DLDT: NHL Draft Lottery Edition

[Derrico]

http://sports.espn.g...tory?id=2114426

 

 

 

Any CP sightings in NYC recently?

 

Hahahahah thats great.

[Hoss]

Six times in the 19 NHL lottery's has the team that finished with the worst record been awarded the 1st pick. Being the worst obviously is no guarantee.

 

6/19 is 31%. Even better than the 25% already given. Imagine that.

[Samson's Flow]

6/19 is 31%. Even better than the 25% already given. Imagine that.

That's statistically bad news for our #1 pick hopes. Theoretically there is a regression to the stated odds of 25% coming :bag:

[Derrico]

6/19 is 31%. Even better than the 25% already given. Imagine that.

 

Ya we're ###### :P

Edited April 15, 2014 by Derrico

[carpandean]

Six times in the 19 NHL lottery's has the team that finished with the worst record been awarded the 1st pick. Being the worst obviously is no guarantee.

 

I thought that the whole "they only have a 25% chance" thing made it pretty obvious, but if you prefer statistical evidence to underlying probabilities ...

 

:P

Edited April 15, 2014 by carpandean

[Trettioåtta]

I thought that the whole "they only have a 25% chance" thing made it pretty obvious, but if you prefer statistical evidence to underlying probabilities ...

 

:P

 

What's the opposite of gambler's fallacy? :P

 

Buffalo's veracity?

[carpandean]

That's statistically bad news for our #1 pick hopes. Theoretically there is a regression to the stated odds of 25% coming :bag:

 

Speaking of gambler's fallacy.

 

Poor misunderstood regression to the mean.

[LastPommerFan]

Speaking of gambler's fallacy.

 

Poor misunderstood regression to the mean.

 

A set deviating from the mean will always regress to the mean with the next marginal item added to the set.

 

Correct?

[dEnnis the Menace]

A set deviating from the mean will always regress to the mean with the next marginal item added to the set.

 

Correct?

 

let's say no for this year, and yes next when we're like 4th or 5th worst :rolleyes: :flirt:

[carpandean]

A set deviating from the mean will always regress to the mean with the next marginal item added to the set.

 

Correct?

 

The further a set of draws has been from the mean, the greater the chance that the next draw (or set of future draws) will have a value (or mean for a set of draws) that will result in the overall sample mean being pulled in the direction of the true underlying mean. This is simply due to the fact that there are more possible outcomes between the current sample mean and the actual population mean.

 

Consider rolling one die. You do it five times and get a sample mean of 4.2 (say, 2,5,6,5,3.) The real mean is 3.5, so 4 of 6 outcomes will pull you toward that population mean (i.e., the sample mean for the six points will be lower than 4.2.) If, instead, your first 5 had a sample mean of 5.2 (say, 3,6,5,6,6.) Then, 5 of 6 outcomes would pull you toward the population mean. Your chance of "regressing to the mean" is 83% instead of 66%, but the probability of rolling a 1 or a 2 or ... has not changed; still 1/6.

 

The draws are independent, so the value of one (or more) cannot affect the probabilities for another.

Edited April 15, 2014 by carpandean

[shrader]

Do they actually show anything besides which team won (since nothing else matters)?

 

They're not going to show anything because what they actually do would make almost no sense to anyone. It isn't as simple as one ball coming out of a machine with a team logo on it. In order to set that up, they'd have to have 1000 balls in the machine to produce the odds listed above. They're not wasting their time on that kind of setup.

 

Maybe my google skills suck, but I can't seem to find anything online that states exactly what the league does. Whether it is computer simulation or drawing a random number from the ping pong ball setup. Either way, the process would bore or confuse just about anyone, so there is nothing worth airing on TV. This is always going to bring up the conspiracy theories, but it's impossible to avoid those.

 

the probability of rolling a 1 or a 2 or ... has not changed; still 1/6.

 

2/6 ;)

[carpandean]

2/6 ;)

 

Yeah, yeah, you know what I meant.

 

"The individual probability of rolling any single value (e.g., 1) is still 1/6." Better?

[LastPommerFan]

The further a set of draws has been from the mean, the greater the chance that the next draw (or set of future draws) will have a value (or mean for a set of draws) that will result in the overall sample mean being pulled in the direction of the true underlying mean. This is simply due to the fact that there are more possible outcomes between the current sample mean and the actual population mean.

 

Consider rolling one die. You do it five times and get a sample mean of 4.2 (say, 2,5,6,5,3.) The real mean is 3.5, so 4 of 6 outcomes will pull you toward that population mean (i.e., the sample mean for the six points will be lower than 4.2.) If, instead, your first 5 had a sample mean of 5.2 (say, 3,6,5,6,6.) Then, 5 of 6 outcomes would pull you toward the population mean. Your chance of "regressing to the mean" is 83% instead of 66%, but the probability of rolling a 1 or a 2 or ... has not changed; still 1/6.

 

The draws are independent, so the value of one (or more) cannot affect the probabilities for another.

 

You could not be further from reality. And I thought you were a chart/statistics guy. You completely forgot to factor in the Bettman Corollary to the Costas Ocular Condition.

[carpandean]

You could not be further from reality. And I thought you were a chart/statistics guy. You completely forgot to factor in the Bettman Corollary to the Costas Ocular Condition.

 

Dammit, I always forget the BC to the COC. (There's a dirty joke in there somewhere.)

[26CornerBlitz]

SIX CANADIAN TEAMS IN THE MIX FOR NHL DRAFT LOTTERY ON TSN

During the 30-minute TSN original production, Hockey Insider Bob McKenzie will unveil his consensus Top 10 prospects and TSN's Director of Scouting Craig Button projects how the first 13 picks will unfold.

[Jsixspd]

1 in 4 odds for a Buffalo team is almost a certain loss, given Buffalo teams' star crossed histories.

 

Is it true the worst we can pick is 2nd then if we lose 1st? I thought I read somewhere the most anyone can drop is one place.

[Derrico]

1 in 4 odds for a Buffalo team is almost a certain loss, given Buffalo teams' star crossed histories.

 

Is it true the worst we can pick is 2nd then if we lose 1st? I thought I read somewhere the most anyone can drop is one place.

 

That is correct.

[carpandean]

1 in 4 odds for a Buffalo team is almost a certain loss, given Buffalo teams' star crossed histories.

 

Is it true the worst we can pick is 2nd then if we lose 1st? I thought I read somewhere the most anyone can drop is one place.

 

Yes, the winner moves up to first and every team in between their old spot and the top moves back 1 spot. So, either we win and pick 1st or someone else wins and we pick 2nd.

[darksabre]

Yes, the winner moves up to first and every team in between their old spot and the top moves back 1 spot. So, either we win and pick 1st or someone else wins and we pick 2nd.

 

You know we're picking 2nd. It's gonna happen.

[dudacek]

 

 

You know we're picking 2nd. It's gonna happen.

 

That's a given. The only question is if it will be the Leafs that jump ahead of us, the Canucks, the Oilers or the Canes.

[Jsixspd]

You know we're picking 2nd. It's gonna happen.

 

Even more likely given the stats that Kishoph posted in the "Leino Must Go" topic. He wrote that in the 19 years of the lottery, the number 1 pick has won 6 out of 19 times, or 31.6%. That's quite a bit higher than the 25% that should result

 

So our probability of losing is higher than 1 out of 4, because the distribution will trend back to 25%.

[Eleven]

Even more likely given the stats that Kishoph posted in the "Leino Must Go" topic. He wrote that in the 19 years of the lottery, the number 1 pick has won 6 out of 19 times, or 31.6%. That's quite a bit higher than the 25% that should result

 

So our probability of losing is higher than 1 out of 4, because the distribution will trend back to 25%.

 

The probability of losing is exactly 3 out of 4. Considerably higher than 1 out of 4. The stat geek has the explanation upthread about how the concept of regression to the mean does not affect a single event.

[TrueBlueGED]

You know we're picking 2nd. It's gonna happen.

 

Fine by me. Gotta save up our luck for next year because let's face it, no way we'd win two in a row! :P

 

[Jsixspd]

The probability of losing is exactly 3 out of 4. Considerably higher than 1 out of 4. The stat geek has the explanation upthread about how the concept of regression to the mean does not affect a single event.

 

Thanks - I miswrote that - I should have written - our chance of success is lower than 25%.

[Eleven]

Thanks - I miswrote that - I should have written - our chance of success is lower than 25%.

 

But it isn't. It is exactly 25%.