You throw around that word, regression, a lot... but I don't think there's a team in the league that's more primed for a massive regression than the Sens.


A lot of people compare the Leafs to Colorado and Calgary, saying they regressed and the Leafs will too... but Ottawa's underlying stats are a lot closer to those teams than the Leafs.
- Feeling Glucky?

Keep it real simple? Best performing possession teams missed the playoffs. High perking possession got knocked out in the first round. The conversation on regression is entirely bizarre.

There is no correlation between possession and standings from last year. So, what does regression actually mean.

If you have great numbers like LA you can miss the playoffs. Is this not a joke?

Keep it real simple? Best performing possession teams missed the playoffs. High perking possession got knocked out in the first round. The conversation on regression is entirely bizarre.

There is no correlation between possession and standings from last year. So, what does regression actually mean.

If you have great numbers like LA you can miss the playoffs. Is this not a joke?
- spatso

I guess we can just wait and see.

I've got facts, you've got wishful thinking.

now here is a question for you. Craig Anderson had a career year (for playing 35+ games). if he doesn't replicate that, and say, falls back to his career average of .916, how many games do you think that will cost the turds? he's got a long history of being a streaky goalie. not only is it possible. its likely, at the age of 37 or whatever.

over 40 games, that's 10 to 15 goals. I'd say that's FAR more likely to lead to a regression than a supposed sophomore slump which barely exists, if at all. especially for a team that had such trouble with goal differential. and pretty fortunate skills competition results. oh and just lost the defensive side of their first pairing.
- prock

Great point, but although Anderson is getting older (and that could start to catch up with him,) his sv% during his time with OTT is actually .924, so you may not want to rest all of your "I hate Ottawa" hopes on him regressing back to his terrible Chicago numbers from 15 years ago... Also,this all has zero to do with my original comment that TOR had some bad luck last year, but also lots of good.

Wow,a lot of useless arguing here on this article. In reality we should all be happy that Canadian teams are on the upswing and hopefully heading into the right direction.

Both still need a lot and wont be contending for a cup for a while. No point in arguing when both the Sens and Leafs would be able to compete with the Caps, Hawks, etc.

The Leafs will be dog crap next year and will miss the playoffs, and for the 10th time in 12 years will finish behind the Sens. I don't care if you don't believe me, it will happen.

A significant portion of their offense relies on 3-4 good rookies who will all face potential sophomore slumps. While it is not necessarily a given, there is a small regression in points totals for players in their second year compared to their rookie seasons, or nil to very marginal improvement in most cases, even for elite players.

Despite what the cherry picked comparable analysis in the article below tells you, for most of the players listed there was either no change at all, or significant declines in their output, with some extreme outliers of improvement like Kovalchuk and Kopitar bringing up the averages significantly (it's not the cherry picked variables, but the numbers behind the numbers, that matter in analytics like this).

The Leafs should expect similar output from Matthews, Marner and Nylander (60-65 points each) as what they got last season, if it all falls into place and the team stays injury free, but to expect Matthews and Marner to turn into the second comings of Crosby and Kane in their sophomore seasons is just plain homerism, nothing more.

https://www.pensionplanpu...erience-a-sophomore-slump

They'll be a bubble team again, finish anywhere between 9th and 12th in the conference, but the regression in the offensive output for their rookie scorers (plus given how they've done nothing to upgrade their weakest position with their defense) will be their undoing. They'll still be a fun exciting team to watch, but don't expect division title challenges or even a second round appearance until they can figure out how to play team defense, which they haven't yet.
- ahjnkn

These sound like the incoherent, angry ramblings of a guy who's heard one too many "Go Leafs Go" chants in his own barn....

You'd better hope the Sens get even luckier this year than they did last year. Because after this season, you're dealing with Karlsson occupying 20% of your cap space. That, or you'll lose Karlsson all together. For a team that doesn't spend anywhere near the cap, you'd better hope it's the latter...

Great point, but although Anderson is getting older (and that could start to catch up with him,) his sv% during his time with OTT is actually .924, so you may not want to rest all of your "I hate Ottawa" hopes on him regressing back to his terrible Chicago numbers from 15 years ago... Also,this all has zero to do with my original comment that TOR had some bad luck last year, but also lots of good.
- wilkobecks

And yet, in seasons he's played more than half a season (just 3), even with the turds, the best sv% he's had is .916. (.911, .913, and .916).

So, if you play him as your backup, he'll do just fine.

Besides, it's not .924. It's .920. So, if YOU'RE counting on him to play to your fictitious average in a turds uniform, you may not want to rest it all on that either.

The Leafs will be dog crap next year and will miss the playoffs, and for the 10th time in 12 years will finish behind the Sens. I don't care if you don't believe me, it will happen.
- ahjnkn

I disagree with the idea of the Leafs being dog crap next year. Under Babcock the team has gotten progressively better over the course of the season. They were much better at the end of the year than they were at the beginning. So the progression should carry over.

I don't believe they'll win the division; but I do think they'll make the playoffs even if they sneak in by a single point again...

A significant portion of their offense relies on 3-4 good rookies who will all face potential sophomore slumps. While it is not necessarily a given, there is a small regression in points totals for players in their second year compared to their rookie seasons, or nil to very marginal improvement in most cases, even for elite players.

Despite what the cherry picked comparable analysis in the article below tells you, for most of the players listed there was either no change at all, or significant declines in their output, with some extreme outliers of improvement like Kovalchuk and Kopitar bringing up the averages significantly (it's not the cherry picked variables, but the numbers behind the numbers, that matter in analytics like this).

The Leafs should expect similar output from Matthews, Marner and Nylander (60-65 points each) as what they got last season, if it all falls into place and the team stays injury free, but to expect Matthews and Marner to turn into the second comings of Crosby and Kane in their sophomore seasons is just plain homerism, nothing more.

https://www.pensionplanpu...erience-a-sophomore-slump

- ahjnkn

The team still hasn't really gotten worse though. Matthews probably won't score 40 next year, sure. And Nylander and Marner might only be 60 point players again at best. However they still haven't gotten significantly worse. If anything they got better up front with the addition of Marleau, and the backend has improved simply by Polak not coming back (OH PLEASE DON'T BRING POLAK BACK...).

They'll be a bubble team again, finish anywhere between 9th and 12th in the conference, but the regression in the offensive output for their rookie scorers (plus given how they've done nothing to upgrade their weakest position with their defense) will be their undoing. They'll still be a fun exciting team to watch, but don't expect division title challenges or even a second round appearance until they can figure out how to play team defense, which they haven't yet.
- ahjnkn

I agree with the fact that they may just be a bubble team again next year. But with the experience of the late season push to the playoffs I think they have a better idea of how they have to play to get there first. I'll say this; if the Leafs finished 9th in the East I wouldn't be shocked, but I do think they can make the playoffs.

Their team defence (systems, positioning etc) will be improved. Their backend hasn't "improved" how Leafs fans would like, but the deletion of Polak should lead to taking less penalties in the very least.

The Leafs ran Andersen ragged because they didn't trust their backup until MacElhinney showed up. But with Mac in place now Andersen won't play back to back and they'll be at least slightly more willing to give Andersen a night off with a more trusted backup.

I think the regression in Brown will be made up with Marleau, and Hyman basically has almost no room to regress anyways. They can afford a few injuries, but as long as they don't get too crazy they can probably survive them (Especially on the wings where their depth is much better than at center). Moore replacing Ben Smith as the 4C for a whole season will make up some room as well.

I don't think they'll be as good as some Leafs fans think, but I don't think they'll be as bad as most Sens fans hope.

yeah, I thought it was an odd reference too. An article that basically says there is little to no regression from year 1 to 2, and he's using it as a basis to say there will be regression.

I'd question whether the very little regression that is shown there, is actually statistically relevant. I'd say there is a good chance that any variance there probably isn't enough to qualify as statistically relevant, and likely falls into the category of standard variance.

There are formulas to measure that, I just don't know what they are. Someone should ask lumlums to calculate it.
- prock

*Sigh* You didn't actually read the article and charts fully, did you?

Some of what you are saying is true. What you call standard variance is standard deviation (the Greek symbol for sigma, σ), and you can calculate it by finding the square root of the variance from the mean or average of the group you are comparing. I'll skip the calculation steps for you (good link here if you want to do it yourself: http://www.mathsisfun.com...a/standard-deviation.html) but the results are that the standard deviation for the palyers compared in the article between year 1 and year 2 is (rounded up from 0.096) 0.1 points per game. Any result above that total (positive or negative) means that there is statistical significance to the result. Everything else in the sample is just white noise like what you describe.

The author picked a single comparable to each of the Leafs rookies (Matthews=Kovalchuk, Marner = Kane, Nylander = Brad Richards, etc.) and shows that those individual players improved in their second year, and since they have similar attributes the Leafs rookies output will likely be the same in their second seasons. What was overlooked there is that the cherry picked comparables were always among the best performing players in each group assessed. If you read the background charts for each group compared, there are a larger number of players who reported declines or no significant change in their output in the next season. In statistics terms that is taking the most extreme positive outlier in a sample and trying to frame it as the normal average for the group. It's BS and does nothing but confirm the point I was making, that most hockey analysis in the media and blogosphere (people like Yost being exceptions) is done to suit a narrative, not to see what the most probable outcome is likely to be.

Here's how the players in each grouping actually performed when broken down by actual results for each player. My method is any player with more than a 0.1 Pts/GP ratio is a statistically significant improvement or regression, while those within the 0.1 range are seen as no change year over year:

Matthews Comparables:
Stat. Significant Improvement: Nash, Kovalchuk
No Change: Kane, Toews, Kopitar, Gaborik
Stat. Significant Regression: Lecavalier
% of all comparable Players assessed with statistically significant improvements in their second season: 28.6%

Nylander Comparables:
Stat. Significant Improvement: Thornton, Draisaitl
No Change: Monahan, Richards, F. Forsberg, Nugent-Hopkins, Mackinnon, E. Kane (barely)
Stat. Significant Regression: Lecavalier
% of all comparable Players with statistically significant improvements in their second season: 22.2%

Marner Comparables
Stat. Significant Improvement: Kovalchuk, Nash, Eichel, Duchesne
No Change: P. Kane, Toews, Kopitar, Gaborik
Stat. Significant Regression: Lecavalier, Skinner
% of all comparable Players with statistically significant improvements in their second season: 44.4%

Connor Brown Coomparables:
Stat. Significant Improvement: Arnason, Plekanec, Brouwer
No Change: Cole
Stat. Significant Regression: Donski, Chiasson, Carcillo, Janmark, Purcell
% of all comparable Players with statistically significant improvements in their second season: 33.3%

So, based on all the Leaf rookie comparable players, the large majority of comparables saw either no significant improvement or a regression in their second season compared to their first. To me that says that it is very very optimistic (unrealistic, actually) to expect that in each case the Leafs rookies will see improved results in their play over last year.

You are right, there is also very little statistical significance in terms of output changes for a lot of the players but all that means is that no significant improvement should be expected from this Leafs group. It's also important to point out that in each case, the players used in the sample comparison are also cherry picked by the author to suit their conclusions, rather than being representative examples. This is p-hacking and is generally seen as unacceptable for data analytics work like this.

I didn't saying that a regression was definitely going to happen, only that the Leafs should not be expecting any significant improvements in terms of individual scoring, or any extra points in the standings as a result. If the team is going to finish any higher, it will in all likelihood be on the back of improved defensive play, and they have not done much of anything to address that yet.

I'm sure you will not read any of what I just wrote above and will continue to disagree with me just on principle, but the article does nothing but confirm what I said if you know how to read the numbers and how to figure out when they are being cherry picked to suit a need. It's not that the Leafs should expect a regression, just no improvement from last year.

Anyone else think it would possibly be a great fit for Markov to play with Ottawa?

He could be money on the PP point with Karlsson and if it's a 1 year contract, he could rack up a lot of points for next summer UFA, so he would have incentive to come here as well. It would be the perfect stop gap until Chabot is ready to be on the top pair

I didn't saying that a regression was definitely going to happen, only that the Leafs should not be expecting any significant improvements in terms of individual scoring, or any extra points in the standings as a result. If the team is going to finish any higher, it will in all likelihood be on the back of improved defensive play, and they have not done much of anything to address that yet.

I'm sure you will not read any of what I just wrote above and will continue to disagree with me just on principle, but the article does nothing but confirm what I said if you know how to read the numbers and how to figure out when they are being cherry picked to suit a need. It's not that the Leafs should expect a regression, just no improvement from last year.
- ahjnkn

Oh, I see, so you weren't really saying that a regression WILL happen when you said:

The Leafs will be dog crap next year and will miss the playoffs, and for the 10th time in 12 years will finish behind the Sens. I don't care if you don't believe me, it will happen.

A significant portion of their offense relies on 3-4 good rookies who will all face potential sophomore slumps. While it is not necessarily a given, there is a small regression in points totals for players in their second year compared to their rookie seasons, or nil to very marginal improvement in most cases, even for elite players.

So, you're saying the individuals will not necessarily regress, but the team most certainly will?

You may think you're making sense, but you're not.

Additionally, they explained quite clearly, they took the top 5 Calder winners of each year. It's not really cherry picking, they took ALL rookies from years past, that finished in the same bracket as the Leafs rookies. You know, at the top, like the Leafs' rookies.

I should probably also point out, that all your stuff about players over and under your calculated standard deviation (and by the way, that's not really how it works, you apply the standard deviation test to your sample as a whole, to find out if the entire thing has any relevance, not to each individual player), there were more in the improved row you had, than in the regressed.

All of your ramblings amount to salty tears. It's amazing how the Leafs having one fo the best young teams in the league is turning so many fans of certain other teams in rambling messes.

And yet, in seasons he's played more than half a season (just 3), even with the turds, the best sv% he's had is .916. (.911, .913, and .916).

So, if you play him as your backup, he'll do just fine.

Besides, it's not .924. It's .920. So, if YOU'RE counting on him to play to your fictitious average in a turds uniform, you may not want to rest it all on that either.
- prock

Oh God you're right, sorry I just took a quick look at hockeydb and got .924, but I guess if you're really intent on trolling then you'll put in the legwork and work it our exactly. Those 6-7 "extra" goals that he saved were definitely the difference in OTT's season this year, and I am not counting on anything, just enjoying life and looking forward to next season. (One of the benefits of not being an angry troll)
BTW I hope you don't ind but I am going to start using the word "turds" as a sports team insult, and pretend that I made it up because it really sounds well thought out and clever

Oh God you're right, sorry I just took a quick look at hockeydb and got .924, but I guess if you're really intent on trolling then you'll put in the legwork and work it our exactly. Those 6-7 "extra" goals that he saved were definitely the difference in OTT's season this year, and I am not counting on anything, just enjoying life and looking forward to next season. (One of the benefits of not being an angry troll)
BTW I hope you don't ind but I am going to start using the word "turds" as a sports team insult, and pretend that I made it up because it really sounds well thought out and clever
- wilkobecks

Given that they finished 4 points up on 9th, they absolutely could have been. Toss in the skills competition points, and they absolutely are.

Anyone else think it would possibly be a great fit for Markov to play with Ottawa?

He could be money on the PP point with Karlsson and if it's a 1 year contract, he could rack up a lot of points for next summer UFA, so he would have incentive to come here as well. It would be the perfect stop gap until Chabot is ready to be on the top pair
- AlfieisKing

I had the same thought but word is that he still wants to be paid btw $5-6 Mill, which would be almost as bad as signing Marleau to a 6+

Given that they finished 4 points up on 9th, they absolutely could have been. Toss in the skills competition points, and they absolutely are.
- prock

Well if that is the reasoning then those goals must have come directly out of Bobby Ryan's pocket because he was well below his regular average. problem solved, world back to normal, and we don't have to readjust last year's results to #procksworld.
I am also not sure why you're so obsessed with "could have's" when TOR would have missed the playoffs for the millionth 82 game season in a row, if TBL or NYI had been able to secure just one more loser point in the season

Serious question, what makes you happier, Leaf success or Sens failure?

Well if that is the reasoning then those goals must have come directly out of Bobby Ryan's pocket because he was well below his regular average. problem solved, world back to normal, and we don't have to readjust last year's results to #procksworld.

Serious question, what makes you happier, Leaf success or Sens failure?
- wilkobecks

Leafs success, and the amount it drives some of you and your ilk nuts. Like I say, I love that it's turning so many into a rambling mess.

Did you even read the drivel that started all this? It's hilarious. The twisting of an article dispelling sophomore slumps, trying to use that as a basis for the Leafs regressing? It literally has people becoming nonsensical. It's awesome, I love it.

Leafs success, and the amount it drives some of you and your ilk nuts. Like I say, I love that it's turning so many into a rambling mess.

Did you even read the drivel that started all this? It's hilarious. The twisting of an article dispelling sophomore slumps, trying to use that as a basis for the Leafs regressing? It literally has people becoming nonsensical. It's awesome, I love it.
- prock

totally agree on that, I have no such ilk, crazy ppl are just crazy, regardless of who they cheer for. I actually don't actively dislike any teams except for whoever OTT is playing on any given night, and PITT, well just because.
I think this season is going to be awesome, McDavid is going to absolutely light it up, and Vegas is going trade anyone who threatens to derail their tank plan.

I had the same thought but word is that he still wants to be paid btw $5-6 Mill, which would be almost as bad as signing Marleau to a 6+
- wilkobecks

Or trading for Phaneuf.

Anyone else think it would possibly be a great fit for Markov to play with Ottawa?

He could be money on the PP point with Karlsson and if it's a 1 year contract, he could rack up a lot of points for next summer UFA, so he would have incentive to come here as well. It would be the perfect stop gap until Chabot is ready to be on the top pair
- AlfieisKing

Not a bad idea, I just think he'll want a big contract.

I've suggested this last year but I think picking up a guy like Adam Mcquaid would be a good replacement or Methot for a year or two and then move it around so our top defensive lines would be...

Chabot - Karlsson
Dion - Mcquaid
etc

Could get him at a reasonable price and he would it into Ottawa's systems well.

Or trading for Phaneuf.
- systemtool

Like it or not that trade (so far) has worked out well for both teams. Phaneuf played great for OTT last year and they likely don't get as far as they did if Wideman or whoever is playing in that spot. (Also, OTT would have still had Greening, Michalek and maybe Cowen on the roster, and missed the playoffs altogether)
Moving Phaneuf made TOR's D worse in the short term (and their forwards too), which enabled them to complete the tank for Matthews, so everybody wins.

Oh, I see, so you weren't really saying that a regression WILL happen when you said:



So, you're saying the individuals will not necessarily regress, but the team most certainly will?

You may think you're making sense, but you're not.

Additionally, they explained quite clearly, they took the top 5 Calder winners of each year. It's not really cherry picking, they took ALL rookies from years past, that finished in the same bracket as the Leafs rookies. You know, at the top, like the Leafs' rookies.

I should probably also point out, that all your stuff about players over and under your calculated standard deviation (and by the way, that's not really how it works, you apply the standard deviation test to your sample as a whole, to find out if the entire thing has any relevance, not to each individual player), there were more in the improved row you had, than in the regressed.

All of your ramblings amount to salty tears. It's amazing how the Leafs having one fo the best young teams in the league is turning so many fans of certain other teams in rambling messes.
- prock

I am saying that the Leafs should not expect any improvements in output compared to last year for the former rookie players who are expected to carry the offense, and they have a better chance of missing the playoffs as a result. As good as they were last year, they barely squeaked in, and a similar or worse outcome next season is likely, especially if the defensive results they had last year are similar. The team may not regress but the teams around them have improved a lot more (TB, Boston, Buffalo).

This is something Ottawa also has to worry about next season too like someone earlier pointed out, and I don't deny that. I just believe the Sens are currently built more to address the challenge because they have a more experienced group of players now compared to the Leafs.

It is cherry picking for two reasons. 1) The years of the Calder winners selected are arbitrary. 2) Your point about the Leafs rookies being at the top makes no sense when you consider the Connor Brown comparison in the article as an example. I also didn't know there were 5 Calder trophies awarded each season, that fact was very enlightening, thank you (see? misunderstanding people's statements is fun!!).

And that IS how it works for standard deviation calculations. The standard deviation calculated (from the sample as a whole, which I did) is then used to show the variance of each individual result from the average of the group. The intent of standard deviation calculations is to determine how much each result observed deviates from the average of the group as a whole, not how the average of a group deviates from the average of another group. Nice try, but you're wrong on that point.

You are correct, however, that there were more players who showed improvement then regression, but you conveniently ignore that they still represent a significant minority of all the results observed from each sample group. I was trying (unsuccessfully) to show that the previous results observed would lead most people to conclude reasonably that they should not expect any drastic improvement in performance from the Leafs rookies next year. Is this a regression? Not necessarily, but it will likely not be enough to show any significant improvement next season for any of the rookies, or the team as a whole. It's hockey though, and as the Sens showed this year anything can happen with good timing, a bit of luck, and a couple of really tuned in players leading the way.

All of your ramblings amount to misinterpretations of data and what it's actually telling you only for the purpose of making me look like a dumbass (*insert response about how I am doing that more than well enough myself here*). Don't believe me if you want, it doesn't make the observed outcomes and correct methodology I used any different.

I could not give any less of a care about how awesome the cadre of Leafs rookie talent is (good as it is right now, I admit), I was only saying that to expect the level of improvement out of them next season that some of their fans are (those saying they can challenge for division championships and have Matthews winning the Art Ross) incredibly optimistic and the outcome I predicted of them falling out of the playoff picture is a much more realistic scenario.

I am saying that the Leafs should not expect any improvements in output compared to last year for the former rookie players who are expected to carry the offense, and they have a better chance of missing the playoffs as a result. As good as they were last year, they barely squeaked in, and a similar or worse outcome next season is likely, especially if the defensive results they had last year are similar. The team may not regress but the teams around them have improved a lot more (TB, Boston, Buffalo).

This is something Ottawa also has to worry about next season too like someone earlier pointed out, and I don't deny that. I just believe the Sens are currently built more to address the challenge because they have a more experienced group of players now compared to the Leafs.

It is cherry picking for two reasons. 1) The years of the Calder winners selected are arbitrary. 2) Your point about the Leafs rookies being at the top makes no sense when you consider the Connor Brown comparison in the article as an example. I also didn't know there were 5 Calder trophies awarded each season, that fact was very enlightening, thank you (see? misunderstanding people's statements is fun!!).

And that IS how it works for standard deviation calculations. The standard deviation calculated (from the sample as a whole, which I did) is then used to show the variance of each individual result from the average of the group. The intent of standard deviation calculations is to determine how much each result observed deviates from the average of the group as a whole, not how the average of a group deviates from the average of another group. Nice try, but you're wrong on that point.

You are correct, however, that there were more players who showed improvement then regression, but you conveniently ignore that they still represent a significant minority of all the results observed from each sample group. I was trying (unsuccessfully) to show that the previous results observed would lead most people to conclude reasonably that they should not expect any drastic improvement in performance from the Leafs rookies next year. Is this a regression? Not necessarily, but it will likely not be enough to show any significant improvement next season for any of the rookies, or the team as a whole. It's hockey though, and as the Sens showed this year anything can happen with good timing, a bit of luck, and a couple of really tuned in players leading the way.

All of your ramblings amount to misinterpretations of data and what it's actually telling you only for the purpose of making me look like a dumbass (*insert response about how I am doing that more than well enough myself here*). Don't believe me if you want, it doesn't make the observed outcomes and correct methodology I used any different.

I could not give any less of a care about how awesome the cadre of Leafs rookie talent is (good as it is right now, I admit), I was only saying that to expect the level of improvement out of them next season that some of their fans are (those saying they can challenge for division championships and have Matthews winning the Art Ross) incredibly optimistic and the outcome I predicted of them falling out of the playoff picture is a much more realistic scenario.
- ahjnkn

It wasn't arbitrary. They took ALL of the top 5 finishes in Calder voting for every year over the time period they took. It wasn't arbitrary at all.

I decided to look at the top five rookies in Calder Trophy voting from every year from 2005-2006 until 2015-2016, giving us a total of 55 players. I figured this will provide a large enough sample size, while ensuring the list consists of rookies with similar production to the four Leafs. I then separated the data into specific forwards defensemen graphs.

You really are grasping.

You're also 100% wrong on standard deviation. The standard deviation in an experiment, is applied to the entire thing, to see if the findings are statistically pertinent.

In other words, you carry out an experiment, with a certain sample size, where the control group has an average of 10, and you have a standard deviation of 1, then any findings with an average between 9 and 11, mean the experiment has not had a significant enough deviation from the norm to conclude that the experiment had any effect at all. In other words, if you're average goes from 10 to 10.5 in that scenario, a scientist would deem that inconclusive.

And beyond that, over the last 50 or 60 games, they were within a few points of the top team in the East. They don't need much improvement over what they accomplished most of last year. In fact, a few more shootout goals would do it. Which isn't really improvement at all.

Like it or not that trade (so far) has worked out well for both teams. Phaneuf played great for OTT last year and they likely don't get as far as they did if Wideman or whoever is playing in that spot. (Also, OTT would have still had Greening, Michalek and maybe Cowen on the roster, and missed the playoffs altogether)
Moving Phaneuf made TOR's D worse in the short term (and their forwards too), which enabled them to complete the tank for Matthews, so everybody wins.
- wilkobecks

I was talking about contracts

*Sigh* You didn't actually read the article and charts fully, did you?

Some of what you are saying is true. What you call standard variance is standard deviation (the Greek symbol for sigma, σ), and you can calculate it by finding the square root of the variance from the mean or average of the group you are comparing. I'll skip the calculation steps for you (good link here if you want to do it yourself: http://www.mathsisfun.com...a/standard-deviation.html) but the results are that the standard deviation for the palyers compared in the article between year 1 and year 2 is (rounded up from 0.096) 0.1 points per game. Any result above that total (positive or negative) means that there is statistical significance to the result. Everything else in the sample is just white noise like what you describe.

The author picked a single comparable to each of the Leafs rookies (Matthews=Kovalchuk, Marner = Kane, Nylander = Brad Richards, etc.) and shows that those individual players improved in their second year, and since they have similar attributes the Leafs rookies output will likely be the same in their second seasons. What was overlooked there is that the cherry picked comparables were always among the best performing players in each group assessed. If you read the background charts for each group compared, there are a larger number of players who reported declines or no significant change in their output in the next season. In statistics terms that is taking the most extreme positive outlier in a sample and trying to frame it as the normal average for the group. It's BS and does nothing but confirm the point I was making, that most hockey analysis in the media and blogosphere (people like Yost being exceptions) is done to suit a narrative, not to see what the most probable outcome is likely to be.

Here's how the players in each grouping actually performed when broken down by actual results for each player. My method is any player with more than a 0.1 Pts/GP ratio is a statistically significant improvement or regression, while those within the 0.1 range are seen as no change year over year:

Matthews Comparables:
Stat. Significant Improvement: Nash, Kovalchuk
No Change: Kane, Toews, Kopitar, Gaborik
Stat. Significant Regression: Lecavalier
% of all comparable Players assessed with statistically significant improvements in their second season: 28.6%

Nylander Comparables:
Stat. Significant Improvement: Thornton, Draisaitl
No Change: Monahan, Richards, F. Forsberg, Nugent-Hopkins, Mackinnon, E. Kane (barely)
Stat. Significant Regression: Lecavalier
% of all comparable Players with statistically significant improvements in their second season: 22.2%

Marner Comparables
Stat. Significant Improvement: Kovalchuk, Nash, Eichel, Duchesne
No Change: P. Kane, Toews, Kopitar, Gaborik
Stat. Significant Regression: Lecavalier, Skinner
% of all comparable Players with statistically significant improvements in their second season: 44.4%

Connor Brown Coomparables:
Stat. Significant Improvement: Arnason, Plekanec, Brouwer
No Change: Cole
Stat. Significant Regression: Donski, Chiasson, Carcillo, Janmark, Purcell
% of all comparable Players with statistically significant improvements in their second season: 33.3%

So, based on all the Leaf rookie comparable players, the large majority of comparables saw either no significant improvement or a regression in their second season compared to their first. To me that says that it is very very optimistic (unrealistic, actually) to expect that in each case the Leafs rookies will see improved results in their play over last year.

You are right, there is also very little statistical significance in terms of output changes for a lot of the players but all that means is that no significant improvement should be expected from this Leafs group. It's also important to point out that in each case, the players used in the sample comparison are also cherry picked by the author to suit their conclusions, rather than being representative examples. This is p-hacking and is generally seen as unacceptable for data analytics work like this.

I didn't saying that a regression was definitely going to happen, only that the Leafs should not be expecting any significant improvements in terms of individual scoring, or any extra points in the standings as a result. If the team is going to finish any higher, it will in all likelihood be on the back of improved defensive play, and they have not done much of anything to address that yet.

I'm sure you will not read any of what I just wrote above and will continue to disagree with me just on principle, but the article does nothing but confirm what I said if you know how to read the numbers and how to figure out when they are being cherry picked to suit a need. It's not that the Leafs should expect a regression, just no improvement from last year.
- ahjnkn

I really like that you have put some effort into this.

Unfortunately, I see some flaws in the methodology...

Hand-picking any group of "comparables" is not scientifically sound. Just because point totals or size or "style" ("style" is not scientific) are comparable, we have no idea whether team or coaching changes, individual player conditioning, physical maturity, attitude etc. are comparable. These things ALL have a significant impact on a player's development, consistency and statistics.

On a similar line of thinking, I would note (albeit, anecdotally) that the opinions of commentators or those who write about hockey seems to be that young players in general are much better prepared than they were in the past, even as recent as 10 years ago.

Without picking the entire thing apart, I would suggest your final line regarding the Leafs points may be wrong. You ignored a very significant change in the team's performance from the first couple months of the season to the second half. The Leafs led the league in blown 3rd period leads... they improved. They also were the worst extra time team in the NHL (I think?) and I believe even that improved a little. The Leafs lost a lot of points early in the season. The 3 on 3 and shootout performance has a very large luck component (not entirely) but also has a very large experience component. Inexperience and luck should not continue.

I respect that you have attempted to predict using analysis. I just do not believe that the analysis can possibly include all the variables and I think the analysis is ignoring the evidence that the Leafs already started growing and maturing last season.

Also... the Leafs defense is not done yet (probably). You can't say who is even on the team next year but I think you would agree it is unlikely to be worse. The Leafs best defenseman also had a horrible season. That is very unlikely to continue.

Like it or not that trade (so far) has worked out well for both teams. Phaneuf played great for OTT last year and they likely don't get as far as they did if Wideman or whoever is playing in that spot. (Also, OTT would have still had Greening, Michalek and maybe Cowen on the roster, and missed the playoffs altogether)
Moving Phaneuf made TOR's D worse in the short term (and their forwards too), which enabled them to complete the tank for Matthews, so everybody wins.
- wilkobecks

Phaneuf was 7 million for your second pairing. He was not "great". Your team also tried to give him away to Vegas and then tried to give him to anyone for a discount.

I have a hard time reconciling "great" with the team wanting rid of him. I have read a ton of times how his salary does not hurt the team because the Sens aren't a budget team... then why are they trying to get rid of him?

It is cherry picking for two reasons. 1) The years of the Calder winners selected are arbitrary. 2) Your point about the Leafs rookies being at the top makes no sense when you consider the Connor Brown comparison in the article as an example. I also didn't know there were 5 Calder trophies awarded each season, that fact was very enlightening, thank you (see? misunderstanding people's statements is fun!!).


- ahjnkn

Oh and, for the record, I do appreciate the analysis, but the arrogance in this paragraph should have stayed out of it.

Your analysis is not perfect. It doesn't earn you the right to patronize and insult others.