### Author Topic: Exercise for logic  (Read 2022 times)

0 Members and 1 Guest are viewing this topic.

#### MrPerfect.

##### Exercise for logic
« on: June 11, 2017, 01:12:53 PM »
In the file attached there is an example how quants look for roulette . Find a logical error.

The following users thanked this post: kav, Herby

#### MrPerfect.

##### Re: Exercise for logic
« Reply #1 on: June 13, 2017, 03:22:55 PM »
It looks complicated,  right? Lots of unnecessary math with whatever...
If we focus on words and charts only, we can see that there are some nice numbers there.
Autor just pick up 2 best and decided " this is it".
Unfortunately 2 numbers  bets do have huge variance. They located right next to each other , wich don't help much as well. Testing Kelly in this case is stupidity. Simply he had to choose more numbers and create a bet with different denominations , depending on numbers performance and stability.
This is clear example of how good methodology and correct math in wrong hands do produce whatever results. He is not a player, that is his main problem.

#### Janusz

##### Re: Exercise for logic
« Reply #2 on: June 13, 2017, 06:40:26 PM »
It looks complicated,  right? Lots of unnecessary math with whatever...
.....
This is clear example of how good methodology and correct math in wrong hands do produce whatever results. He is not a player, that is his main problem.

As I understand you didn´t understand anything at all.
How can you criticize math, not understanding what it is about?

One question: What (in Terms of Roulette) is "mean reversion level"  and "mean reversion speed oder mean reversion rate" ?

Or do you just ignore that by simply use "2" numbers?

#### MrPerfect.

##### Re: Exercise for logic
« Reply #3 on: June 13, 2017, 06:51:14 PM »
My point was that after all that bla bla bla, he ( autor) just choose 2 numbers standing next to each other as his target.
Have no clue what these terms mean in his interpretation.

#### Janusz

##### Re: Exercise for logic
« Reply #4 on: June 13, 2017, 07:31:57 PM »
My point was that after all that bla bla bla, he ( autor) just choose 2 numbers standing next to each other as his target.
Have no clue what these terms mean in his interpretation.

So you don´t know, why he chose only 2 numbers standing "near" to each other?
See Markov: If the "hidden movement" (HMM) creates a distance between 2 numbers, it is possible that this distance is created also between the next spins.
The "sequence" (HMS) which creates this distance is unknown, but it has a higher probability to repeat.

Pls note, that this is statistical Math, not really transferrable to roulette.

#### MrPerfect.

##### Re: Exercise for logic
« Reply #5 on: June 13, 2017, 10:56:56 PM »
Janusz, sry asking, but did you read a doc l posted?  Autors choice of numbers has nothing to do with Marcov or distances between numbers... it's simply due the fact that these numbers pass 3% filter.
There is no errors in autors math besides logical ones. I have no clue whatsoever about first half of document, where he makes unnecessary math show off. There was no need to prove that probability of each number to hit is 1/37, he could simply say so, l would belive him , wouldn't you?
Since when stats math is not transferable to roulette?

#### fiben7

##### Re: Exercise for logic
« Reply #6 on: May 19, 2019, 06:57:59 PM »
MrPerfect,

As per your kidn request, I had a look at the paper in this topic discussing a quantitative trading approach to a specific roulette wheel.

I have to agree with your main point here, that the logic of the writer when choosing straight numbers to flat bet just because the in-sample probability of appearance during the first 5,000 spins is higher than 3%, looks rather naive. Kelly's comparison to FBS looks poor as well, it is not well put and tested through robust sensitivity analysis.

A few points of my own

• I do not believe that 10,000 spins are enough for any statistically significant result and system development, whatsoever if this system is as simple as bet on some straight numbers that looks like they appear more often than usual
• Math is nice and relevant to the topic under study here. The author tries to formalize his concept in a concrete way, no harm there. However, math should be used with a wise mind based on the actual practicality. Example, it makes no sense that the author uses a Chi square test to check whether the roulette is biased, the result of the test is that the roulette is not biased but then the author says that he could exploit short-term biases of the roulette on betting on certain numbers. whaaaaat???
• the quantitative approach is evident here and as per my post, I fully agree on the point to use quantitative tools in order to analyze, evaluate and compare roulette systems and strategies; maximum drawdown, calmar ratio etc are powerful enough to lay the truth upon our eyes
• in order to use such quantitative tools, I agree with you that there is no need for players to understand or follow such advanced math; as you said, prob is 1/37 and payouts per type bet is also known and simple to understand for all players. after the real play, one could just use the series of recorded spins and analyze such performance metrics through simple excel functions or relative finance software packages
• in addition to my previous point, given the fact that when playing live on a roulette, the player is not allowed to have any application, laptop, tablet etc using software that features all such complicated and difficult math to generate predictions but is only allowed to have a pen and a paper, then the complicated math stuff becomes unapplicable in reality. Whatever system one has, needs to be practical, i.e. all calculations and rules, trigger signals etc. should be relatively easy to be performed at the limited time between each spin.
Kind regards MrPerfect.

The following users thanked this post: UnlikelySam