Forum

Leaderboard

So how is that fare if I have received 40 less points than others?

ouzocoke
29 May 17 8:17 PM

When teeing off to give player's on the fairway a sporting chance, remember to call fourrrrrrr! Fair enough? :)

Lemond
29 May 17 10:57 PM

First of all if that joke was supposed to be funny - its not. If it is random why do you see the same names at the top?

ouzocoke
30 May 17 6:19 PM

First of all you're forgetting rule 1.

Rule 1 : Competitions are all about the promoter.
Rule 2; Read the T&C's. If you're going to play the game know the rules.
Rule 3: See Rule 1

Competitions achieve what the promoter perceives will benefit them.

The leader board is weighted system that's predominately skilled based with a random element to decide the winner from members of similar "skill". The skill is being an active member filling in all the "prerequisites" and completing the daily tasks. You can call the skill determination, persistence or active member. Overall not a complex skill thou seems to weed out the hundreds of less active members and left with approximately 50 determined active members each month. So for starter's the leader board is biased towards the most active members.

You could argue that the penalties for not completing a task are excessive i.e. missing a day of the guessing game (-50 potential points) and are virtually a prerequisites to have a genuine chance of the top 3. That is likely by design to achieve rule 1 and encourage members to be active.

Now the random component excluding any skill components for competitions viewed etc which continually raises eyebrows.

I underlying burning question is there an issue with the random generator?
Most random number generator use pseudo random numbers generation. Not all random generators are equal. They simulate random numbers according to an algorithm usually based on a seed value. There are various tests you can use whether a random generator is robust enough for it's purpose.

One method is examining it's distribution/pattern. That may spot potential issues when scores frequently exceed expected outliers.

The total monthly daily random points for follow a cumulative distribution function. The shape of the distribution bit like a slippery slide which tappers off at each end. I.e. The odds of achieving a score higher or lower then the expected extremities tappers off fairly quickly. The majority of scores will be around the average monthly score. The expected scores depend on the number of days in a month.

So your absolute monthly score score and difference is meaningless by itself. You need to compare it to the expected average for a month. i..e. average daily score is (10+1)/2 = 5.5. and days in this month 31, 5.5*31 = 170.5. So if end up with a score around 171 expect to end up midway of active members. 50% mark

For 31 days the average top score is approximately 206+- about 5 points and the lowest is about 126 +-5. The +- is the level of confidence. I think with around 99.9% confidence top scores was around 220ish. Yes it's possible to achieve scores higher or lower but highly unlikely. The range of expected monthly scores for 31 days is around 90 points. So being 40 points behind the leaders is well within expected scores.

Looking at this months the top 3 scores 216, 208, 206 three scores equal or above the average expected top score. "interesting" I might have to dive into the simulation spreadsheet I did last year for rough odds of having that many. From memory possible and probably due for some scores higher then average. My impression of the last few months have been slightly below average top scores. The top score is getting close to limit where would start getting concerned thou in isolation nothing unexpected.

The combination probability of having the same 3 winners
Will have to think back 20 years. Could be some errors.

See for combination and permutations calculators:
http://www.calculatorsoup.com/calculators/discretemathematics/combinations.php or
http://stattrek.com/online-calculator/combinations-permutations.aspx

Presume minimum 30 active members
nCr = n! / r! (n - r)! i..e n = 30, r = 3
30!/(3!*27!) = (30*29*28)/(3*2*1) = 24360/6 = 4060 combinations each month where order of top 3 not important

So odds of picking the top 3 in any order would be 1/4060 and in order 1/24360

For picking two out of three winners : (3*2)/(2*1) = 3 combinations i.e. (1,2; 1,3; 2,3)

Two members out of 30
C(n,r)=C(30,2)
C(n,r)=C(30,2)
=30!(2!(30?2)!)
=30!(2!(30?2)!)
= 435

For 2 members 435 combinations of those 3 combinations are top 3
so 3/435 = 1/145 both are in the top 3

So given member X and Y the odds of them being in the top 3 is 1/145.

Something like that will think it next month :)

Lemond
31 May 17 8:36 PM

eg For a 6 sided die if roll a 1 then probability of rolling a one next roll is 1/6. Given 1 then next two rolls 1/36.

I think using last month as your base reference ie given member X and Y then the probability is 3/435 chance they will be top 3. Nothing way out of the ordinary. If the same two end up the following 2 months =(1/145) ^ 2= 1/21,025

See how it goes in June.

The birthday bonus component is another point you and argue if fair or not. i.e. birthday distributions aren't equal and higher bonus the greater bias against higher then average birthday months. Think about it if set too high they on average they share the top prizes, the more birthdays the more sharing the same prizes and in effect lower odds compared to other months. Then you got the effect members wising up an going AWOL if feel the number of birthdays in a month is high. Might explain why it's posted as late as possible now.

Hmmm maybe still a touch high from memory with a slight disadvantage to higher number of birthday months. Something like 5% will have look it up later. 25 points thou is vastly better then 50 bonus points that was initially set last June. If less then 5 birthdays in a month I don't think was a significant issue. At some point it might need revising by lowering that bonus or increasing the number of prizes to compensate, if the number of active members increases enough.

ps From memory almost forget about winning in June, If it's not your birthday. If it's like last year which had at least 6 birthdays about,
you're likely to need close 210 points, a slim thou possible chance for non birthday members. Will place bets will see the return of @moreofachance etc being more active in June and at least two birthday winners. :)

Lemond
31 May 17 9:22 PM

Hmm one thing noticing is lack of lower scores expected around the 135-150 range.
Struggling to find many, I think should expect a handful at least. That will have to check on average roughly how many scores would expect in that range.

Possible due to the multiple fixes have pushed those active members off the leader board which usually show up.

Might be hinting the distribution might be skewed towards the higher end this month caused by abnormalities with the random daily points. Might also explain why top scores are a touch on the high side.

Lemond
31 May 17 10:14 PM

Can u please explain all that one more time lol

Brookemaree93
01 Jun 17 9:21 AM

@Brookemaree93 lmao. Well the odds of any two members winning in a month is 1 which was the what @ouzocoke overlooked.

Argh worse thing knowing there're so many June birthday's about. I've got Buckley's chance this month. :(

Note for next year note there didn't appear to be many if any May birthdays for the active members.

Lemond
01 Jun 17 3:11 PM