Proposal: Fire DJ Smith

[HSF]

Here's an even crazier take. I like to judge a successful season by looking at the standings at the end of the season.

yup after the pressure is off and teams win meaningless games?

 

[Micklebot]

I'm not.

You're right, in your example, that if they had allowed no bad goals (assuming 6 is the correct number of "bad" goals allowed this year), the SV% would be .916%.

My example was that if they had made fewer extraordinary saves (let's say 6, to keep that number the same), that number could also easily be 86 goals allowed on 885 shots, which would make the SV% .903%.

If we're talking hypotheticals, you can't remove the bad goals but keep the exceptional saves. Yes there have been bad goals, but there have also been great saves, so it balances it out and that .910 is a pretty accurate representation of the goaltending we've gotten.

And at the end of the day, a combined .910 SV% is solid. Not amazing, but solid.

Which is my point.

If Dorion legitimately thinks we should/could maintain a combined .915%+ SV% over the course of an entire 82 game season, based on who we have in net and how we play, I think he's out to lunch.

Last year, the Lightning had a combined SV% of .913%. With an amazing forward group, defense, and the best goalie in the league.

Now, if we can maintain a .910 SV% and get SA/game down to ~30-31.5, then we're talking...

My bad, I missread your post,

My point wasn't that they should be at .916 because they let in some softies that we should exclude, it was more about how the current sample is small so a small change can get us there. Goaltending is pretty erratic year to year, guys have swings that could easily result in a jump from .910 to .916 one year to the next, it's a high mark, but I don't see it as unachievable.

Also, I don't think he set it as an expectation for success, I think he meant it as a mark that would guarantee success.

Just as an aside, since the 19-20 season, Talbot and Forsbergs combined sv% is .9135,

 

[aragorn]

I like to judge a successful season by looking at the standings

Crazy take I know

You are what your record says you are.

 

[Golden_Jet]

EN are included in team stats but are not goalie sv% which is what Dorion referenced. Still a fairly lofty expectation to hit .916-.918, but the question of "this season will be a success if:" could be interpreted as what guarates a success rather that what is a requirement of success.

Thanks. I knew goalie ones didn’t, but assumed team did, but wasn’t sure.

 

[swiftwin]

yup after the pressure is off and teams win meaningless games?

All games count the same. Wins are worth 2 points, OTL 1 point, and 0 for a loss. Doesn't matter if it's October or March.

 

[DaveMatthew]

All games count the same. Wins are worth 2 points, OTL 1 point, and 0 for a loss. Doesn't matter if it's October or March.

You're right. At the end of the day, a win is a win and a loss is a loss. And really, that's all that matters.

It's interesting though because so often, your posts argue that this season, each loss hasn't been a loss. There have been many times we're you've said that even though we received 0 points, you couldn't just judge the team based on that.

You can look at and analyze an entire season with nuance, just like you can a game.

A winning streak after you've been eliminated from playoff contention can be looked at differently from a winning streak in the midst of a playoff battle down the stretch.

Good news is, we haven't yet been eliminated from the playoffs yet. And this is a big week. We can be .500 on Sunday and have beat two of the teams in front of us. Let's see if the team can step up.

I'd value a 5-game winning streak right now (to push our record to 16-14-2) more than a 5-game winning streak at the end of the season if we're 30-45-2 with 2 weeks left to play.

 

[swiftwin]

You're right. At the end of the day, a win is a win and a loss is a loss. And really, that's all that matters.

It's interesting though because so often, your posts argue that this season, each loss hasn't been a loss. There have been many times we're you've said that even though we've received 0 points, you couldn't just judge the team based on that.

You can look at and analyze an entire season with nuance, just like you can a game.

A winning streak after you've been eliminated from playoff contention can be looked at differently from a winning streak when you're battling for a playoff spot down the stretch.

That's not the reason I fuss over the underlying metrics. My reasoning is this:

1. The only thing that matters is our standings at the end of the season.
2. Underlying metrics is a better predictor of future standings than current standings, especially this early in the season.

 

[Golden_Jet]

Unbelievably he’s had 2 months with a winning record since he’s been here.

Trouble with charts eh
Oh let me guess your counting the bonus points in the loss column, that’s not How NHL does it unfortunately.’

Even if you do count it the wrong way, it would be 4 not 2.

21 months total
10 months months P% >= 500

 

[DaveMatthew]

That's not the reason I fuss over the underlying metrics. My reasoning is this:

1. The only thing that matters is our standings at the end of the season.
2. Underlying metrics is a better predictor of future standings than current standings, especially this early in the season.

Underlying metrics are a very useful way to understand and put into context the situation you're in at present, but they're not necessarily a great predictor of the future because in the future, the underlying metrics can easily be different.

The entire season is 82 games, which is an incredibly, incredibly small sample. Not really enough time for things to average out. I mean, the "deserve-o-meter" you often reference runs 1000 simulations of each game.

 

[swiftwin]

Underlying metrics are a very useful way to understand and put into context the situation you're in at present, but they're not necessarily a great predictor of the future because in the future, the underlying metrics can easily be different.

Sure, they can be different. But it's still a better predictor than wins and losses. With underlying metrics, you're using thousands of datapoints. For W/L, and ultimately our standings, you only have 26 datapoints.

Our underlying metrics have only improved in recent weeks. We're now in the top 10 in just about every metric out there (xGF%, CORSI, PP+PK, etc.). Even our xGA/60, was hovering around 25th a month ago is now 18th. Our xGF/60 is still 4th best in the league.

 

[DaveMatthew]

Sure, they can be different. But it's still a better predictor than wins and losses. With underlying metrics, you're using thousands of datapoints. For W/L, and ultimately our standings, you only have 26 datapoints.

Our underlying metrics have only improved in recent weeks. We're now in the top 10 in just about every metric out there (xGF%, CORSI, PP+PK, etc.). Even our xGA/60, was hovering around 25th a month ago is now 18th. Our xGF/60 is still 4th best in the league.

If the season had 1000, or better yet 10,000, games left, we could expect that things would average out and we'd end up winning more than we lose, based on the underlying numbers.

But we only have 55 games left. Impossible to predict.

The season is finite. So it's appropriate to criticize the actual results (wins and losses) of the first 27 games, 33% of the season.

The fewer games that remain, the less relevant the underlying metrics become. "We'll be fine look at the underlying metrics" held much more weight after 5 games than it does after 27. In 10 more games, it'll mean even less.

 

[swiftwin]

If the season had 1000, or better yet 10,000 games, left, we could expect that things would average out and we'd end up winning more than we lose, based on the underlying numbers.

But we only have 55 games left. Impossible to predict.

The season is finite. So it's appropriate to criticize the actual results (wins and losses) of the first 27 games, 33% of the season.

The fewer games that remain, the less relevant the underlying metrics become.

You're absolutely correct. That's why all the games we're playing right now are important, meaningful games. At a certain point, we'll run out of runway.

Fortunately, 55 games is still alot of games, but we have very little margin for error, and we can't afford another big losing streak.

Moneypuck still shows us with a 28% chance of making the playoffs.

 

[Alex1234]

Here's an even crazier take. I like to judge a successful season by looking at the standings at the end of the season.

Did exactly that the last 4 years

Was fun

 

[swiftwin]

Did exactly that the last 4 years

Was fun

Cool. Fortunately, our standings the last 4 years have no impact on our standings this year.

 

[HSF]

That's not the reason I fuss over the underlying metrics. My reasoning is this:

1. The only thing that matters is our standings at the end of the season.
2. Underlying metrics is a better predictor of future standings than current standings, especially this early in the season.

It might be better but its not that good with a small game sample size.

 

[YouGotAStuGoing]

Cool. Fortunately, our standings the last 4 years have no impact on our standings this year.

If past results can't be used in the calculation, then what difference do the underlying metrics this year serve? We're not going to have the same team next year, or the year after. Hell, even by the trade deadline I'd be surprised if we looked exactly the same. So where's the value in using underlying metrics as prescriptive?

 

[swiftwin]

It might be better but its not that good with a small game sample size.

It's still the best thing we have. What do you suggest is a better predictor of future results?

 

[swiftwin]

If past results can't be used in the calculation, then what difference do the underlying metrics this year serve? We're not going to have the same team next year, or the year after. Hell, even by the trade deadline I'd be surprised if we looked exactly the same. So where's the value in using underlying metrics as prescriptive?

Like I said, they are a predictor of future results. Earlier in the season, during our losing streak, when were in 2nd last place, I pointed at the metrics, saying we are not the 2nd worst team in the league, and that the metrics show we were going to have better results in the future. It proved to be true, because now, today, we're 5-2-1 in our last 8, and are up to 6th last place. I'm still using the metrics to show that we are still going to keep rising in the standings.

 

[HSF]

It's still the best thing we have. What do you suggest is a better predictor of future results?

the numbers for the last 4 years under DJ and Smith

 

[Golden_Jet]

the numbers for the last 4 years under DJ and Smith

DJ and Smith are 2 people now.

 

[swiftwin]

the numbers for the last 4 years under DJ and Smith

Lmfao. Different personnel. Different systems. You can't make an apples to apples comparison across seasons like that, other than maybe to show progress. Like, are you going to compare Jared Bednar's 47 point season with the Avs to today? Of course not.

 

[HSF]

Lmfao. Different personnel. Different systems. You can't make an apples to apples comparison across seasons like that, other than maybe to show progress. Like, are you going to compare Jared Bednar's 47 point season with the Avs to today? Of course not.

Went from 48 to 95 points. Would be nice to have a coach with those kinda results

Is our current record even better than last year?

 

[DaveMatthew]

Like I said, they are a predictor of future results. Earlier in the season, during our losing streak, when were in 2nd last place, I pointed at the metrics, saying we are not the 2nd worst team in the league, and that the metrics show we were going to have better results in the future. It proved to be true, because now, today, we're 5-2-1 in our last 8, and are up to 6th last place. I'm still using the metrics to show that we are still going to keep rising in the standings.

They are not.

They can be used as reasoning that helps you formulate a prediction, but they are not, in and of themselves, predictors of future results.

Someone else could back up their prediction with different reasoning (basic stats like W/L, GF/GA), and it wouldn't be any less valid.

 

[swiftwin]

Went from 48 to 95 points. Would be nice to have a coach with those kinda results

Is our current record even better than last year?

We were 9-17-1 at this point last season. We're 11-14-2 this season. Also, we were 25th in xGF% last year, and are 10th in xGF% so far this year, which means not only are we better than we were at this point last year, it's likely the gap between this year and last year is only going to get better as the season progresses.

 

[Golden_Jet]

They are not.

They can be used as reasoning that helps you formulate a prediction, but they are not, in and of themselves, predictors of future results.

Someone else could back up their prediction with different reasoning, and it wouldn't be less valid.

That’s what xGA , xGF, GSAX, and others stats all due now, try to predict, and yes nothing is guaranteed, I think we all know that.