AfroThunder396
[citation needed]
Who gives a shit about the draft pick, no one here has a time machine. Zacha is the guy the Devils have and he's turning into a very nice player.
If the ‘15 draft was so incredible, it wouldn’t be a bad pick. But it was, and it was.
, before I get into it, may I ask what context you’re referring to... the poster you quoted gave a sample size whereas you did not.No it does not.
, before I get into it, may I ask what context you’re referring to... the poster you quoted gave a sample size whereas you did not.
Speaking is smaller samples- no, it’s not predictive. The bias in the data should be quite evident when comparing to historical data. A good basis for Corsi, in my opinion, is 1/3 of the season to really see where that player is.
The spread for a predictive stance should be small enough to account for outliers, unless the player goes on a tear offensively for 10+ games or so (they’re playing better than they normally do).
Do I agree, yes in the context that it doesn’t take a whole season, but I don’t really know where you stand since you never put up a platform
Ummm, the R^2 is .14 which is quite horrible. If you want me to create a standardized model of what the regression accounts for, I will do so, but the variability of the data is not described by the regression: therefore, even in a predictive stance 14% of data accounted for in volatility doesn’t really give you an answer.Yes, the poster that I quoted gave a sample size. They said that Corsi requires at least one season's worth in order to be predictive. They did not add ANY context to that comment, so I didn't feel the need to add any context to mine.
You want a platform? Here is every team's 5-on-5 Corsi For % in October, and their corresponding point total at the end of the season.
The R^2 and p-value show you that while the predictive value of Corsi in just October not extremely strong, it still certainly does have predictive value and the R^2 for that predictive value is very significant at any confidence level. And remember, this is just October.
If you want to look at the interactive chart on Tableau, follow this link here.
Ummm, the R^2 is .14 which is quite horrible. If you want me to create a standardized model of what the regression accounts for, I will do so, but the variability of the data is not described by the regression: therefore, even in a predictive stance 14% of data accounted for in volatility doesn’t really give you an answer.
While the P value suggests you can reject the sentiment of no correlation, the data points go against that because they’re pushing such a high volume of time-correlates data. You don’t need all of those sample sizes. That person who did the model seems to be unsure of their model, like a cook who drowns their food in a sauce to cover up any flavor discrepancies. Throw 2 data points up for each team then see what the line of best fit looks like. It surely won’t look like that.
Honestly, that model looks like a graph of Corsi-to-Points relationship and then someone just threw a line of best fit then used the parameters thing in Excel to gather statistical data. Kinda weird.
I will agree with you, but using ~18% of games to measure the impact that will have in your season leaves a lot of room for error.
Because it’s hard to tell if there actually is a correlation.The point was that it only used such a small percentage of the games played. Literally just the month of October. And that it is still predictive even if the predictive value is as small as 14%.
I have no idea what you are talking about with your silly simile of a cook who drowns their food in a sauce and I have no idea why you are suggesting that you don't need sample sizes. Why not use all of the data that we have available to us?
I remember Craig Button saying this guy should be picked at the end of the 1st round at best
but he guaranteed he'd be picked in the top 10
It requires 20 to 30 games. Not a whole season or more.It's stupid because corsi as a stat requires at least a season's sample size if not more in order to properly be predictive. People are misusing it so much nowadays.
Source?
https://www.tsn.ca/craig-s-list-provorov-leaps-into-top-5-1.204945Source?
Don't look now, Zacha is on a 40 point pace! I stand by what I said before.Always a believer and even as a #3C, can be an important cog on the Devils. Him taking harder minutes will be very valuable to Nico and Hughes going forward as they can get easier ones.
If you look at his points, it's stagnant at 24, 25, 25. But games played show something different because he has missed some.
3 years ago 28 point pace
2 year ago 30 point pace
last year 34 point pace
So it's nothing groundbreaking, but it is improvement and he is just not an offense guy. He is good defensively and good on the PK. That doesn't show up in easy to look at stats, but teams value and need those players.
Couturier never paced for more than 40 points until his 5th season...then jumped to 42 and 52 pace for 2 years...then on to 76 point seasons. So maybe Zacha will have a similar path. Even if he gets to that 40-50 range that will be a very good player for the Devils.
He's looked good between gusev and bratt.Don't look now, Zacha is on a 40 point pace! I stand by what I said before.
How has he looked this year in getting those points?
Not great but lately with Bratt and Gusev he looks great, he is driving holes with the puck and generally being more aggressive. then again Gusev is making whoever he plays with look great it seems. Everyone has confidence playing with him.Don't look now, Zacha is on a 40 point pace! I stand by what I said before.
How has he looked this year in getting those points?
Don't look now, Zacha is on a 40 point pace! I stand by what I said before.
How has he looked this year in getting those points?
He's looked good between gusev and bratt.
Hes on a crappy team, somebody has to put up points. Its the same as Duclair, Spooner etc.