Forum

@Lemond knowing my luck as i'm near the top of the leaderboard all the August birthdays will appear during the last week of the month :(

@spyglass lol. Yeh the birth stats in Aust are fairly even thou I think August to October slightly higher. Ie New Year + 39.7 weeks .

I noticed a couple of birthdays earlier in the month looked liked they had missed least one day. :).

I've updated the model for the leaderboard with dense rank for each month, including leap years, randomise/manual birthday months and sorted, inputs for birthday bonus, number of participants up to 100 and number of prizes. I had to split the data generation from the rest due to old PC so modelling different daily bonus would have to manually copy the data set in. Then stats comparing birthday verses non birthdays wins even for each month. Handy as sort the birthdays and can easily compare how different number of birthdays in a month compare. It's evolved lol.

50 bonus points from memory is about 75-80% bonus. So if 1 birthday most likely end on top 3 time out of four years and the other year 2 or 3rd. The mean for 31 days is 171 (31*5.5) . The top score around 206 +or -10. So a monthly score for 31 days around 156 is approx score needed to beat the top non birthday score. So for a non birthday you need around 210+ once you get a few birthdays, you might scrap in with bit lower if got a few extra points from comps viewed. You'll would need every single point.

For this month looking like there might not be any birthday members that have done all the guessing game. You might be in luck.

I think someone ask last month the odds for 5 birthdays at the top.
We'll probabily 3/4 a success and can work the rest out using combination probabily to get the result etc. for 0, 1, 2 .. 5 etc etc

Now the major issue apart from the overzealous birthday bonus of 50 points means the probabilities for each birthday month are virtually 1/(number of birthdays) for each of the 3 prizes. Eg if 6 birthdays just below 3/6 chance of a prize. Note if only a couple they sharing first and second most times.

The more birthday members in that month the more you have the share the three prizes with on average. So if only one or a couple in that month it's a cash cow compared to other months. Ie it's a baised system due to having the birthday bonus so high. Non birthdays usually wouldn't get a look in unless 0 or only a couple birthdays and then most likely be only 2nd or 3rd.

I stats I previously put up only looked at an average number of birthdays eg 5. Improbable would ever be evenly distributed over a year. The result is some members are greatly biased towards compared to others. It's can be multiple times put it that way.

Next few days will see what birthday bonus might be a 'fair' least remove most of the bais. Most likely under 20. Easiest solution scrape it thou preference for a slight birthday bonus.

Question is why have it at all. Promotional exercise to lure in 'new' members Definitely discourages been active for most of the year except birthday month and maybe this once off. Least till the August members wake up next year!

Also with the dense rank method used and extending to 50 places the could equate to 70-100 members with the scoring system so tight.

@Lemond that's a lot of work there! I'm happy if they scrap the birthday bonus......but only in December :)

@spyglass....then I'm out of the running! lol Cancelling NYE's celebrations of my birthday bonus forthwith! ;)

@toxicgherkin @spyglass lol. Hmmm yeh bit of work for the modelling and statistics. I've also added in expected payout per person via for each month. Might also add in group the payout according to number of birthdays in a month. Would save me having to sort that to compare.

All data driven flexible to model most things. I possible could modify/add for different ranking systems. Site uses dense rank could also compare to competition rank. With the tight scoring system and how prizes are awarded, I would be little nervous if I was paying out in some scenarios. All I'm saying is OPHRA!

I've been slowly doing the model during my self imposed comping exile. Waiting on Wondercom for an Internet connection . Still wondering after 15 working days for a connection appointment.

Just making sure the probabilities are calculated correctly, I rushed up last months calculations. Also adding some statistics to help analyse what an appropriate birthday bonus, if any would work. Then need to expand all the sheets to take in a larger sample. Lot easier using a small sample until got the model done.

I've got some ideas on a birthday bonus and how might work. Think it could be related to the average number or likely maximum number of birthdays in a month then examine the average top position monthly score and subtract from it the birthday bonus.. Would have to consider also how well it scales with greater numbers.
Eg from limited data set working with atm.

For 50 participants:
Assume likely max birthdays is 7 then 1/7 = 14%
For 30 days the top 2% score ie first position the average score is approx 199, 14% is approx 182.
The difference 17 points. Would have to model it to see the effects for if 1, 2, 3... birthdays in a month and how they compare for expected payout per person. Also comparing against no bonus.

For 50 bonus points
For the one birthday example verses other months the preliminary results showed it over 3 times better off. It went from being one of the worse performers with a totally random system 0 bonus points to clearly the best with 50 bonus points. Note only a small sample of 20 years.

If you think about it, if it's set to high on average most/all of the birthdays end up at the top of the leader board and share the prizes available.

Eg if one birthday first on average most years
2 birthdays every 2nd year etc win/ alternate 1st and 2nd etc etc

Eg For 50 participants
For 0 bonus
Odds roughly 1/number of participants to win 1/50 (1 in 50 months, 4.2 years) . For any prize 3/50 ( once every 17 months). Or odds for winning on your birthday month once every 50 years, once every 17 for any prize. Odds are slightly better due to the increased chance of multiple winners with the tighter points system.

For 50 bonus points:
1 birthday win every 3 out of 4 years and then other most likely 2nd or 3rd. I think under 8% chance to miss out totally.

If got 6 birthdays in a month be 1/6 (1 in 6 years) to win . For any prize 3/6 or once every 2 years on average. Chance to miss out would expect to be low ie probability of 3 members out of 6 with each roughly 75 to 80ish% each to be on top of the leaderboard. Feel free to calculate that see combination probability. So basically if you going to miss out, your in the bottom 3 birthdays anyway. Lol

And if more birthdays in a month! hopefully you get the idea.

So for months with less birthdays then average, those indivuals should win more often. So the winners of the current system would be biased towards those members.

So you first question should be: Is your birthday month better off then the average? Then decide! Lol. Otherwise your better off with a no birthday bonus system.

@toxicgherkin Will have to drum some support for December birthdays to come out of the woodwork! That way you can whack a mole them before they multiply.

Also if touch unlucky what would be the expected worse case scenario for how long it might take to win comparing systems? Are you better off putting all your eggs in one basket once a year? Definitely if only a couple of birthdays and 50 bonus points. Thou if 10 birthdays might be a while between drinks. And that's hoping they don't increase too much over time.

I was thinking of putting up a non binding poll once completed the leaderboard model and statistics.

The problem with plebiscites, self interest and uninformed opinions may overrule the common sense for a fair system. That's where the @admin need show some leadership. Lol

@admin The birthday bonus should be removed until you're confident that it doesn't cause a biased and unfair system. Possible a reasonable bonus amount might only have an insignifant or limited bias.

Reducing the daily points I can understand as increases the importance or viewing competitions, however the birthday bonus wasn't adjusted correspondingly. If you pull one string it can effect things unexpectly.

I'm not saying revert to the old system. It might also had some bias will have to generate the numbers for 50 daily points. Educated guess boost was about 20% which corresponds to roughly to less then 20 points for the 10 daily points. You can calculate that for n participants for 1/10 and 1/50 probabilities for an average month of 30 days etc to get expected values. See PMF and z values I think. I haven't dived into the mathematics as yet. 20 points is probably still on the high side. It's best I don't prompt an arbitrary bonus points until I've put a large set of data in. Overall it involves some heavy maths!

At the moment using only a small sample of 20 years on the updated monthly modelling. Intend to expand that 10 times the amount. Shouldn't take long unless the PC complains.

Before making any decisions best to model or get the maths brains out. Hints how many participants on average do all the guessing game, and birthday distribution, PMF and expected values for an average month, combination probability and a dart board.

I don't claim to understand it all - my statistical days are well and truly behind me - though I hope @admin and @zedmin are perusing Forum content. It would be nice to hear from the site's admin team every now and again.

How about running a monthly 'Caption This..' competition. Introduce Weekly Riddles, Enigmas, Puzzles and such, just to engage users more. The Forum has dried up a tad in terms of numbers of respondents. Let's open the doors and invite everyone back to have fun, contribute, and lift the site's social profile. Thoughts?

@toxicgherkin Bit rusty myself, why took the 'easier' path of doing a spreadsheet to model it. Just the last couple of weeks with no birthdays of regular members compared to last month a light bulb moment. As expected the sample data confirm it's biased greatly. My feeling even if a bonus works probably will need tweaking occasionally, if get substantial increase of core members doing the leaderboard competition everyday or influx of birthdays.

You would think with such an advantage members be coming out of the woodwork for their one shot a year. Maybe they missed the big advertising banner announcing the changes. Only noticed one 'new' member as a result of the changes and they have since done a runner. So from a 'promotional' point of view it's a failure so far! More likely to kill off the site and turn members away. It's kinda the glue that keeps members coming back regularly.

The forum has dried up. The capation this competition used to have bit of a regular following on Facebook. That's a good idea or something similar. Maybe a photo comp. Creative comps are a good change of pace. Also another forum run competition. Or find the hidden bugs in the site lol.

Last we heard from the @admin they were exterminating bugs by unleashing some cane toads! New month approaching hope they got the exterminators on standby.

@Lemond I agree, this forum used to be fun

@Anna2077 @toxicgherkin @admin
It might reflect a drop off in members actively using the site.

The leaderboard, guessing game and forum differentiate it from other comping sites. The leaderboard is glue that encourages members to actively using the site daily. The guessing game while may attract members daily , doesn't necessarily encourage active use the site. Thou you think they would explore a bit. Why question some winners of the guessing game occassionly, especially if are new members. Least question for the promoter's side why the lure occassional fails. Guess more about put out the lure and see how many you can catch. Bound to be a few that get away.

The website glues together the other social media sites they use. Like all comps it's all about the promoter! How to attract members and keep them interested. Other exclusive site comps are a good idea as well, thou as not daily keeps them on their toes only once in while. Be only short lived and would need regular commitment for them.

Bound to be some effect from having the leaderboard becoming primarily a "birthday" competition. As it's set disapportionly too high, apart from possibly discouraging members for most of the year, it's an biased and unfair system. Ie the the number birthdays each month unlikely/improbable/impossible (unless number the of *'committed/active' members is divisable by 12) to be evenly distributed.

*regularly do all requirements daily to have a genuine chance. Due to the high vale of points for the guessing game once miss a day realistically out of the running. (Unless birthday month and not many other birthdays)

Done some reading, looks like the modelling done so far is leading towards a Monte Carlo simulation. That might be a good idea to start there, instead of dealing with the complex maths involved. Also there are other similar methods that revolve around the Monte Carlo Simulation technique.

I will have to read up on it especially what statistics to generate. Guessing coefficient of variation for a normal distribution for wins or expected payout.
Aim would be find the most appropriate birthday bonus or other what other the variables can alter appropriately. Can play with random daily points, number and value of prizes, points for competitions viewed etc. They are all things "play" with. Pull on stick and see how it effects the outcomes, before creating a system the falls in a heap!

Could to like higher daily points, increase points for the competitions viewed, decrease points for the guessing game etc as an example. It's a fine balance depending on what your aims are. Guessing encourage to view the competitions posted. Lol. For appropriate values you need careful selection and testing.

Ps too high of a birthday bonus falls in a heap!

What is everyone's suggestion for birthday points? Remove them?

What's the problem with birthday bonus? If people change birthday date to get extra points the system would pick it up.

Without the birthday bonus some will never get to the top 3.

@Anna2077 It's a biased system! Ie it's favourable to the groups of members that have on average, less birthdays in thier month, compared to those with more. In one word UNJUST.

The baised system is more apparent since changing from 50 random daily points to 10 and leaving the birthday bonus as 50 points. Assuming the previous system of 50 daily points had an insignifant bias, if the birthday bonus was reduced to reflect the changes it should produce similar outcomes. It not as simple as dividing by 5. You would need to understand the distribution of expected monthly scores and their variation for 50 daily points and 10 daily. I'm not stating a 'reasonable' birthday bonus value does/doesn't exists just other variables eg the random daily points, comps viewed etc start having more effect on outcome. The only guaranteed solution is 0 birthday bonus points for no bias. 50 points it' becomes obvious.

Possible an appropriate range of birthday bonus might result in an insignificant bias according the a random daily points system. It's a complex mathematical problem. You need to consider all the variables like the expected number of ; participants, variation of birthdays in a month, random daily points value.

Statistically untrue, eventually even with a baised system it may happen, just happens less often and may take longer then average to be in the top 3! Currently the chances are once a year plus the slight chance the number of birthdays in a month are less then about 4. 0 birthdays in a month everyone has equal chance, which is the same as having a 0 birthday bonus for this month that month.

Many variables can change for example, decrease the points for the daily guessing game. That way members arent penalised as much and loose interest. Reducing the daily points to 10 in my opinion is a good idea as makes viewing competitions more important.

To demonstrate how biased will put up a sample of expected outcomes grouped according to the number of birthdays in a month.

@admin maybe at the end of this year..? I don't want to be (further) disadvantaged. My birthday's NYE (31/12)! Lol Remove it from 1 January 2017..?

@toxicgherkin @Lemond Have either of you gentleman entered featured comps and are they safe to enter?

@toxicgherkin lol Jan 1st bound to be heaps of 'birthdays' that day for the members that don't want to enter in thier real date. I find that day an easy one to remember if I don't trust a site for personal details.

Just having a look through 1 example of a birthday distribution and the effect of birthday bonus. Possible can have a 'reasonable' bonus. Technically should to redo that simulation numerous times before making any assumptions. The bias is fairly clear the higher the birthday bonus is. 50 points isn't much difference from guaranteing with say 300 points with 10 daily random points. The bias is uneven birthday distributions. More importantly the 50 point birthday bonus is weighted too heavily towards the current leader board outcome. That ballon needs some serious deflating to reduce its bais to reasonable amount.

I haven't automated running numerous simulations for different birthday distributions and birthday bonus. I've pushed using spreadsheet to its outer limits. It's an impressive prototype for a database implementation. PC Starts cracking a wobbly when attempting more then 100 years of random data at once. Then you need to automate that to generate for various random birthday distributions to fiind recommend birthday bonus.

It's most likely only valid until the excepted maximum number of birthdays in a month, for rolling 12 month period increases. So occassional would need minor adjustment unless a sudden increase of active members.

Looks roughly 40-70 regular members do all daily requirements so maximum be around 11 for a month. Surprised a few birthdays haven't shown up at the top of the leader board given the big advantage. Only ones I saw early in the month had missed at least a day.

Ok now for @ admin for an official reply regarding the birthdays bonus will do a new post. If want to 'ask' the forum an important questition best not to hide in down in existing post.
Done one report of expected payout. Will post that up today. Hope you understand why it's currently biased.

@Anna2077 Hi, there. Yes, I've entered featured comps. I made the mistake of using my personal email account. I wished someone had advised me to setup a throwaway email address just for comping, in general. My personal email is bombarded with hundreds of junk email offers daily . It's easy going on a deleting binge to inadvertently delete something of importance. Yes, I've most certainly done that before, and on more than one occasion. Some of the more spam-type emails I'm hesitant to click on the 'Unsubscribre' link, just in case it redirects to something nasty, or attempts to infiltrate the laptop's securities. Easier for me just to delete. I do report a lot of spam to ACMA but you just quickly receive a 'reply acknowledged' from them. Great! Another email to delete! Grr! Lol. When it's all said and done, no, I don't recall having won in a featured comp on this website or on random comp entries.

@Anna2077 I would start with ones where you initiate contact verses the ones emailed claiming to be competitions mostly likely SPAM, unless you have a subscription for a site. I wish I kept a list of sites I've subscribe to for competitions. As I haven't touched my emails for 2 months, I bound to indentify those spam to permantly bin. I just hope haven't missed any wins from before mid June.

I think the ones that start asking long survey's for things you're not interested would be high on the list to be wary of receiving unwanted emails and calls. Best safety if afraid of financial/identity fraud is to limit what correct personal details given out, used separate emails for personal, business and comping.

Layer bank accounts or even use multiple banks so harder to hit all at once. What I've done. Never know when one bank systems wants to chuck a wobbly. Of course different pins for each account and bank. God just what we all love another PIN to remember. The joys of using a bank account you don't use regularily. Like the guessing game, oh so close but no cigar.

Limit amounts kept in accounts if its a debit/credit card and only leave just enough as required. Saves being stung by even "reputable" companies (cough TPG or any ISP) trying to double dip for a bill.

Don't use the phone to visit unknow sites. Maybe use different device for comping and business. Good idea there. For financial things I might reboot into a clean OS which I use as a backup. Hmmm maybe create another OS boot, one, for personal/comping, financial and also backup. At least the IPAD Pro I won helps when the PC overheats and those safeguards are worthless. lol

@toxicgherkin Hear Hear!!! Same I wished I used another email for comping.

Any volunteers to clean up my personal/junk email account. To be nice you can do it for the last two months hehe. Think its grown from 9,000 what I had previously in mid June to around 15,000 give or take a few.

@Anna2077 I think it would be irresponsible to list any sites as safe. Most companies wouldn't have the resources to guarantee any personal details are safe. Even the Australian government has had issues keeping data safe. USA milarity secrets for planes going to China, Wikileaks. The list goes on and on with security breaches it's an endless battle. From Samsung, Sony been lots of cases of stolen user and password lists, compromised credit cards etc etc.

Even for the Census data, if identifier keys are separate from the personal data and then put one down a mine shaft with no connection, you might think it was secure. Well ask IRAN and their issue with nuclear centrifuges. They were on an independant network system, cut off from everything else. The weakest link was not the technology safeguards but people. I.e. employees home machines were targetted and some stupid fool used a USB key brought in from home on the work machines, for that case. Well the malware accidently brought in was designed to destroy important hardware without raising the alarms. Kinda of like the philsing scams of old, with dailup. One neighbour ended up with an $1,000 phone bill. The malware would silence the modem speakers and then proceeded to dailup an overseas number.

EG. for call centres if tricked or careless could accidently give out personal information etc. Cases of Indian call centre empolyess selling off personal data etc.

Best safeguard, assume all are potientially unsafe. What can you do? Hmm .. least make it hard as possible, and put up enough hurdles, so they move on and go after the easier targets first. The exception to that if you got something of "value" enough to attract hardcore hackers. Like a few billion sitting around, like the Federal Bank of Bangladesh had earlier in the year.

@toxicgherkin thank you! Wasn't sure if they are safe to enter.

I used a social media email address for comping and now I get at least 100 spams a day. Not unsubscribing & just deleting. As unsubscribing will bring 100000 more. Learned the hard way in the days of hotmail :-)

Went on a deleting binge and deleted an invite.. apparently had to RSVP. Sadly missed the event.

(did my select the page and delete) lol .. took only few minutes.

@Lemond thank you! Most ask too many questions I have no desire to answer.

1. Birthday
2. Location
3. Where do I bank

and few others cannot think of right now.