NHL Draft Software: 2018 Rankings

[ProspectsFanatic]

DY: Draft year. DY-1: 1 year before draft. DY-2: 2 years before draft.
The first section (with DY, DY-1, DY-2) are the evaluations of each prospect season performance calculated by the software based on the data entered.
The second section shows the percentage taken into account in the final score, it varies depending on the games/tournaments played for each season.

Click to read more...

 

[ProspectsFanatic]

You can compare with last year and have some infos about the software here (the comparison isn't perfect though since I keep on improving the algorithm):
http://hfboards.mandatory.com/threa...draft-eligible-players-data-analyzer.2350541/
Software evaluating draft eligible player



___
If some of you guys are interested, I could enter in the software any player you desire, you can sort of test the software with hard to evaluate ones if you so desire (the software can analyze player statistics from practically any men leagues and development leagues, USHS is the main one the software can't analyze). Based on the final result the software can give you a range of where the prospect should be drafted at.

You also need to give me a scouting evaluation of the player on those 7 criteria: Skating, Shooting, Puck Control, Playmaking, Competitiveness, Physical Play and Defensive Play (Skating is the most important to have right since it weights the most in the software), it needs to be judged on that scale: Exceptional, Exceptional/Excellent, Excellent, Excellent/Very Good, Very Good, Very Good/Good, Good, Above Average, Average, Below Average, Bad, Very Bad. The average 3rd rounder, for example, should normally average a rating of Very Good/Good, maybe on the lower side if the major upside in his game is in his size. Extreme evaluations (trending towards Exceptional or Very Bad) should in most cases be avoided. Those ratings will affect how the software will evaluate the player statistics.

It needs to be his first draft eligible season (I am updating the algorithm for overager).

 

[Sens of Anarchy]

Very interesting ... How can I find out more about this?

 

[StatisticsAddict99]

Quick question, why is there two DY boxes and what are they valuing

 

[ProspectsFanatic]

Very interesting ... How can I find out more about this?

It has been less than a year that I started developing the software, still working on improving his precision. I believe a well programmed software has the potential to give better results than the standard way of evaluating prospects, because of the high complexity of player evaluation; you need to be heavily accurate while factoring many variables at the same time which are all relative to many other variables affecting one another (points relative to league to age to ice time to quality of teammates, etc, etc and in comparison with scouted attributes and projection). I am currently in a discussion with a programmer to embed the software within a website so people can experiment with it. Feel free if you have any more questions.

Quick question, why is there two DY boxes and what are they valuing

DY: Draft year. DY-1: 1 year before draft. DY-2: 2 years before draft.
The first section (with DY, DY-1, DY-2) are the evaluations of each prospect season performance calculated by the software based on the data entered.
The second section shows the percentage taken into account in the final score, it varies depending on the games/tournaments played for each season.

 

[Rasmus CacOlainen]

Is winger versus center taken into account? Normally being a C is a privilege over W if all else is equal.

Also...looks quite interesting and I like last years results from your extract.

 

[Sens of Anarchy]

It has been less than a year that I started developing the software, still working on improving his precision. I believe a well programmed software has the potential to give better results than the standard way of evaluating prospects, because of the high complexity of player evaluation; you need to be heavily precise while factoring many variables at the same time which are all relative to many other variables affecting one another (points relative to league to age to ice time to quality of teammates, etc, etc and in comparison with scouted attributes and projection). I am currently in a discussion with a programmer to embed the software within a website so people can experiment with it. Feel free if you have any more questions.


DY: Draft year. DY-1: 1 year before draft. DY-2: 2 years before draft.
The first section (with DY, DY-1, DY-2) are the evaluations of each prospect season performance calculated by the software based on the data entered.
The second section shows the percentage taken into account in the final score, it varies depending on the games/tournaments played for each season.

Really cool.. you may be on to something here... good results

 

[ProspectsFanatic]

Is winger versus center taken into account? Normally being a C is a privilege over W if all else is equal.

Also...looks quite interesting and I like last years results from your extract.

Good point, I did consider it, but since the software greatly values scoring and that center normally get a few "free" assists by taking faceoffs I decided not to give them a further advantage. Otherwise, if you are referring to the fact that centers tend to be more complete players with better 2-way game, that's already taking into account in the players evaluation where you need to judge the player offensive and defensive IQ. Also, many of them end up becoming wingers anyway at the NHL level because more centers are being drafted proportionally (better players tend to play center and drafted prospects are the better ones at the junior level).

 

[JThorne]

Maybe add in their position? And country(for those who really care)?

 

[ObscureAlien]

will you be adding more players into the software?

 

[tony d]

Looks interesting, thanks for sharing.

 

[StatisticsAddict99]

What year did you start doing these rankings?

 

[ProspectsFanatic]

will you be adding more players into the software?

Yes, but entering new players is very time consuming, I need scouting evaluation for those players, so I am not sure when that would be.

What year did you start doing these rankings?

Last year was the first time I produced rankings with the software.

 

[ProspectsFanatic]

Rankings updated in OP (stats updates, some players evaluation changed, minor tweaks to the algorithm)

- Dahlin and Svechnikov currently clear cut at #1 and #2 respectively.
- 5 players race for #3 with Zadina leading in front of Wahlstrom, Boqvist, Tkachuk and Bouchard.
- Players competing to round out the 3 spots left in the top10 (in order): Dobson, Hughes, Veleno, Kotkaniemi, Merkley and Smith.
- It is very close in the mid first round, players could easily move up and down.
- Interesting inclusions rounding out the first round: Jan Jenik, Bulat Shafigullin and Filip Hallander.
- Notable omissions that didn't quite make it in the first round: McBain, Groulx and Denisenko.
- Player to watch, on the rise: Kotkaniemi

Let me know what you guys think of this list?

 

[Pierce Hawthorne]

Honestly... This looks really, really good. I'd be curious about all the statistics that are used and how they're weighted to produce the data.


But just looking at the numbers and the rankings, it looks highly accurate to me.

 

[StatisticsAddict99]

DY: Draft year. DY-1: 1 year before draft. DY-2: 2 years before draft.
The first section (with DY, DY-1, DY-2) are the evaluations of each prospect season performance calculated by the software based on the data entered.
The second section shows the percentage taken into account in the final score, it varies depending on the games/tournaments played for each season.

So what your saying is the first Draft Year box is overall season statistics and the second box is they’re stats in tournaments and playoffs?

 

[ProspectsFanatic]

Honestly... This looks really, really good. I'd be curious about all the statistics that are used and how they're weighted to produce the data.

But just looking at the numbers and the rankings, it looks highly accurate to me.

You can see a sample of all the data taken into account in last year draft here:
Software evaluating draft eligible player

So what your saying is the first Draft Year box is overall season statistics and the second box is they’re stats in tournaments and playoffs?

The second section simply shows how much each season are weighted in the software, the total of all 3 seasons will always be 1 (100%). Tournaments and playoffs performance are taken into account with the season statistics in the first section.

 

[ProspectsFanatic]

If some of you guys are interested, I could enter in the software any player you desire, you can sort of test the software with hard to evaluate ones if you so desire (the software can analyze player statistics from practically any men leagues and development leagues, USHS is the main one the software can't analyze). Based on the final result the software can give you a range of where the prospect should be drafted at.

You also need to give me a scouting evaluation of the player on those 7 criteria: Skating, Shooting, Puck Control, Playmaking, Competitiveness, Physical Play and Defensive Play (Skating is the most important to have right since it weights the most in the software), it needs to be judged on that scale: Exceptional, Exceptional/Excellent, Excellent, Excellent/Very Good, Very Good, Very Good/Good, Good, Above Average, Average, Below Average, Bad, Very Bad. The average 3rd rounder, for example, should normally average a rating of Very Good/Good, maybe on the lower side if the major upside in his game is in his size. Extreme evaluations (trending towards Exceptional or Very Bad) should in most cases be avoided. Those ratings will affect how the software will evaluate the player statistics.

It needs to be his first draft eligible season (I am updating the algorithm for overager).

 

[Kevin Musto]

Hey look! It properly ranked Kotkaniemi and Kupari!

 

[ProspectsFanatic]

Hey look! It properly ranked Kotkaniemi and Kupari!

Yep, not only do I have Kotkaniemi over Kupari, but by a pretty big margin, will be interesting to follow how this will be developing.

 

[slumpy43]

Nice job on the modeling. Impressive that you combine more than stats in the models which is likely way more difficult to weigh the different factors. How did you start?

Did some modeling myself in the early 2000's and got 3 variables of importance: age, ppg and proportion of teams points. Based on that old relationship, had Robertson in the top 10 last year...time will tell. Now that ESP and PPP are out there, have you found any relationship with those indices and NHL success? Any differences on the importance of the factors between Dman and forwards?

 

[ProspectsFanatic]

Nice job on the modeling. Impressive that you combine more than stats in the models which is likely way more difficult to weigh the different factors. How did you start?

Did some modeling myself in the early 2000's and got 3 variables of importance: age, ppg and proportion of teams points. Based on that old relationship, had Robertson in the top 10 last year...time will tell. Now that ESP and PPP are out there, have you found any relationship with those indices and NHL success? Any differences on the importance of the factors between Dman and forwards?

How did I started? Well like 6 years ago I started doing my own draft has the Habs to see how I would compare, over the years it became apparent to me that I could achieve better results with a software which could juggle with more variables at once all of which to greater precision and find further consistency in prospects evaluation. I started on a blank sheet and slowly progressed, at the beginning it looked horrible and I wasn't sure if it would actually amount to anything, but I persisted, and with my very perfectionist nature and having a certain natural ease with numbers I managed to produce something which I consider to be of decency.

Yeah Robertson was a challenge, he made me reconsider how the software was evaluating players, he had such terrific statistics, even more so considering his young age and poor teammates. Scouting evaluation being factored in by the software is a must in order to achieve representative results. The problem with Robertson was his questionable skating, you can get away with subpar skating in junior by compensating with some other tools, but this handicap will accentuate has you move higher up the pro levels. Skating is the attribute which I am valuing the most in a prospect by far, that's why he is where he is at in my rankings.

About the ESP and PPP I am not sure if I found any type of relationship between them and NHL success. The important thing on that subject is heavily factoring ice time(particularly PP time) when looking at offensive production, that's how I ended with Formenton being ranked even a bit higher than where he actually got drafted in my rankings I believe. I even improved the algorithm to further account players with limited ice time this year.

For Dman; I believe the software should perform a bit better with forwards since their contribution translates more into numbers which can be easily traceable by the software. For Dman, I am valuing more highly the scouting evaluations, particularly the defensive side, it is a matter of founding the right balance I believe. Unless you are 6'7" if you can't produce at the junior level I can't see your game translating to the NHL level, so there is value to be had there in numbers. I also look much more in the +/- for Dman.

 

[slumpy43]

Could only find limited information but no stats on time on ice. Scott Reynolds brought an interesting concept of estimating a players time on ice (How Well Do Scoring Events Reflect Ice Time?)
Unfortunately, I could only locate limited data at Prospect-Stats
Do you know of any other sites that carry such information?