Forum

The following is a output of a simulation for the current leader board for (1-10) random daily points with various birthday bonus and how they effect expect outcomes.

@admin appologise if the image isn't too clear. Working on that.

Imagine lot of questions to make sense of it all.

Simulation based on 1200 randomly generated monthly totals using the daily random points (1-10) for up to 100 participants.
For birthdays I randomly allocate each participant a birthday and obtain a randomly generated birthday distribution.

For the simulation I've selected one random birthday distribution as shown in the table i.e. 2,3,3,3,3,4,4,4,4,6,10 for the sum of birthdays in a month. The order of the number of birthdays for each doesn't correspond to the order of months as randomly generated.

For the birthday distribution applied various birthday bonus points using the same set of random monthly totals for 50 participants.

The results are sorted initially for the scenario 300 birthday bonus points, according to the number of birthdays in a month then and expected payout order. I.e. 300 birthday points would guarantee being at the top of the leaderboard. The progessively did this for the various birthday bonus points of: 300, 50, 40, 35, 30, 25, 20, 15, 10, 5, 0.

Where 0 bonus points is a totally random outcome.

@Lemond I agree with @toxicgherkin if changes are to be implemented then they should start 1 Jan (midnight).

I don't have a problem with birthday bonuses as they stand now. I presume there is a something in place that monitors people changing details like birthdays in order to get a few bonuses a year.

@Anna2077
Yes you hope there are checks and blances changing birthday dates. The sacastic humour must of been lost. I.e. was a dig at the greatly bias system due to reducing the daily points and not balancing the birthday bonus correspondingly. I'm sure if I changed mine it wouldn't go unnoticed by some members. :)

@admin

@Anna2077 So you're supporting to continue a greatly biased system, favouring some groups of members over others? And which side of the coin to you think you think you fall on? The "fool" or "unjust"? That would be the first question, if making a decision based on self interest. Being pre-emptive, falling on the finely balanced edge scenario is improbable!

@toxicgherkin
Removing the birthday and ASAP may sound like the Grinch who stole Xmas. lol With 50 birthday bonus the only members expecting a Merry Xmas would be probably be those with a birthday in December. If more then about 4 birthdays, most likely someone will miss out. Do you feel lucky this year. :) There is the case of multiple winners, and/or the "OPHRA" scenario. For that scenario, likely someone would have to "organise" New Year celebrations in advance.

Just of Easter please can have a birthday bonus of 125 or more and while at it excluded those members that won this year for that month. Just incase also exclude any new members with birthdays in that month that have joined since then. Of course while keeping the random daily points at (1-10). (Sarcasm included)

If read to the bottom there is some GOOD news for reducing the birthday bonus.

My suggestion for the current system of (1-10) random daily points, is to scrap the Birthday Bonus of 50 daily points. This could POSSIBLY be replaced it with an appropriate value between 0 and 15 bonus points. 0 being guaranteed to be unbiased and the simple solution. To recommend an appropriate bonus would require further analysis and modifications to the simulation, to model various random birthday distributions, according to the number of expected **"active" members.

Saying that, going way on a limb, given the likely hood of say 10 being the current maximum of birthdays in any month for approximately 50 **"active" members for the next 12 months, 10-15 bonus points MAY have insignificant impact on the overall outcomes. 15 birthday bonus points would be maximum limit I would suggest. This is loosly based on the limited simulations done for different birthday distributions as shown in the one example for the expected payout and difference from the average payout tables. That value MIGHT work until the maximum number of birthdays exceeds approximately (8-10) in a month for **"active" members. The ONE simulated birthday distribution displayed, hints that MAY be a possible ASSUMPTION.

Goes against what my year 11 maths teacher drummed into students "Never ASSUME because ...."

Probably will only work under strict ranges of variables, according to the number of participants/maximum expected number of birthdays for a month in a rolling 12 month period. Once exceeded, it may require adjustment to maintain balance. A couple of ways could do that either manually once in a while or possibly some sort of weighted system to calculate the adjusted birthday bonus correction, by a few points. This would have to be calculated just before the end of the month.
So depending if below or above the average number of birthdays in a rolling 12 month period, the birthday bonus would be adjusted according each month. Would be a complex system for members to accept, having to wait until the end of the month for their adjusted Birthday Bonus. If the number of **"active" members is fairly static I would expect it to remain a constant value and no need to correct. It's sudden unexpected increases might cause some issues.

**Active members:
Are those who complete the daily requirements to maintain a probably chance for the leader board competition. i.e. The current 50 points for the daily guessing game, if you miss a day of points in most scenarios wouldn't have a genuine chance. (Unless it's your birthday month and 50 bonus points)

Roughly looking at the monthly leader board over time, there appears to be around 40-50 members, on average for each month, that meet that requirement.

Notes:
Might be a case to reduce the daily guessing game bonus points to encourage members to remain active if miss a day or two. With the current (1-10) daily points based on the average expected values for the Top 3 for 30 days, there's about 10-15 points to play with before starts having significant impact on your chances for the Top 3. So if allow up to two days before "disqualification", 5 points or under is a suggestion for the guessing game. If combine that with likely missed the (1-10) random daily points as well, average 5.5 per day, maybe only about 2 points per day, for the daily guessing game. i.e. for each complete day missed penalises you on average 7.5 points lost. Miss 3 days or more, I wouldn't like your chances, of course unless it's birthday bonus that gets you over the line. Lol

For example the top monthly score for August was 215, however they were heavily penalised by missing a couple of days. This score was roughly expecting for the top score (207 for 31 days) + 5 due to the correction on the first of August. Generally once someone misses a days' points, whether through personal or technical problems with the points they most likely to dropoff in activity as a result.

Now couple of major impacts could be:
1) Increases the **"active" member count.
2) Consequently decreases chances for the leader board for **"active" members.
3) The likely hood of multiple winners may increase slightly.

Now for point 3.
Some possible scenarios should be aware with the current system. It's only a small chance, thou significant enough. If can read the table of expected payout for 1200 months of data (100 years) would notice that the average total payout per year varies with various birthday bonus values.

Dividing the values values by 100 gives expected payout per year also for members the average payout received. The variation for total payout per year is about $100 between 0 birthday bonus points and 50 bonus points. This is due to the higher the birthday bonus, the less likely for multiple winners. (i.e. The smaller group of birthdays has more significance then larger group of non birthdays group, due to the bias as it's increased.

The chance for multiple winners slightly increases the more participants, with the tighter scoring system of (1-10) random daily points. $100 may be insignificant depending on the budget. The modelling done shows the probability for multiple winners.

For 0 birthday bonus points: first about 6% , second 12% and third 17%. Overall on average of 11% (about once every 9 months) expected multiple winners for each top 3 prizes.

For 50 bonus points, for the birthday distribution simulated as shown in the table:
Overall it's approximately 5% (once ever 18 months) for a top 3 prize. Breakdown 4%, 6%, 7% for first to third.

Being cryptic, whatever the birthday bonus it could result in an "OPHRA" moment and Banking Royal commission. Hopefully haven't given too many ideas to members.

Do have some possible solutions to avoid those scenarios, thou would be a small chance of them "naturally" occurring.

GOOD NEWS
So the GOOD news for having 0 birthday bonus points compared to 50, from a members perspective, the likely hood of winning a prize increases, due to the increase chance of multiple winners. :)

The chances for those groups disadvantage dramatically improves, of course at the expense of those groups that are advantaged. Only fair!

The periods between not winning MAY be also be reduced with the odds more evenly spread out, instead of all in one basket. Would have to check the expected outcomes/maths to validate that claim. Odds with 0 bonus for a totally random outcome are about once every 16 months for a win for 50 **"active" members.

@admin
The "Compromise" which MAY have validity is 15 or less birthday bonus points. Comes with all the above conditions. WARNING may contain nuts. Consume at your risk. :)

It's a balancing system, if change one variable, like the daily random points you have to recalculate appropriate points values for other variables to maintain a reasonable "fair" system while still achieving the aim to encourage members to be active viewing the competitions posted. The daily points were reduced to (1-10) without balancing the 50 Birthday Bonus points accordingly. This has highlighted and exaggerated their bias, due to birthday distributions naturally being uneven.

Will post up more details explanation of the tables etc provided over the coming week.

From The Wrath of Khan, "Logic clearly dictates that the needs of the many outweigh the needs of the few"... "Or the one"

@Lemond Serious Chill!!

Did you get your bonus points this year and now you don't want others to get theirs?

Since I've joined I've not won guessing game nor came anywhere within top 3. Have you?

You don't speak for majority of the compers, majority haven't commented.

If @admin, want's to remove them then 1st of the yea, otherwise they stay until majority have commented

@Anna2077 Yeh I got my "bonus" points. It was before the June change when they had the comparative value of a couple of pesos. Yeh the leaderboard loves me, NOT!. Well today it does. :) Also no cigars for the guessing game , Fortuna teased me 3 out of 4 weeks couple months back, one away each time and all around.

Won an IPAD PRO the following week. There's Karma for you. OK entering around 250 competitions in those couple weeks might of helped as well :)

It's not about whether I received my bonus points or self interest.
It's about fair competition! Everyone having an equal chance over time.

Warning! I umpire cricket. It's not professional or recommended I show any bias. They are like a pack of hounds when they smell any weakness. lol

It's about the unforseen effect from decreasing the daily points from (1-50) to (1-10) in June. The birthday bonus of 50 points should of been reduced accordingly at least, to maintain the same bias. It's that what is at the heart of the problem.

Any previous bias, due to uneven and naturally occuring birthday distributions, has now been exaggerated to have noticable effects on the outcomes. Favoring those groups of members with less birthdays in the month, then the average, over a 12 month period, then other groups.

From the tables provided, sorry not clear atm, aim to fix up tomorrow evening,
50 bonus points has almost the same bias as the extreme case of 300 bonus points. It's quite noticable in both cases.

For 50 bonus points:
The most disadvantage groups, the ones with higher number of birthdays it's expected payout/probability is approximately halved, OUCH!
The advantaged, smaller groups, approximately increased by 50%. Happy days for thoses cases.

The bias between groups has some mininal mixing between 25-30 bonus points range. The bias is still quite noticable at those values. Around 15 points and below its starting to look more random as it approaches the total random case of 0 points.

So could concentrate simulations from around the 10-17 point range for closer inspection for acceptable weight for the birthday bonus, and under what conditions that assumption would be valid for.

Do have some ideas why at that range it starts appearing more random and possibly an acceptable bonus value. Will explain later after having a good think about it. Got to be careful I'm not attempting to justify an arbitary value to be valid.

As stated previously, and hope the point is clear: The greater the weight of the birthday bonus towards the leaderboard outcome, the greater the bias!

I think it's quite resonable to except the leaderboard a dominately random competition to be fair for all. Then you got the dilemma it's not a "professional" look running a greatly biased competition. Forgiven as it was probably, unintentional. Those bugs keeps @admin on their toes. hehe

@admin For cases of equality, a plebiscite wouldn't be appropriate. That's where leadership needs to be taken. The mob are always free to share opinions. Hopefully grounded on informed ones verses emotive.

Decisions for what is fair or unjust we trust @admin will do their best.

@Anna2077 It's nothing personal and wouldn't expect on facevalue to accept the claim or @admin, it "may" be a bias system. It's complicated to explain the mathematics.

I need some "simple" analagies to cut through the heavy maths lecture. lol

Uneven birthday distributions caused by many natural things eg Xmas and New Year "celebrations" those birthdays should be cropping up around the end of the month or early next month. lol Also just random sample of compers. :)

Attempting by using modelling and pushing the bonus to it's extremities eg from 300 points to 0 demonstrate how it is currently greatly biased.

PS I'm probably better off with the bonus being so biased. That's assuming by next year there isn't a dramatic increase of birthdays for that month and the greatly baised system still going.

@Lemond Sorry, but agree with you

1. Let's just say you were born on the 28th. that means you don't get points till that day.
2. All month you get rather low daily points
3. Others get7-10, hence they are ahead of you on leader board.

With me so far??

Now your 50 bonus 'happy birthday' points takes you to the top and if you are lucky you at least come in 3rd

Why should you miss out on your birthday month?

I've been trying to win an ipad for years and still no luck. One day might just splurge, then again it;s a luxury.. will keep on trying to win.

@Anna2077

@toxicgherkin

25 points - Rewarded on your birthday (just before mid-night AEST).
If noticed the update.

Read your question @Anna2077 had a good thought about it. Can see the conceptual flaws overlooking. First start on the maths and analogy later when done my cleaning and after Internet connected later today.

Revolves around the number 1 rule of probability. The sum of all the probabilities each outcome EQUALS 1.

The probability distribution graph for the accumulative monthly score follows an approximate normal distribution. ( I.e bell graph).
Where the area under the graph is the probability I.e total area represents 1.

First draw a line through the middle to the Base line. ( that's the average point) eg for 30 days and (1-10) daily points ( 1+10)/2*30 = 165.

If marked each outcome on the base line ie the accumulated monthly score (ie. 30-300) . Ok for the leaderboard monthly example need a big floor to draw on. Now due to the shape of the distribution ie it's represents how the probabilities change. If draw a line from the a point on the graph to the base the longer the line represents the bigger the probability. Eg score 30 very short line be a dot, score 165 the longest, and other extreme 300 another dot line. I.e the probability isn't the same for all monthly scores. It DECREASES THE FURTHER AWAY YOU GET FROM THE AVERAGE.

Now that point also helps me understand how the bias is working. I.e the further the away the number of birthdays in a month is away from the average it bias increases/decreases. Now there is a relationship between two statements. Ok need to clean. Edit later and should of started with the analogy involving pies and pyramid schemes. Ie when you start guaranteing someone to win it starts taking on the shape of a pyramid scheme. Missing the whole picture everyone expecting the same possible outcome verses focussing on one person. Ie guaranteed one or a group at the expense of others. (Bias)

@Lemond you are overthinking it...

stop using maths and start thinking logically

@Anna2077 lol yeh putting up my thought processes. It's handy. Good reference. By "debating" it helps me understand whats happening.

Ok simple demonstration. And hard to get away with no maths to explain how trying to manipulate winners by guaranteing effects other participant. Outcomes.

Similar to a Pyramid Schemes and how they work. The winners are the ones at the top of the pyramid. I.e The few benefiting from the ones below and that continues down the ladder to the bottom feeders with the left over scraps in the scheme/competition. If you advantage one or more groups it's at the expense of others.

For 50 bonus points the few are the smaller birthday groups that would of got up to a 50% bonus and each level in the Pyramid towards the average gets incrementallly less of a bonus. On the other side of the coin the further away from the average the incrementally disadvantaged until to get to the maximum amount of birthdays.

Now that that logic helps work out the challenge what is a reasonable birthday bonus. Picking an arbitary value can have undesired effects from major bias to physcological ones. Those physocological ones are also important to attact and keep members interested. Remember it's all about the promoter.

Would need to recheck the odds for birthdays and non birthdays. If the non birthdays odds are too low (i.e.) length of time to expect a win, it may having of the effect of discouraging members for the majority of the year, and only get motivated when they expect a realistic chance. 50 points could of had that effect. Apart from August expection of 0 birthdays. Once there are 4 to 5 birthdays in a month for *"active" members non birthdays had a slim chance. Maybe if a few slip up and miss a day. Stephen Bradbury moment. lol Lesson there hang in there you just never know. Just don't expect that to happen too often. :)

Yeh got FTB installed. :) And the leaderboard loves me two days in a row. Now there's a first. Hopefully not too many birthday's or need of few of them to slip over.

Now for the new 25 points will check the spreadsheet model see what it can tell me about the birthday/non birthday odds. Anyone got any ideas how many birthdays to expect this month?

For the tables I put up the scenario was maximum of 10 birthdays. That random distribution could of been more of an expection. Why would have to run that 1000's of times to make realiable claims about a birthday bonus amount.

My gut feeling it might work up to about 6 birthdays maximum in a month in a rolling 12 month period. I think I now how to work it out from what I said from above. I.e. analyse the variation of expected payout for each birthday number group, according to the average number of birthdays in a month. The sounds a lot like the coefficent of variation. Used to analyse trends etc. I was eyeballing that trend for the obvious bias. Once started dropping it below 25 birthday bonus, tehnically hard to "eyeball" and make conculsions with becomes les reliable. The assumptions for the variables become more critical to make any reliable claims, without using mathematics to anaylse trends. That I haven't implement YET.

Give me time. And migrate it all in a database to automate the process for random simulation of birthday distributions. :)
Thou I got 5,000 emails for spring cleaning. Might be some wins in there. Probably too late to claim.

Would help if had what the expected maximum of birthdays would likely be for the next 12 months. We had 6 in june. Also the number of likely participants which gives the average number of birthdays for a month.

Also got to allow from some wriggle room for growth. Why working from 10 birthdays might be a good idea. Most likely wouldn't need to adjust the bonus for quite a while until that limit got passed. Disadvantage bonus is set a touch lower. Advantage non birthdays keep interested with better odds. Disadvantage birthday months whinge why didnt I win on my birthday month. lol.

Then you got the ones that never win, or the ones that win consecutively for a few months in a row. Now that case last year, thou never officially stated at the time, some "bugs" have been detected during earlier this year and that first of last month which got corrected. So hopefully that bug has been exterminated. :)

I wish I could say the same for my cockroach war. :( Damn saw a couple this morning. Good news thou they are only bunkered down in one area.

Now if anyone got any suggestions for those bugs?

Done the spring clean over a month, that should help. Been persistent over 2-3 months, slowly have reclaimed fridge, bathroom, laundray area, bedroom, walls, ceiling, microwave, couch, tv unit, stove and oven. Just the kitchen in one corner near the oven. I thought I had reclaimed the kitchen. There's nothing to eat in the cubboards I removed everything to lure them out.

@Lemond back to my example... you only get daily scores under 5 for 20 days on your birthday month. It will never get you into the top 3, so that you get something for your birthday,

Your birthday falls on the 28th as an example and suddenly you are in the top 3. You get a small gift.

Without that birthday 50 points you'd still be trailing somewhere and people who don't have a birthday have been lucky that month and received gifts

it's September. You've received your gift . Don't take it from others. As @toxicgherkin told @admin that 1st January is fair.

@Anna2077 Ok 50 bonus points 5 perday is just below the average. 30*5=150. For 30 days expected top score is 200+_ 7 So good chance might be above the top non birthday. Be close those competitions view difference would most likely make the difference in that case.

@Anna2077 What gift? And you can't compare 50 birthday bonus points for the old system of 50 daily points compared to 50 birthday bonus points for the (1-10) system. It doesn't translate due to mathematics and distribution of scores. I.e old system average for 30 days would of been 30*(1+50)/ 2 compared to 30*(1+10)/2.

And it's been never about self interest. About the system been fair for all. If change the daily points to whatever eg (1-20) ... (1-100) or (1-1000) etc etc. If wanted to have the birthday bonus to have a comparative effect it needs to recalculated. I.e. would have to check the percentage rank influence of 50 points in the previous system. Then look at expected scores at that percentage rank in the current system and the difference from the top position would roughtly relate to the new birthday bonus. This would have same comparative effect as previously. The advantage scores are now closer with (1-10) over the old system. So can understand why might of changed to make competitions view more important.

Could of also left it at (1-50) and work out what appropriate points for each competition viewed would of had the same effect. Complicated. It's a fine balancing system when changing the random daily points systems. Where need to know your maths or do some simulations to calculated expected month score distributions.
It's not incremental relationship in a straight line it follow the Bell graph because it's a random distribution. :)

@Anna2077 Hmm fair for ALL was scraping the 50 points are at least reducing the greatly bias system ASAP. The only way it was fair was under strick conditions. One equal birthday distribution which is improbable. There is more cases didn't initially think about. Equal groups either side of the average number of birthdays. Still unlikely. The more the bonus is reduced towards 0 it squashes the pyramid scheme until certain of no bias and works for all scenerios. From 2 compers to billions. Like lotto. 25 points might work well under certain strick conditions. It may have weakness depending on number of participants, max birthdays. Depends on if expect a stable growth or low growth of particpants.

@Anna2077 Your logic for continue a greatly bias and unfair system, which didn't even work well for a low number of participants is flawed.

If had the approach for umpiring cricket, if I make a error of judgement or one of the laws of cricket then recommending, I should continue that making that mistake at least for that game, or until a number of months later. That way all teams get "fair" treatment. Well at least "consistent". hehe OMG

I could just imagine the umpires report. Make for some good reading. It would be unprofessional. When I use that word that's being dipolmatic. "Unprofessional" on a scale 1-10 is 10.

@Anna2077 Using a simple example.

Say 36 particpants, 36 prizes awarded each year. For simplicity no mulitple winners.

Might be a recommendation there to control payouts blowing out in some "scenarios" most natural and not a huge issue depending on if the how tight the budget is.
Could potientially be "unnatural ones". Best not go too much into that scenario. I call that the OPHRA scenario/Royal Banking Commission.

ok if even birthdays each month with maximum bonus they all get on average 1 prize a year and over 3 years on average 1 of each prize.

However if say a couple of months had 0 birthdays and couple of months had 6 birthdays those with 3 get on AVERAGE get 1, and those in the groups of 6 birthdays on AVERAGE get 1 prize every 2 years. Or on avergae one of each prize EVERY 6 years. You see the problem there? The smaller groups below/on the average number of birthdays are better off on average, then those with greater, then the average number of birthdays. If was smaller then 3 birthdays then they would do better then the groups of 3.

Why 50 points, which almost the same as unliminated bonus points bias was a problem.

@Lemond Sadly you are missing the 1st of Jan factor

It's even sadder that the argument has to only be your way....

you've never complained about it till now, have you?

So settle down, chill out and let people get at least a small chance of winning something.

For instance I hardly ever win comps, but still trying... have never guessed the lucky number, Never came within top 3

Many compers are like me... so do you believe that we should continue like that?

If you are going to throw bias around then take this:

1. A lot of competitions that ask you for age, do not let anyone over certain age win.
2. A lot of companies go by the name and only some names win
3. 25 wol doesn't always depend on how good your answer is, but if you are male of female.

the list goes on.... you cannot always have it your way when suddenly you decided that something is not the way you want it to be.

Have you noticed that no one is replying to your posts? Apart from me?

@Anna2077 lol. It's not is argument one way. I can't see a logical or professional reason to continue any unfair situation. Call intregity, professionalism. If only could give everyone a prize a year. Sadly have no control if more then 36 people of average enter the leader board competition. hehe

It's about EVERYONE having EQUAL chance to win over time. Verses some with greater at the cost of others with less. Thats a fair and reasonable point. Simple as it gets. Called competition.

If make a mistake which i.e. umpiring would acknowledge and try and correct, it and hopefully doesn't happen again. Instead of repeating it on purpose for "fairness".

The admin has acknowleged the unintentional great bias and made some attempt to correct it, with the limited information at hand. Possibly picked an arbitary value. It's a more reasonable value for the (1-10) points.

Long as aware of the flaws and what circumstances it works for and when it starts to fail. They need vist to a maths professor to get pick their brains. Get the appropriate math formulas out or simulation to see if has integrity under the anticipated conditions. EG number of participants, max birthdays, etc etc.

If 50 points was intentional for (1-10) they most likely weren't aware of the flow on effect when the system was changed. It's a finely balancing act when start changing the variables available. Eg For points for competitions viewed, points for the guessing game, birthday bonus points.

For those decisions thou tempting to dive in and change them. Need first decide why want to change and desired effect. I.e. promote use of site. The lure a reasonable fair chance to win for all. Then model excepted outcomes or calculate them and see if maintains integrity within the expected limits.

@Lemond So you are saying you'll continue talking about the unfairness until 50 birthday bonus points are gone?

You've never complained till now... so ask yourself why now and not before?

@Anna2077 It's gone the 50 bonus points. lol. See the updated conditions. It's now set to 25 since yesterday.

Thought I had posted that. Was tired last night and PC crash and busy this morning.

@Anna2077 Because it wasn't apparent the bias. It was a lot less obvious with the (1-50) random daily points. You can't compare then two systems on facevalue for the birthday bonus. Major % difference influence of the 50 birthday bonus points. One is about 20% the other about 80%.

It was only after the change in June then midmonth thought calculate the probabilities for birthday/non to see what the if was any realistic chance for non birthdays. With 6 birthdays in a month for non birthdays group lol was 500-1000's of months for a likely win. 0.5% or less. At 0 points its once every 17 months. 6%

Then during August when 0 birthdays started looking likely. I realised the probabilities I posted where only valid for once scenario. So I expanded to calculate for every month according to a random birthday distribution. Was after a week or so the BIAS issue clicked. Probably looking at expected outcomes and thinking what was the cause? Why some participants seem way better off then others SO much from the modelling. What was relationship causing the BIAS. It was noticably not within random variation.

Could work it out. lol. God would have to generate random monthly scores based on (1-50) then paste that into the model. I've had to separate the data generation from the model due to the PC complaining.
Would take me a 1/2 hour. Could do that for all possible daily random points. I should dive into the mathematics. I do know roughly or could google what formulas to use. Too much brain power to calculate and chance to make mistakes.

@admin Thankyou for looking into the Birthday Bonus issue and adjusting it to a more appropriate value to reflect the changes made in June (1-10) random daily points. I hope it hasn't caused too many headaches and has been interesting reading. :)

Become a bit of pet project since January on and off. Aim has always been to be constructive and suggest improvements and/or highlight possible flaws, if possible appearing to behave in as an "unfair" system. Eg the possibile first day bug issue and the early birds v the rest. :) .

Hmmm where do I start for more possible tweaks. Apart from the birthday bonus it's probably a touch high. Will know when get motivated to look the model. The good news should be distracted catching up entering competitions for a while. :)

Hope the leader board has long and posperous life. Maybe even pick up a win a two. On average expect to be about every 18 months depending on number of *"active" members of course. Hopefully get lucky in the next 12 months. Still waiting on my welcoming gift.

@Lemond guess you being selfish and complaining cost many people myself included ever getting to the top 3. Well done!!!

@Lemond You have way to much time on your hands.. time to find another hobby

@Anna2077

1) Not getting paid to look into leader board issues.
@admin what does the reward for spotting bugs pay? Based on random value greater or equal to $0? hehe
2) Had a couple months off comping so had plenty of spare time.
3) Previously had focused attention on 1st of the month bug that alledgely popping up.

lmao. Seriousl! You must of missed the "Early Bird's" V the Rest alleged issue early this year.

I usually log in very early. It's realated to best strategy for the guessing game. It only help chances by getting in early. Would be nice if I actually guessed right. That responisbile has been now passed on to random.org :)

In January I started looking into the leader board. It's been an interesting project. hehe. There appeared to be some abnormilites from when I joined from September to December. One being never could get into the top 20. The other never ended higher up then @toxicgherkin for months on end. Not even for a day. Ok got to equal score one day. Then sent tumbling down the ladder. hehe. Thou most likely could explained with random variability and bad luck. Thou looking at your own scores out of context doesn't have any validity. You got to look at the bigger picture of a larger group of scores.

Then for 3 consecutive months at had same group of winners at the top. Extremely improbable, thou possible. Probably better odds for Lotto then that happening. Also the distribution of monthly scores didn't "smell" right. They seemed consistently higher and skewed above the average the same amount for a few months. Possibily due to, like last month, on the first day the wrong daily points was issued. Then it was corrected if recalled. I was suspicious from January something wasn't right just causing the skewed results above average consistently.

Not till February the 1st, I noticed what was the likely cause. Early on the day low scores in range only (1-10) till 8:29 am, then range (40-50) till about 5pm. Then looked random for the rest of the day. It was more by chance spotted it by logging in early. I was suspicious before that something not right thou. Hightly improbable to see a pattern like that. Can't be explain away within random variation.

I actually recorded the approx login time for most members, from first competition viewed log for the whole day. OMG . After the first day it's impractical and hard to spot if an was a daily or reoccuring issue. May of been an reoccuring issue each month on the first day unnoticed for a while. So possible explanation why seem like patterns for groups of members doing consisently better then others and skewed above average. I.e. the early birds v's the later ones. At least if there was an issue @admin was then aware of it or probably keeping an eye on it.

The skew above average can be explained with ie. more members log in later in the day then the early birds. Like having extra birthday bonus for most members each month. So unintentional due to human habit of times when logged in. Some members may of consistently been fortunate.

Coincidently Feb 2nd the system went down for maintanance and from March to May the monthly scores for the first time appeared ok. The skewed results above the average had disappearred. No offical word on that until the first of last month. When the bug reappeared with the daily scores being incorrect ranged and fixed the following day. @admin if I highlighted that issue any chance of a bug spotting rewarded. lol Hopefully that bug has finally been put to bed. :)


Some background
An old job was for implementing, basically solely, a financial modelling system for public and industry funds. I'm the IT side with some maths. Didn't design the model. Thou implementing/living and breathing it over numerous years got understand it, and the maths behind it. If you change the importance/weight of one variable, it has consquences on the overall output. I.e Was the Rating of the Funds.

Early on recall the boss had rush off to a Fund Manager to show off the reports for whatever reason. Realising that something wasn't right on the way out. Became of game of anticipate the Boss's requirements.

The rating system he designed wasn't showing expected results for the performance for those funds. First reaction it must of been implemented wrongly. I'm way to through with testing and modelling. It was that his design was flawed. Wasn't a major mistake, thou it's impact had major consequences on the overall Rating system. One variable was weighed wrongly or having the opposite intended effect on the model.

Designed the system to be flexible enough to cater of any request for mincing the stats for adhoc reports. In those days PC's were slow. It took 24 hours to process all the data and produce all the statstics for reports. Once a month new monthly data would arrive on cassette (the dark ages of dailup hell). Would come in over the weekend, on the Sat run stats over the new data checking for any obvious errors to correct. Where variation/standard deviation comes in handy. It highlights the obvious errors. EG fund price instead of $1.23 is $123. Data errors would distort the whole modelling for that Fund Category. Then once happy found most of the errors would rerun the model on the Sunday night and go home. So by Monday morning the Boss could happily play with all the new stats for new adhoc reports. Another IT person did most of the report designs. They would be published in a monthly booklet and a yearly book.

The modelling delt with dealing with large number of fund performance data then had to group into appropriate categories and compare performances and their variation etc. Then use that in a weighted model to rate each fund. EG combine performance compare to peers, risk/variation , also compare to industry indexes like ASX etc etc. Then combine performance measure with finely balanced weights, to highlight more important statisics for the Fund Rating.

Perfermances of funds categories follow roughtly normal distributions. You can analyse things like are they adding value for their service/fees according to benchmarks like Bank Bills, ASX etc.

So bit of intuititon when a random distribution isn't behaving as expected. All things being equal everyone should have the same chance to win over a period of time.

@Anna2077 lol yeh comping. Cricket season coming up. It's been winter hibination.

@Anna2077 Gee "selfish". That's harsh. Won't take it personally. Will have to do way better then that to offend me. Remember I umpire cricket. lol. All complements.

Hmm complaining. I wouldn't put it that way. Wonder if the suffragette's were thought of just annoying complainers, for a justice and equal rights to vote. Thou it did take them many attempts and petitions for politicians to eventually see the light.

At least politicians didn't put it to plebiscite.

@Anna2077 I can only infer from your emotive remark that you perceive you have being disadvantaged by the change, or possibly you've got a birthday coming up very soon. Or depending on which side of the coin you perceive would of landed on the favourable or disadvantaged from reducing the bias to a lessor amount.


Approached as objectively as possilbe. Let the simulation do the talking. Thou in your defence you don't the spreedsheet modelling in front of you of examine, nor admin yet. :)

If I thought I was being advantaged over anyone else due to the changes, I would of said something about it. Probably as my month had 3 or maybe 4 birthdays might of better off under the 50 bonus points. Good chance better off then the 6 in June group.

When I noticed the possbile "Early bird" issue on February 1st, I could of tried and taken advantage of it. I was aware that even with testing for it, and if was reoccuring. I could be taking advantage with that knowledge. That's why I posted that publically. At least if I perceived I may be advantaged, everyone else had the same opportunity to use the information according. I.e. delay logging in early. Confirmed last month might of been issue early on the first day. At least if was an issue they are aware an ontop of it. :)

The modelling just followed exactly how the leader board works. I'm fairly confident implemented correctly. Maybe some doubts regarding what happens with payouts for multiple winners. Thou that doesn't have an major impact on statements made regarding the obvious bias with too high of a bonus.

Thou suggested removing the birthday bonus to 0 or suspending it until an appropriate value could be decided or calculated. Was trying to be prudent in not suggesting an inappropriate value. Trying and see what value mantains the integrity of the leader board competition. Had to be careful own personal preferences for a value and justify a birthday bonus value to fit a case. Only limited to a few sample birthday distributions. Not where not enough data to suggest appropriate range of values.
Also don't have the stats the the admin would have on likely expected growth of members being active. Might be a touch of growth in the last year.

I did suggest some of further analysis at the points where limited data was from facevalue looking promising.

The modelling of outcomes doesn't have any feelings, just 3,600,000 random numbers to generate the expected monthly scores for 100 years. Currently it is however limitated to fully analyse what an appropriate birthday bonus ranges would be. Can only eyeball what ranges might be further analyses to focus initial. Obvious bias can be seen when the outcomes are clearly grouped within birthday number groups and then compare that 0 random totally random case. Just in case they happened to be in the same order. It would of been inappropriate to make an outright claim for a birthday bonus value without anything concrete to base it on. Need extend the modelling and add extra statistical analysis.

Unconstructive "self interest" thoughts. Sounds like someone needs a win badly. Or a short break from comping. Does wonders. Helped me for April. :)

Wasn't until April had a couple of wins. After almost continual comping from last September. Took a month break think in Feb.

I know can get frustrating. When you least expect it a win seems to pop up. :)

Good luck.

@Lemond you are still carrying on!!

Long winded posts that constantly say same thing are on same level as very long web pages = no one reads them in full.

You might think that you did the right thing, but you didn't. Everyone who had a birthday in the first 8 months received 50 points, for the last 4 months it will be 25.

Would it have ruined your life if the new system came into play from 1st Jan?

Now you don't know who had or didn't have a birthday in August. As you don't go through every compers account. If you do then you have way too much time on your hands. You can only see around 35 people, not everyone... not everyone play's the guessing game either.

The thing that bugs me is that you spoke for everyone, no one elected you speaker.

@Anna2077

The @admin took the appropriate steps, to the best of thier knowledge, and with the information available at the time, to rectify the unintentional biased leaderboard system changes. Possible the new value a bit arbitrary.

Guess a failure to educate you why a change was necessary ASAP was lost in your sentential and emotively based argument.

Where do you begin about why an ASAP 'product recall' would be appropriate for a demonstrated faulty product, and taken off the self, to replaced and repaired.

It's not something you vote on. It's not a religious argument or about what colour to paint the walls. It's based on complex maths. If anyone could of provided the mathematically proofs/calculations it would of been appreciated. Thou mathematics can be used to help calculate the probabilities it's a complex weighted system of numerous variables. A lot easier using random simulation and modelling. One of my fortes. :)

Using simulation based on numerous random samples is the Law of Large numbers.

See https://en.m.wikipedia.org/wiki/Law_of_large_numbers.

I don't need to know every account. Just the rough amount of ones that do all of the daily requirements, to have a genuine chance for the leader board competition. I.e. Missed a couple of days of the guessing game etc. It helps me to make reasonable assumptions for the modelling. Ie. the number of participants/active members. Thou allowed for up to 100 participants in the modelling. Spreadsheet has it limitations, so does my PC.

Possibly the weight for the guessing game could be reduced as suggested previously, if they felt it might encourage members to keep active. Has some merit that suggestion. Then it a lot harder estimating the figure directly. Thou the excepted mean score for 30 days would give me a good indication as well and what position it translate to.

For the last two months because it was so biased birthdays were likely to appear on the leader board, especially early in the month. There were a couple you may of not noticed. I noticed as it had an impact for the probabilities I posted up in June, especially if were no birthdays. I was hoping there would be bunch to demonstrate the effect on the likely outcomes. Didn't expect the scenario for none to appear at the top was likely. Least by them all slipping up it was a equal chance for all others.

Regarding the 50 birthday points, the first 6 months system of (1-50) random daily points CAN'T be compared with the last 2 months of (1-10). Like comparing a paper plane to the space shuttle.

Most likely, the 25 birthday bonus for the (1-10) daily points has more value/weight then the 50 BD points for the (1-50) system. Would have to check the model with (1-50) points per day or calculate it.

The simple logic you provided, translates as: if it was (1-50) before and now (1-10) it 'should' be changed to 10 points ie (50/5) to give the same relative value. Maybe it's 10points, unlikely thou. It doesn't work as simply as that. See normal distributions and calculating expected values. Educated guess be around the 20-25 points.

Well it was for the @admin to decided the appropriate action. It's thier reputation and integrity on the line in that case.

Also other people eg @spyglass and others had commented previously of the their concerns at the time. Without the modelling at the time the information wasn't available. Before making any statement or claims, I had done the research and modelling, and backed up with the mathematical methods to highlight the obvious bias. It's not something I worked on over night, put it that way.

The leader board is an interesting/complex probability question and has a lot of similarities how it functions with a previous fund ranking modelling. Pulling things apart to see how they work can be educational and a 'fun' hobby. Putting them back together ... takes patience. Did I mention that umpire cricket? Lol

Maybe if the @admin had announced the upcoming changes before June could of looked into the side effects. It appears was an unintentional mistake. If want to point fingers or any grievances your energy is focused at the wrong person.

I accept your apology in advance. :)

Knowing the rough probabilities for competitions you can make decisions how much effect is it worth putting into it or what your realistic chances are.

@fourleaf made some great comments regarding whether consciously/unconsciously they focused on types of competitions more likely to have a chance according to their forte.
Is the effort worth the reward. Eg Spend a hour for a WOL for steak knives with a 1,000 entries or an hour for a WOL for a car with a hundred entries etc.

Initially was examining the probabilities of the new system for winning in a birthday month verses non birthdays as were comments unhappy with the change. The odds had changed dramatically. Impact basically meant don't bother unless less then 4 birthdays ie 6 in July. August with 0 birthdays on the top of the leaderboard made be upgrade the basic modelling to calculate them for every month according to a randomly simulate birthday distributions. This was based on the assumption of 50 genuine contenders. With flexibility up to 100.

@Anna2077 And Yes. Will probably at a later date will 'carry on' with suggestions to help improve the integrity of the leaderboard competition once all modelling/analysis and reports completed. Definitely if there any 'obvious' issues with it. It's had teething issues.

May even publish the model on a site so you play around with variables at some point in time. At the moment not user friendly. Getting there. :)
Be an interesting project refreshing/learning skills.

That way @admin can play with suggestions in a model before making changes.

I would like it to continue where everyone has an EQUAL chance over time, while still able to achieve the aims for the competition..

Thank you for your appreciation. :)

@Anna2077 lol no wonder so cranky 7 daily points for 3 days. Hmm Karma. Early days it can only improve. :)

Wouldn't also be your birthday month or before January by any coincidence?

Good luck

@Anna2077 Have you have a profile name change Annaar?
For many reasons I recognized it could be you. :)

Good Luck when you come back