It will not happen in your life-time!

[Kon-Fu-Sed]

Hello all,

"It happens only once in a million bets. You'll never make one million bets in your life-time!"
You've heard it before?

Sure you have.

But do you know what's really enlightening in that statement?

- That the one who said it uses probability but doesn't know what probability really is!


Probability may say that an event only happens once in a million bets.
And that may be correct.

BUT!

It doesn't say ANYTHING about WHEN that event will happen.
It may happen right now or it may happen "never in your life-time".
But we don't know.

Let me give you an example.

Suppose that one thousand and one hundred (1,100) of the members of this board made (OK - recorded) one thousand spins/bets each.
1,100 members x 1,000 bets = 1,100,000 bets in total.
1,000 bets isn't much for one person.

So I made 1,100 sequences of 1,000 RNG single-zero results and counted losing sequences for the even chances.
To lose RED means a sequence of BLACK and ZERO.
To lose BLACK means a sequence of RED and ZERO.

I checked all six bets and for each sample, I recorded the longest losing sequences.

The WORST LOSING SEQUENCES in 1,000 RNG spins
(1,100 samples of 1,000 spins = 1,100,000 spins checked)

[table=,]
Length,   L Blk,   L Red,   L Hgh,   L Low,   L Evn,   L Odd
   6,      12,      12,      10,      11,       7,      13
   7,     101,      94,      85,      76,      78,      77
   8,     249,     196,     224,     210,     252,     225
   9,     249,     284,     254,     299,     262,     289
  10,     204,     200,     215,     201,     226,     191
  11,     124,     136,     149,     127,     134,     137
  12,      79,      71,      72,      86,      71,      79
  13,      35,      47,      47,      38,      35,      46
  14,      23,      25,      28,      24,      15,      21
  15,       9,      20,      10,       9,      12,      11
  16,      11,      11,       3,      11,       4,       5
  17,       3,       4,       3,       2,       3,       4
  18,       1,        ,        ,       5,       1,       1
  19,        ,        ,        ,        ,        ,         
  20,        ,        ,        ,        ,        ,         
  21,        ,        ,        ,       1,        ,       1
[/table]
Example: In 94 samples, the worst losing sequence betting Red was 7 consecutive losses.

As you can see; some were very lucky getting only a rather short losing sequence, like 6 or 7.

But quite a lot got quite un-healthy lengths... And that's not good...
Note especially the 21-sequences - thats a 1 / 1,197,872 chance that one such will appear!
And this unfortunate member had only 1,000 bets to make!
That's not even ONE PER MILLE.

But think of it: We checked 1,100,000 spins...
Do you see the connection?


- BUT THAT'S RNG-SPINS!


As I know that someone at some point sooner or later will point that out, I also did this:

I took all the spins I have from Wiesbaden (table #3, April 1 2003 to July 31 2008) and Hamburg (Table #1, 1 Jan 2001 to Dec 31 2005)
That is a total of 1,145,828 spins.
So the number of spins are approximately the same as the RNG-spins...

But this time I checked DAY BY DAY instead of 1,000-spins chunks.
In total I checked 3,689 days.
Making an average of 311 spins/chunk instead of 1,000. One third, approximately.

Here you are:

The WORST LOSING SEQUENCES IN ONE DAY at Hamburg and Wiesbaden
(1,145,828 spins in 3,689 days checked)

[table=,]
Length,   L Blk,   L Red,   L Hgh,   L Low,   L Evn,   L Odd
   2,        ,        ,        ,        ,        ,       1
   3,        ,       2,        ,       1,        ,       1
   4,      26,      21,      26,      20,      24,      28
   5,     244,     222,     224,     228,     231,     236
   6,     693,     674,     630,     631,     635,     630
   7,     877,     902,     900,     866,     929,     932
   8,     721,     724,     775,     724,     752,     757
   9,     532,     495,     461,     541,     475,     508
  10,     288,     291,     292,     338,     304,     284
  11,     166,     172,     181,     164,     147,     145
  12,      68,      85,      80,      90,      95,      86
  13,      39,      60,      53,      36,      56,      39
  14,      19,      25,      35,      23,      23,      23
  15,       8,       7,      18,      12,       9,      13
  16,       3,       5,       5,       7,       4,       3
  17,       1,        ,       4,       5,       3,       1
  18,       1,       1,       3,        ,        ,       2
  19,       1,       3,        ,       2,       2,         
  20,       1,        ,       1,        ,        ,         
  21,       1,        ,       1,       1,        ,         
[/table]
Example: In 222 days, the worst losing sequence betting Red was 5 consecutive losses.


Do you notice?

There are STILL 21-long losing sequences! But now there are THREE OF THEM!
And the checked sequences were only 311 spins in average...

And what about all those short sequences - 5 and shorter?
Why are they so many when we check live-spins?

The average of 311 spins instead of 1,000 makes us find all those shorter sequences.
If I make a new RNG run but use 3689 sets of 311 spins (the same as the average live days), I get this table:

[table=,]
Length,   L Blk,   L Red,   L Hgh,   L Low,   L Evn,   L Odd
   3,        ,       1,        ,        ,        ,         
   4,      12,      13,      12,      10,      11,      12
   5,     196,     182,     211,     182,     186,     203
   6,     665,     676,     647,     692,     655,     637
   7,     915,     916,     951,     901,     890,     909
   8,     766,     775,     727,     795,     808,     747
   9,     545,     551,     524,     495,     478,     523
  10,     309,     272,     271,     306,     308,     320
  11,     145,     138,     149,     133,     173,     157
  12,      64,      84,      96,      81,      78,      73
  13,      34,      30,      54,      41,      41,      66
  14,      26,      25,      22,      28,      31,      16
  15,       6,      14,      12,      15,      16,      14
  16,       1,       9,       7,       6,       7,       4
  17,       2,        ,       3,       2,       4,       5
  18,        ,        ,       2,       1,       1,       1
  19,       2,       1,       1,       1,       1,       2
  20,        ,       2,        ,        ,       1,         
  21,       1,        ,        ,        ,        ,         
[/table]
Example: In 182 samples, the worst losing sequence betting Red was 5 consecutive losses.


Do you think live spins differ very much from RNG?



But now... Prepare!


How many bets do we REALLY HAVE TO WAIT for that "never in a life-time" event?


I supposed betting on HIGH (on a single-zero wheel) and counted how many results I saw until a sequence of 21 losses appeared.
To find 21 un-specified 18/37 losses is - as said above - a 1 / 1,197,872 chance. Mathematically.
That means I SHOULD have to wait, in average, something like 2.3 million trials before a 21-sequence of HIGH losses would appear.
Mathematically.

Well, I counted until I had found 50,000 such losses.

(I was aming at 1 million but I realised that only finding 10,000 was more than 6 hours computer time... 50,000 losing sequences meant a total of 113,279,546,087 trials )

BTW!
50,000 sequences giving a total of 113,279,546,087 trials is an average of 2,265,590 trials/sequence.
For how long did I say, above, should I mathematically have to wait for 21 High losses?


The results:

Waiting trials before the "never in your life-time" event starts
(Counting spins between sequences of 21 consecutive losses on the "High" bet)

Wait Interval   Sequence of
From - To       21-losses
   1 <   1K       27  <== Shortest = 135 spins!
  1K <   2K       29
  2K <   3K       12
  3K <   4K       18
  4K <   5K       15
  5K <   6K       19
  6K <   7K       16
  7K <   8K       16
  8K <   9K       25
  9K <  10K       15
10K <  20K      210
20K <  30K      201
30K <  40K      208
40K <  50K      225
50K <  60K      216
60K <  70K      203
70K <  80K      185
80K <  90K      179
90K < 100K      203
100K < 200K    1,972
200K < 300K    1,909
300K < 400K    1,802
400K < 500K    1,754
500K < 600K    1,743
600K < 700K    1,620
700K < 800K    1,616
800K < 900K    1,523
900K <   1M    1,545
  1M <   2M   11,579
  2M <   3M    7,418
  3M <   4M    4,809
  4M <   5M    3,052
  5M <   6M    2,028
  6M <   7M    1,488
  7M <   8M      926
  8M <   9M      514
  9M <  10M      318
10M - ....      459


So much for "It happens only once in a million spins and you will never place one million bets in your life-time!"
It CAN happen after 135 spins!


I hope you can see now, that a very small probability of something to happen, doesn't prevent it from happening NOW.

And many times it does.


Now you have been warned.

So don't be surprised when the s**t hits the fan!
KFS

[Kon-Fu-Sed]

Addendum:

The probability for a 18/37 bet to lose 21 times consecutive is, as said above, 1 / 1,197,872

That figure is for an UN-specified bet.
Meaning that in 1,197,872 spins we could mathematically expect...
... one such losing sequence for Black OR Red
... one such losing sequence for High OR Low
... one such losing sequence for Even OR Odd

So mathematically we could expect close to one in each PAIR of bets in the first, 1,000 RNG spins, table.
There were two in total.

And in the Live-spins table as well as in the 311 RNG spins table, we could expect very close to one in each pair as the number of spins is very close to 1,197,872.
Live spins: 1,145,828
RNG spins: 311 x 3689 = 1,147,279
There were three in total in the Live-spins table.
But only one in the 311 RNG spins table...
(That's called "random" )

KFS

[Adriatik]

Can we say for something that it will NEVER appear ?

example: can we say that one nuber will not repat 1000 x in row? I think we can say that, because of  the law of big nubers.

So if we can say that, can we say that one nuber will not repeat 100 x in row, or 50 x in row .... Where is the border between something that will not happen for sure, and something that is most likely that it wont happen ?

[Kon-Fu-Sed]

Hello Adriatik,

Thanks for your interesting thoughts.

I think, my opinion only, that as long as the probability is anything greater than 0, we have to accept that it is possible to happen.

But one problem regarding the 1000 times repeat of one number is that we haven't seen enough spins yet, to neither prove nor disprove the possibility.

Look here:
To find a number repeating "only" 16 times, we need 4,373,282,969,801 spins, mathematically.

Suppose roulette has been recorded for 200 years. That's 21,866,414,849 spins to record each year.
Suppose doing it for 24 hours a day, 365 days a year. That's 2,496,166 spins per hour.
Suppose 60 spin per hour and wheel, and we need 41,602 wheels running.
Since day one; 200 years ago.

Now, that was only a 16-repeat.
A 20-repeat would require 1,874,161 times as many: 77,968,845,922 wheels

Running 24/7 for 200 years...
...


Anyway, I like this question:

So if we can say that, can we say that one nuber will not repeat 100 x in row, or 50 x in row .... Where is the border between something that will not happen for sure, and something that is most likely that it wont happen ?

Exactly: Where is the border?

You tell me...

Unfortunately I'm more of a maths person than anything else, regarding these subjects , so for me the border for "not happen for sure" is when the probability is anything greater than 0. Anything greater.

But I promise I'll be TRULY flabbergasted when a casino reports a full day of only one repeating number.
I wish it will happen during my lifetime.


Best regards,
KFS

[TwoCatSam]

KFS

Your analysis reminds me of a preacher I once heard.  "Manifest Destiny" he called it.  His theory was you could have anything you wanted in life IF you didn't set a time limit on when you got it.

'Course if you died before you got it, that didn't count.

Sam

[Kon-Fu-Sed]

That's true, Sam.

And anything you want to happen may happen immediately, if you don't limit to whom it should happen...


KFS

[hoper35]

Ya, but it won't happen to me. 

[lucky_strike]

Well it would be nice to see you make the same test with bet selections that is not "static".

Cheers LS

[Kon-Fu-Sed]

Hello Lucky,

I'm not sure what you mean...
I wanted to show that a long, low-probability, losing sequence may hit you right from the start - as well as "close to never".

with bet selections that is not "static".

Would a long losing sequence be delayed because you change the bets from spin to spin?
(If that's what you mean would happen if you use a non-static bet...?)

It would be nice to see you make the test and show us.


/KFS

[lucky_strike]

If you use an "bet selection with an random element with one static rule you get an momentum that will continue to change aginst your previos bet selection and it would never play with or against the same momentum twice "bet selection" and its not the same thing as playing with or against 24 reds or 24 blacks "thats is an static bet selection".

Its like winning the lotteri or be hit by lightning if you lose 6, 8 or 12 bets.
That would change you math and statistics.

If you find this intressting i can make a post with one simpel rule to follow and you will see that the result above will not become the same thing as your new results.

Cheers LS

[Kon-Fu-Sed]

Pleeease show us statistics.
Please.

Regards,
KFS

[lucky_strike]

If you find this intressting i can make a post with one simpel rule to follow and you will see that the result above will not become the same thing as your new results.

Don't get me wrong KFS i love your post and what you contribute to this forum.
But i am not a programmer and i did ask you to do this but you say no.

So its up to you if you find the idea intressting to try out, but now you are the "Probability Meister with static rules" ops just kidding, that was a joke
I just find it boring when some don't think "out of the box" and get stuck with static elements all the time.

And i can admit that i am wrong if some one prove me wrong...

Cheers LS

[Kon-Fu-Sed]

LS,

Yes; I may be the "probability meister with static rules"
That's OK. If we mean the same thing:

1) I know SOME probability theory - but far from everything. And I (think I) reason quite logically.
(But it isn't me who put "Probability Meister" under my avatar.)

2) I always use static rules in tests.
How else can the results be evaluated?

"This test shows that constantly changing every rule is profitable."

And what use do we have for such a conclusion? Where is the value?
Or any result that comes from changed rules.
Either you use the same rules till the end of the test or you break the test and start from scratch with the new rules.

So no offense taken.


So don't take this wrong:
I (and my friend Peter who actually make all those stats) find it boring when someone comes along with an idea and asks for tests without anything at all to show what, how and why that idea will...
* Increase the hit-rate and/or decrease the loss-rate, as that would be a change
* Make the UNITS net worthwhile a long-term use of the method, as that would be a change.

When someone actually can show HOW and WHY the method can ACTUALLY change the -2.7% into + or the n/37 into n/35, Peter will gladly make tests.
That's his promise.


You wrote:

Its like winning the lotteri or be hit by lightning if you lose 6, 8 or 12 bets.

Yes! Exactly. I believe that.

And that's just what I tried to show above -[highlight] that you MAY be hit by lightning TODAY. [/highlight]

But we don't know and therefore we can't say that "it will not happen".
Just look above how many times that once-in-2.3 million event hit before 1000 spins - the shortest wait being only 135 spins.

135 spins at the casino and you're hit by a once-in-2.3 million event
135 steps in your garden and you're hit by lightning.

What would be interesting would be a lightning-conductor that really works.
But as you say yourself; we still may be hit.

And that will hurt.


Best regards,
KFS

PS. I'm sure that someone else is qualified and curious enough to write your program if you just show what in your idea that's unique and out-of-the-box.

[lucky_strike]

What would be interesting would be a lightning-conductor  that really works.
But as you say yourself; we still may be hit.

Thanks for your reply and for sharing.

I just don't agree that its the same thing to win the lotteri with an momentum as using a static sequense.
You wont get the same evaluated results as above and is not hard to code an bet selection with an random element with an momentum, because it has one staic rule it always follows.
So it would give you evaluated results as above, if you code it.

For now on i will claim i have an "lightning-conductor"  until some one prove me wrong.
But it will also fail but it just make it littel harder to do so.

Cheers LS

[Adriatik]

KFS,

can we look at this problem like this:

how may roulette wheels are spinning in world hand and air balled. In my small city there is about 10. And there must be 100000 cities like so lets say we have 1 M roulette wheels. So 1 M x 100 years x 12h of spins per day. Can you calculate now how many repetition of one number can we expect ?

[Adriatik]

-