There are only three magic numbers that matter. One for the Habs to be in the playoffs, the one for the Leafs to be out, and the one for the Bruins to be out.

I know it's silly to start this so early but hey, at least it's something different right? Let's try and keep all magic number discussion in here.

Umm, yeah. Since it's "statistically" possible for every other team in the conference to end the season 32-0-0, our magic number is 54. However, remember that when teams in the same conference meet, only one of them can get the 2 points, and the other has to settle for 1 or 0. "Realistically" or "technically" our magic number should be quite less than 54 to clinch a playoff spot looking at the remaining games (briefly).

Since it is sooo early, it may be better (for speculation) to look at the teams in our conference, check out who they play over the final part of the season, check their record against same conference teams in that list, project their current record vs those teams over the remainder of the season, assume wins against all intra-conference opponents, and see how the points would end up that way. either way, it is too early to mean anything, but the thread had to be started some time!

Umm, yeah. Since it's "statistically" possible for every other team in the conference to end the season 32-0-0, our magic number is 54. However, remember that when teams in the same conference meet, only one of them can get the 2 points, and the other has to settle for 1 or 0. "Realistically" or "technically" our magic number should be quite less than 54 to clinch a playoff spot looking at the remaining games (briefly).

I know it's a bit silly but that's the statistical magic number at this point. Realistically I see it more around 35 ish.

Quote:

Since it is sooo early, it may be better (for speculation) to look at the teams in our conference, check out who they play over the final part of the season, check their record against same conference teams in that list, project their current record vs those teams over the remainder of the season, assume wins against all intra-conference opponents, and see how the points would end up that way. either way, it is too early to mean anything, but the thread had to be started some time!

That's a ton of work...I invite someone to do it though!

I know it's a bit silly but that's the statistical magic number at this point. Realistically I see it more around 35 ish.

That's a ton of work...I invite someone to do it though!

The problem with "magic numbers" is this: if the point gap between first and last place is less than the technical number of points possible through the end of the season, it tells you nothing, and really isn't useful for speculation in and of itself.
that's why the remaining match ups, and the REAL number of points available given every permutation and combination of the remaining matches (taking possible points away from losers with every loss), are so important.

Once there is a gap of, say, 10 points or more between 1st and 9th, with only 5 games to go, THEN magic numbers start to mean something. Again, the thread had to be started sometime. And if the NHL page was better at displaying game results against other teams without being separated into home and away teams, I might give my hand at figuring it out for the current top 8 teams.

Montreal Canadiens Playoff Race
Canadiens beat Devils 4-3, odds of making the playoffs up 2.6% to 91.9%

Not updated yet.

Problem with that is the method in which those numbers are calculated. It's a home-made formula involving a 1000 game simulation "fudge" depending on who he considers to be "top" or "bottom" teams (he forces them to either win or liose 1000 straight games to curb the probabilities down to "reality").

He's probably pretty close, but by no means accurate despite all of his hard work.

Problem with that is the method in which those numbers are calculated. It's a home-made formula involving a 1000 game simulation "fudge" depending on who he considers to be "top" or "bottom" teams (he forces them to either win or liose 1000 straight games to curb the probabilities down to "reality").

He's probably pretty close, but by no means accurate despite all of his hard work.

It's not a "home-made formula," it appears to be a Monte-Carlo simulation using 9.998 million iterations that gets it about as close as possible. He forces 1000 count to winning and losing all games, because otherwise the count would be something like 1 or 0 and wouldn't show up, and reasonably people want to know how a team would do if they won out.
Every team is forced to have 1000 count (i.e. a 0.01% chance.. 1 in 10000) of winning all games, and losing all games. This doesn't curb the probabilities by any significant margin. For instance, if he didn't do that, the probability of Montreal finishing first overall would be 14.246%, a difference of 0.01%.

The real flaw is that he considers every game a coin flip, but save a lot of work and another inherent flaw (that a higher-ranked team has a better chance to win), it's the only reasonable compromise.

I mean, if you know exactly how it's cooked up then by all means tell me, but each time I look at it, it's the most reasonable solution that I could come up with.

The magic number is far more usefull in sports where there are no tie and no point system such as baseball and Basketball. As far as football goes, who really needs to calculate the magic numbers on a 16 game season, right?

I tend to look at several things to see if a team (in hockey) can be a lock for the playoffs or not.

Montreal is having a real good sequence over the last 21 games, keeping a record of 13-4-4 for 30 points, 74 goals scored and 49 allowed. They are +11 on ES in this sequence. They are 7-2-1 in their last 10 games and most importantly, they have reverted bad habits against several teams this season, such as Florida, Toronto (with the help of Price) and especially, New Jersey. Only the New York Rangers are left on the list and they ain't our worst matchup of those I've named.

Also, I then look at the difference in scoring +/-, which is always indicative of a good team or a lucky one (or opportunistic) or just a plain bad one. (EC) Ott +32, Phi +23, Mtl +22, Pitts +12, NJ +8, Buff +5 (I think they may surprise us till the end of the season), and the rest are all minuses. Note that among ALL NHL teams, Montreal is 4th in the entire league for scoring +/-, behind Det +64, Ott +32 and Phi +23.

Then I also look at the points/games differential (and not just the points difference as most lazy jaune-listes do...). In the EC, (1st in conf with div) Ott is +17, (4th) Mtl and (2nd) Phi are tied at +12, Pitt (5th) +10, NJ (6th) +9, Bos (7th) +7, NYI (8th) +3, NYR (9th) +2, Car (3rd) +1, Buf (11th) +1, Was (10th) +/-0, Atl (12th) -2, Flo (13th) -2, Tor (14th) -4, TB (15th) -6.

Now just as a gauge, to attain 100 points in a season, a team must end at least at +18 (82 games/100 points = +18 differential, as 82 points is the even mark). This is the difference right now between Montreal and Tampa Bay. Montreal as a +10 difference with the 9th place Rangers. +9 difference with the Isles in 8th place.

All this is just about numbers. Now include all the facts related to the team, and we can actually say that the chances of repeating last year's month of Febuary is pretty slim. If the Habs can win 2 of the 3 following games (Wash and the Super Bowl week-end annual double afternoon games, this time against the Isles and Rangers), I think they can almost be considered a lock, considering that Montreal is right now on par for 100 points as there are 6 months to the season, and each month requires the team to end that month with a +3 points differential (for the month itself) to stay on par for 100 points. Montreal will end its fourth month at either +11, +12 or +13, all depending on the outcome of the next Caps game, which is very close to the mark.

Now, if an ENTIRE season requires that you have a +18 points differential to attain 100 points, A +10 or +9 difference is pretty steep to overcome. It's not impossible, but very hard as there are only two months of activity left (out of 6.. not including the 2-3 games in April).

Funny it is that the two teams we meet on Super Bowl week-end are the two teams in either the last spot for the playoffs or before access to the playoffs. If we beat both the Isles and Rangers, Montreal will be a pretty sure bet for the post season dance.

The real flaw is that he considers every game a coin flip, but save a lot of work and another inherent flaw (that a higher-ranked team has a better chance to win), it's the only reasonable compromise.

I mean, if you know exactly how it's cooked up then by all means tell me, but each time I look at it, it's the most reasonable solution that I could come up with.

Exactly. The fact that every game is a coin flip is inherently bogus. Despite how the final standings work out, teams always have their boggie teams (weaker teams that seem to beat you overly often), so remaining matchups and historical performance seem to be crucial in determining a "likely-hood" of making the playoffs or how many points will likely (due to the number of games left) be needed to secure a spot. And forget about the fact that team's rosters can still change drastically until the trade deadline every year.

History tells us that the playoff series results can't necessarily be predicted by seeding, and yet aren't a simple 50/50 ("coin flip") situation. The formula's results are interesting, but useless in that they only represent something that is blatantly obvious: taking your current ranking into consideration, the more you win, the more likely you'll make the playoffs. The ACTUAL numbers mean nothing.

Umm, yeah. Since it's "statistically" possible for every other team in the conference to end the season 32-0-0, our magic number is 54. However, remember that when teams in the same conference meet, only one of them can get the 2 points, and the other has to settle for 1 or 0. "Realistically" or "technically" our magic number should be quite less than 54 to clinch a playoff spot looking at the remaining games (briefly).

Since it is sooo early, it may be better (for speculation) to look at the teams in our conference, check out who they play over the final part of the season, check their record against same conference teams in that list, project their current record vs those teams over the remainder of the season, assume wins against all intra-conference opponents, and see how the points would end up that way. either way, it is too early to mean anything, but the thread had to be started some time!

At this site, the author calculates that we have a 95% chance of making the playoffs. We also have a 22% chance of catching Ottawa and winning the division.

So we actually have a four times BETTER chance of passing Ottawa than missing the playoffs.

Exactly. The fact that every game is a coin flip is inherently bogus. Despite how the final standings work out, teams always have their boggie teams (weaker teams that seem to beat you overly often), so remaining matchups and historical performance seem to be crucial in determining a "likely-hood" of making the playoffs or how many points will likely (due to the number of games left) be needed to secure a spot. And forget about the fact that team's rosters can still change drastically until the trade deadline every year.

History tells us that the playoff series results can't necessarily be predicted by seeding, and yet aren't a simple 50/50 ("coin flip") situation. The formula's results are interesting, but useless in that they only represent something that is blatantly obvious: taking your current ranking into consideration, the more you win, the more likely you'll make the playoffs. The ACTUAL numbers mean nothing.

Well I'd say that a coin flip is more accurate than predictions using the standings or any method short of a variable combination of several trends and statistics compared against previous outcomes. For instance, a weighted average of last 10 GP divided by strength of competition ratio, last 5 or 10 GP against that opponent and last 5 or 10 in that arena, standings, comparison of GF-GA ratio among common opponents, evaluation of roster metrics etc. You can include as many figures as you want and a coin flip will probably turn out nearly as accurate, when you account for the amount of work you went through to be able to predict games to something like 5 or 10% more likelihood.

The numbers are as meaningless as you consider them meaningless, but it's inherently true that at 62 points Montreal has less distance to travel to the postseason than someone with 60, or 50, and the percentages reflect that fairly accurately.

If you could simulate who'd win games, there'd be no reason to play them. Statistics isn't supposed to be definite because ANY evaluation of the future has just as many flaws as this does. The author did as well as anyone can.

Well I'd say that a coin flip is more accurate than predictions using the standings or any method short of a combination of several trends and statistics (a weighted average of last 10 GP divided by strength of competition ratio, last 5 or 10 GP against that opponent and last 5 or 10 in that arena, standings, comparison of GF-GA ratio among common opponents, evaluation of roster metrics etc).

The numbers are as meaningless as you consider them meaningless, but it's inherently true that at 62 points Montreal has less distance to travel to the postseason than someone with 60, or 50, and the percentages reflect that fairly accurately.

If you could simulate who'd win games, there'd be no reason to play them. Statistics isn't supposed to be definite because ANY evaluation of the future has just as many flaws as this does. The author did as well as anyone can.

The best way to do the flip coin thingy is to have a predetermined number of coins per team depending on their strenght for each flip and maybe change the number of coins depending on the history of each machtup. Detroit gets 10 coins, Ottawa 9 coins, Phi, Mtl, NJ all get 8 coins.. ect. Then at every matchup, as an example, you flip Detroit's ten coins versus San Jose's 8 coins, and see which one as the most -Win- side up. Repeat the process a 1000 times for each game, fluctuate team coin numbers accroding to sequence, strenghts and injuries. Make it 20 coins max per team if you need to, instead of 10.