: This applies to any game where you can define characteristics that allow you to divide the numbers into 2 groups. They don't have to be 50/50, but that split is easiest for presentation. The same principles would apply for 70/30 or 85/15. The concept is not game specific, it is based on filter performance ratios.

        Pick 3 is the easiest example. Here are some examples by WIN D, to get you up to speed

        In neo-conned Amerika, bank robs you.
        Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.

          Avatar
          Kentucky
          United States
          Member #32652
          February 14, 2006
          7310 Posts
          Offline
          Posted: March 17, 2008, 10:07 pm - IP Logged

          Filters are risky and 50/50 filters are even more, either way it takes only one filter failure to ruin the prediction.

          The only good thing that I see about them is that if you guess right with one of them you cut the complete source of combos in half, that is a lot of filtration.

          As I said, it is safer to cut off 1 or 2 out of 8 or 10 than 1 out of 2.

          That is why it is so important to have so many filters that hold or have 5 or more patterns each.

          We will not be using all those many filters every single time only those few that appear to be safer to use at a particular time for a particular prediction according to the patterns of the past draws.

          Very tricky stuff as many people have already found out, not wise to over use any particular filter nor to use way too many filters at a time as those 2 things compound the rate and chance of filter(s) failure, also not all filters are equaly safe, some filters appear to be more stable than others overall in general, also at a particular time the past draws might guide as to what might be best to use at that time.

          Practice on paper first very many many times.

          I know that I am not giving any particular prediction details here on this post, just some general advice. 

          There are 10 digits in each position in pick-3 games so we can only apply one 50/50 filter per position; we can't equally split 5 odd or 5 even digits by applying a high/low filter or any other 50/50 filter. Using one 50/50 filter per position, there are 6 possible outcomes so the odds against are 5 to 1. Since most state lottery pick-3 games only pay $500 to $1, we can only get a 3 to 1 return playing the 125 combinations; it's not a good bet. If you get $900 to $1 playing online, the odds will shift slightly in your favor. The risk factor is the cost of playing 125 combinations.

          We could eliminate 2 digits by whatever method from each position and have an 80% chance of starting with the correct 3 digits. We can apply one 50/50 filter to each position, the odds against are still 5 to 1, but payoff odds of 6.81 to 1 are now slightly in our favor ((500-64) / 64).

          It would be very risky applying two more 50/50 filters to each position raising the odds against to 215 to 1, but you would get 499 to 1 payoff odds on the one 3 digit numbers.

          "I know that I am not giving any particular prediction details here on this post, just some general advice."

          If a system requires playing 125 straight combinations online to get a 120% return, I don't see how it could benefit the average player so the only application is for predictions. Players seem to want a system having 10 or less straight combos they can play in their state lottery that guarantees a hit every other drawing.

            benmas's avatar - waveform
            Rhode Island
            United States
            Member #56010
            October 28, 2007
            434 Posts
            Offline
            Posted: March 17, 2008, 10:29 pm - IP Logged

            say if a choose a 50/50 filter first time i apply one the odds are 50% if i choose another 50/50 now overall odds become 0.5*0.5=0.25..if i apply a third one the odds become 0.5*0.5*0.5=0.125..(a fourth one gives 6.25% and so on by dividing in half each time)...basically its like guessing three coin flips in a row...its very risky since in P3 you're gambling $125 bucks to win $375 a profit ratio of 3:1 with a win ratio of 1:12.5...most people will not make bets like these..the limiting factor in not making these or similar bets is that payout ratio not high enough to justify the low odds you are choosing to gamble with

              time*treat's avatar - radar

              United States
              Member #13130
              March 30, 2005
              2171 Posts
              Offline
              Posted: March 18, 2008, 3:34 am - IP Logged

              (The examples I linked to focus on p3. That was just to illustrate the concept. My focus is on p5, where this method really shows it's worth and where I developed it.)

              The next step is to drop down to 5 filters. This makes it easier to show details without getting lost in them. Really you can use however many you can keep track of.

               

              filter count5   
              pass rate50%   
                   
              passingfailingturns~odds
              501 3.125%
              415 15.625%
              3210 31.250%
              2310 31.250%
              145 15.625%
              051 3.125%

              As you see, it's very easy to fail one of your filters. The way most people work this, they will only get a 3% success ratio because they put the number against each filter until it fails one of them. The is the simplest way to code this. Here is each possible outcome in more detail...

               

              Outcome#1234567891011121314151617181920212223242526272829303132
              Filter 1PPPPPPPPPPPPPPPPFFFFFFFFFFFFFFFF
              Filter 2PPPPPPPPFFFFFFFFPPPPPPPPFFFFFFFF
              Filter 3PPPPFFFFPPPPFFFFPPPPFFFFPPPPFFFF
              Filter 4PPFFPPFFPPFFPPFFPPFFPPFFPPFFPPFF
              Filter 5PFPFPFPFPFPFPFPFPFPFPFPFPFPFPFPF
                                               
              pass5                               
              pass 44 4   4       4               

              If you code a little differently, you would count the total number of filters passed. This is a little more complex but what happens is that if you keep numbers that pass minimum 4 fo 5, you get 3.125% + 15.625% for a boost to 18.750%. That's a big 6x jump. This is where it will get interesting. I will show what happens when you use filters that are better than heads-tails... As I said before, this could be used for any game, but my focus is on p5.

              In neo-conned Amerika, bank robs you.
              Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.

                Avatar
                NASHVILLE, TENN
                United States
                Member #33372
                February 20, 2006
                1044 Posts
                Offline
                Posted: March 18, 2008, 9:36 pm - IP Logged

                Right now I am working with 12 filters.  I like to think using filters is on a higher plane than dealing with 39 numbers.  Here is my reasoning. 

                Using 39 numbers there are 575,757 possible combinations.  Using 12 filters ( I prefer calling them patterns) and setting each filter to either yes or no, there are 4,096 possible combinations.  Which is the better path?  One out of 575,757 or one out of 4,096?

                The one drawback to doing things like I am are the number of possible number sets that one specific pattern will generate.  The last Pick 5 had 26 possible number sets (including the winning number set, of course).  Below are the total amount of number sets which the winning pattern generated.

                26-5-19-6-33-429-147-374-26-26-41

                Naturally no one in their right mind would play 374 combinations when the next draw is the great unknown.  However, look at the 5 and 6.  Certainly those few are within anyone's budget.  And look at the 26.  That sum of possible number sets hit three times.

                The great problem now is to find which pattern will arise with the next draw.  In time I feel that problem will resolve itself.  So I am tracking the results, appling "what-ifs", and reading the many posts on LP.  So far, I have not guessed the right pattern since Tennessee went with cartoon drawings last July. 

                I could hide behind the fact there have been 98 drawings to date and that is not enough data to make an educated guess.  But I must work with what I have; not with what I want. 

                  time*treat's avatar - radar

                  United States
                  Member #13130
                  March 30, 2005
                  2171 Posts
                  Offline
                  Posted: March 18, 2008, 10:36 pm - IP Logged

                  I want to make sure I get this right. Idea

                  Are you saying your 12 patterns cut 575,757 combos down to between 5 and 429 combos, which contained the winner? Or are you saying one of the 12 patterns generated the winning combo and the other 11 didn't have the winner, and at the low end, that winning pattern had only 4 other combos in it? (In which case, the question would be which of the 12 patterns to use for any given drawing)

                  In neo-conned Amerika, bank robs you.
                  Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.

                    Avatar
                    NASHVILLE, TENN
                    United States
                    Member #33372
                    February 20, 2006
                    1044 Posts
                    Offline
                    Posted: March 19, 2008, 12:05 am - IP Logged

                    I want to make sure I get this right. Idea

                    Are you saying your 12 patterns cut 575,757 combos down to between 5 and 429 combos, which contained the winner? Or are you saying one of the 12 patterns generated the winning combo and the other 11 didn't have the winner, and at the low end, that winning pattern had only 4 other combos in it? (In which case, the question would be which of the 12 patterns to use for any given drawing)

                    The patterns are yes and no questions.  Right now, you look at what came before and make a decision.  Take repeat numbers as a pattern.  Will there be a repeat number in the next draw?  If yes, then set your parameter to one (1).  If you think not, then set that parameter to zero (0)

                    My program then looks at each of the 575,757 combinations and records only those that contain a repeating number.

                    Now I have 12 of these yes/no questions.  Set each one to 1 or 0 and run the program.  Sometimes my 12 patterns generates 5 number sets, one of which WAS the winning number.  Other times my program generates 374 number sets, one of which was the winning number.

                    I study the results of these patterns.  I wish I could say I have found something worthwhile but I can't; not at this time. 

                    For tomorrow night (Wednesday, March 19), I think pattern 940 will prevail.  Pattern 940 contains 36 number sets.  If I were playing money, I would run my program, record all 36 number sets, fill out a play slip with 36 lines, and lay my money on the table.  I have a ways to go yet before I am comfortable with being able to predict the next successful pattern.  So I am hanging on to my money for now.

                    There is beauty in what I am attempting.  This way of thinking about the lottory will work for every game except Powerball and MM. 

                      time*treat's avatar - radar

                      United States
                      Member #13130
                      March 30, 2005
                      2171 Posts
                      Offline
                      Posted: March 21, 2008, 12:58 am - IP Logged

                      O.k. I follow that. Some of my filters are "Must Pass", whereas others I'm being more flexible with. Your system will work fine with what I'll be putting up later. 

                      In neo-conned Amerika, bank robs you.
                      Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.

                        time*treat's avatar - radar

                        United States
                        Member #13130
                        March 30, 2005
                        2171 Posts
                        Offline
                        Posted: March 26, 2008, 10:23 am - IP Logged

                        Moving up from 50/50's, this is what happens with 90% filters. There are a plenty of these in p5 games, which is the level you need to cut down to a playable amount of numbers:

                         

                        filter count5   
                        pass rate90%   
                             
                        passingfailingturns~odds
                        501 59.049%
                        415 32.805%
                        3210 7.290%
                        2310 0.810%
                        145 0.045%
                        051 0.001%

                        Even with each filter containing the winner 90% or more of the time, when you require the numbers to jump through 5 hoops at once, the winner will manage in only 59% of cases. Letting it fail any one filter, i.e. keeping numbers with 5/5 and 4/5 success rates, boosts the overall results to 91.85%. The idea is that you aren't throwing a jackpot away because you were too much of a stickler with the selection process, especially as some of your setups may take a while to materialize. The expected performance is calculable even when the success ratio isn't homogenous. In those cases you'd use the geometric mean as the equivalent pass rate.   

                        In neo-conned Amerika, bank robs you.
                        Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.

                          Avatar
                          NASHVILLE, TENN
                          United States
                          Member #33372
                          February 20, 2006
                          1044 Posts
                          Offline
                          Posted: March 26, 2008, 10:00 pm - IP Logged

                          I would love to know of a 90% filter.  Right now, the best filter I use has a 58% success rate.  My worst, has a 32% success rate.

                          Please give me an example of a 90% filter.

                          And let me comment on a previous post to this thread.  The way I am reading the post may not be what the author meant.  If I am mistaken, he can correct me.

                          Filters, by their very nature, will eliminate the winning combination most of the time, if not all the time.  Hence, the odds of winning that night will be reduced to zero.  That should not mean we ignore filters. 

                          Some of us study the numbers; how they fell, what follows with what, gaps between numbers, and so on.  Why can we not apply the same logic ( perhaps insanity is a better word) to fillters?  How did the filters fall?  What gaps exist between the filters?  Which filter is "High"?  Which filter is "Low"?  What will the filter do with the coming draw?  Things like that.

                          I am having a ball of fun examing the results of Illinois' Little Lotto right now.  I have gone back to January of 2004 and moving forward, processing each draw, recording the filter's results, looking at possible patterns (and being disappointed).  Well, it keeps me off the streets at night anyway.

                            time*treat's avatar - radar

                            United States
                            Member #13130
                            March 30, 2005
                            2171 Posts
                            Offline
                            Posted: March 27, 2008, 4:30 am - IP Logged

                            When people find their "system" doesn't work, I will tease them and say they have a "filter". I sometimes interchange the words "system" & "filter". So I should explain that first... the other part about % filters will depend on your states number history. I haven't done much of anything with TN's. 

                            If you have a means of breaking the (in your case 575,757) number field into two or more groups (they don't have to be equal sized) and you can predict which group the winning combo will appear in with an accuracy greater than raw stats would indicate, then you have what I would call a filter (or systemClown).

                            E.g., you have a technique that makes Group A of 518,182 numbers and Group B of 57,575 numbers. By stats, the winner should show up in Group B 1 in 10 drawings. If your method makes it show up in Group B 2 in 10 drawings, you have something worth paying attention to because it's performing 200% better than random chance.

                            Since you live in an RNG state, you have the option of looking for patterns in the actual draw that happen more or less than random chance would call for. (I think your state calls them *glitches*Crazy) see JADELottery's CGN thread on that.

                            BTW, 32% of 575,757 is 184,242 so if your 32% filter is generating significantly fewer numbers than that (let's say it hits 1/3rd of the drawings but generates 1/6th of the combos), I'd hold on to it. That's not to suggest playing 1/6th of the total combos. I'm always on the hunt for a new filter or idea, so if you don't want it anymore...Big Grin

                            The deal with the pick5s is not to hit or even play every day, it's to hit small until you can hit big. And seeing how your game is "free" on a 2/5 match (I'm jealousRazz), combining a few things in the right order may very well get you playing for free until you hit the big time (those $100K & $200K jackpots are nice, too).   

                            In neo-conned Amerika, bank robs you.
                            Alcohol, Tobacco, and Firearms should be the name of a convenience store, not a govnoment agency.