Analyzing the last 10 games to Nov 6

[Loyal and True]

The common statement from pundits is that the Flames have brutal possession stats and therefore will regress significantly. Looking at last 10 games and the rest of the schedule (67 games left), I'm not sure that's accurate.

I used http://war-on-ice.com/teamtable.html with Even Strength 5v5, in Close Score Situations:

Last 10 games 5-3-2 .600%
5 Home/5 Road
Fenwick% 50.26% (17th overall)
PDO 98.59 (SH% 6.52 + SV%92.06)

There is nothing unsustainable about the last 10 games. The schedule has been fairly balanced. True, 8 of last 10 are against East, but most are the stronger East teams plus a couple of solid Central teams.

Quick look at the Fenwick% (5on5 Close) of our competition over the same time frame:

2 games vs WSH 55.30%
2 x TB 51.37%
2 x MTL 50.00%
1 x CAR 49.19%
1 x CBJ 46.69%
1 x WPG 51.34%
1 x NSH 51.80%

In my book, the Flames are full value for the .600 record of last 10 games. They are not a possession powerhouse, but they can sustain this if the goaltending stays above .920 (which it should)

Looking at the remaining 67 games, 25% will be against heavyweights, 25% will be against weak sisters, and 50% against the middle teams that are comparable to our last 10 opponents.

The first 5 games of the year were bad (possession wise - yet we were still 3-2-0 thx to Hiller & Ramo), but that was tough circumstances. 1 Home game and 4 road games. 5 games in 8 nights. A Vcr/Edm back-to-back, then road game in STL, then a NSH/CHI back-to-back set.

I am not suggesting the Flames will be .600 all year. But I am suggesting that the posession stats argument for regression is not that solid.

[dissentowner]

Even in the loss last night they had plenty of chances and shots against one of the best teams in the NHL in their rink. I think this team will be a bubble playoff challenger.

[Geeoff]

This 9th/10th place finish will be much less painful than the last time we were here.

[united]

10 games is a poor sample size as we all know. Using another metric, score-adjusted, instead of close boosts the sample size from 285 minutes to 531 minutes over those 10 games. Using score-adjusted, the Flames possession was 47.34% - good for 24th in the league in that time frame.

Regardless, 50.26% translates to about 93 points over the course of a season which is 5 points behind the Flames' current point per game pace. If they keep up 50.26% they should finish around 94-95 points which is good for a playoff spot. (47.34% translates to about 85 points FYI).

[Street Pharmacist]

I've said it before, but the metric that best predicts the future is goal differential. I don't think the sample size is there yet though

[Loyal and True]

Originally Posted by united

10 games is a poor sample size as we all know. Using another metric, score-adjusted, instead of close boosts the sample size from 285 minutes to 531 minutes over those 10 games. Using score-adjusted, the Flames possession was 47.34% - good for 24th in the league in that time frame.

Regardless, 50.26% translates to about 93 points over the course of a season which is 5 points behind the Flames' current point per game pace. If they keep up 50.26% they should finish around 94-95 points which is good for a playoff spot. (47.34% translates to about 85 points FYI).

Thanks, and sample size is definitely a factor in all this. We'll see how this measures over more games. I checked out fenwick-stats.com (and the linked Broad Street Hockey Article that I recall reading a while ago) and that looks interesting to me.

Can you tell me how you translate these % into points?

That 47.34% stems alot from 63 minutes of icetime where we were down by 2+ and EVF% was a pitiful 40.51% which is way way below average. A couple of bad periods. I realize it helps to boost the sample size but I think the % ends up skewed too much.

375 minutes (almost 2 periods per game) are either Down 1, Tied or Up 1. And in those 375 minutes the flames are 49.68% which is very respectable and at that level you can expect good goaltending to earn some points.

[Knut]

My Analysis of the last 10 games.



I am thoroughly entertained by the product on the ice.

[united]

Originally Posted by Loyal and True

Can you tell me how you translate these % into points?

I used the formula published by Bob Sullivan of Sporting Charts. As with anything of this nature, it's a very loose estimate.
Points = (249.1 * possession) - 32.6
http://www.sportingcharts.com/articl...tatistics.aspx

[Street Pharmacist]

Originally Posted by Loyal and True

Thanks, and sample size is definitely a factor in all this. We'll see how this measures over more games. I checked out fenwick-stats.com (and the linked Broad Street Hockey Article that I recall reading a while ago) and that looks interesting to me.

Can you tell me how you translate these % into points?

That 47.34% stems alot from 63 minutes of icetime where we were down by 2+ and EVF% was a pitiful 40.51% which is way way below average. A couple of bad periods. I realize it helps to boost the sample size but I think the % ends up skewed too much.

375 minutes (almost 2 periods per game) are either Down 1, Tied or Up 1. And in those 375 minutes the flames are 49.68% which is very respectable and at that level you can expect good goaltending to earn some points.

Again though, we can pick it the bad parts and say that's not how we play. A young team is going to have stretches like the first 5 games. They're going to have stretches where they're going to be down 2+ and not playing well.

In the end, the sum of the whole is far more likely the predictor of the future. If I was a betting man and someone showed me how well Edmonton plays except for the times they played terrible, I still wouldn't very on them.

[Loyal and True]

Originally Posted by united

I used the formula published by Bob Sullivan of Sporting Charts. As with anything of this nature, it's a very loose estimate.
Points = (249.1 * possession) - 32.6
http://www.sportingcharts.com/articl...tatistics.aspx

That's a great article too. Interesting that last year Flames could have had 85 points based on this. Obviously our goaltending was not good enough last year.

[hummdeedoo]

Entered thread expecting my eyes to glaze over by the 10th post...


leaving satisfied.

[Loyal and True]

Originally Posted by Street Pharmacist

Again though, we can pick it the bad parts and say that's not how we play. A young team is going to have stretches like the first 5 games. They're going to have stretches where they're going to be down 2+ and not playing well.

In the end, the sum of the whole is far more likely the predictor of the future. If I was a betting man and someone showed me how well Edmonton plays except for the times they played terrible, I still wouldn't very on them.

You're right. You don't strip out the bad parts. You just be careful that those parts are a fair representation and not weighted excessively in a small sample.

The first 5 games were not a fair representation of the remaining 77 games. The next 10 are a somewhat more balanced representation of the remaining 67 games.
Games 16 to 20 are a bit lighter schedule (knock on wood) and thus coincidentally the entire 20 game sample might be about right... Except of course we are missing 3 centremen (including Backlund) so who knows what will happen the rest of the year.

[GioforPM]

In analyzing the Flames last 10 games I find they are 5-3-2.

[Street Pharmacist]

Originally Posted by Loyal and True

You're right. You don't strip out the bad parts. You just be careful that those parts are a fair representation and not weighted excessively in a small sample.

The first 5 games were not a fair representation of the remaining 77 games. The next 10 are a somewhat more balanced representation of the remaining 67 games.
Games 16 to 20 are a bit lighter schedule (knock on wood) and thus coincidentally the entire 20 game sample might be about right... Except of course we are missing 3 centremen (including Backlund) so who knows what will happen the rest of the year.

But that's my point. Why are the last ten more balanced? Because they're better? We will very likely have another 5 game stretch of bad play. That's normal with any team especially a bad team. I guess what I'm saying is, what evidence is there that the first 5 games skews anything and the last 10 haven't been the skewed portion.

I think the most accurate look includes the whole data set and doesn't presuppose certain portions are less relevant simply because of a biased view of the team.

[Loyal and True]

Originally Posted by Street Pharmacist

But that's my point. Why are the last ten more balanced? Because they're better? We will very likely have another 5 game stretch of bad play. That's normal with any team especially a bad team. I guess what I'm saying is, what evidence is there that the first 5 games skews anything and the last 10 haven't been the skewed portion.

I think the most accurate look includes the whole data set and doesn't presuppose certain portions are less relevant simply because of a biased view of the team.

The reason these last 10 games are more balanced is the schedule. 5 home and 5 away. Strength of opponents are upper-middle class, mostly. It is a small (1/8 of a season) but balanced sample.

The first 5 games are not representative of the 82 game schedule. 5 games in 8 nights. Four on the road and one at home. 2 sets of back to back and a road game in StL sandwiched in between.

Obviously we can't fairly extrapolate from those 5 games alone, because we don't play 80% of our games on the road or 80% of our games on back to back sets. Likewise, at the 15 game mark, one-third of the possession stats fall under that 5 game stretch. I don't think the 15 games are balanced. We haven't played many of the weaker opponents yet and we've only played 6 home games so far. That is about to change.

Based on schedule and opponents, I believe the recent 10 games is better representation of a balanced schedule than the 15 to date.

I think by game 20 you can say the schedule has evened out more. Then you can look at possession stats for the whole 20 games and that would be fairly representative in my opinion.

[united]

Through my own cherry picking and without any historical backing, I too use the 20-game mark as the point when I expect what the stats show to be accurate.

On the other end of the season, I think I remember reading something that possession from the final 20 games of the season outperforms entire season goal differential or entire season possession in terms of expected postseason performance. That could also be made up - memory is hazy.

[djsFlames]

8th place finish, beating Anaheim, LA, and St. Louis on their way to a rematch with Tampon Bay in which they beat them in their own barn for the cup they should've had 10 years ago.

[PeteMoss]

The whole season is too small a sample right now, particularly when you limit it to close. The Flames got whacked by the Hawks and won. With the sample still being small, the Flames would have better 'close' stats if Hiller didn't stand on his end and they got blasted that game.

[GioforPM]

Originally Posted by PeteMoss

The whole season is too small a sample right now, particularly when you limit it to close. The Flames got whacked by the Hawks and won. With the sample still being small, the Flames would have better 'close' stats if Hiller didn't stand on his end and they got blasted that game.

By the same token, the Flames whacked the Canadiens in the first meeting but Price stoned them. (Glad they made up for it next time).

[FlameZilla]

Originally Posted by Hesla

My Analysis of the last 10 games.



I am thoroughly entertained by the product on the ice.

They've been fun to watch & leave it all out on the ice. We'll miss out on McDavid, but establishing a win-hungry culture which rewards hard work is more important than having an individual superstar on the team.