Dedicated rating for the 2.1 engine

This forum should be use to discust players ratings. / Ce forum devrait être utilisé par discuté des côtes des joueurs.
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Dedicated rating for the 2.1 engine

Post by 36Henry »

Note: this post is best read in the voice of Kenny Powers.

This post is intended to inform the community that my NHL ratings for the 2.1 engine will be released around the beginning of August as per usual.

Introduction

These are the only ratings available that are specifically designed for use with the 2.1 Sim engine. Others may claim to have ratings for this engine, but those ratings all have one thing in common, they don't provide a realistic simulation experience.

This forum is littered with examples of people who have tried using the 2.1 engine in combination with such ratings that weren’t designed for it. They bemoan the engine and how it just doesn’t work properly. Well, the outcome will only ever be as good as the input you provide. If you buy a brand new diesel car and fill it with petrol, you wouldn’t expect that car to run very well would you? That doesn’t mean there’s something wrong with the car; it just means you’ve made a mistake trying to run it on the wrong type of fuel. The same thing applies here; the 2.1 engine requires a completely different type of ratings in order to run properly.

And, as I stated above: these are the only ratings available that are purpose made to run with the 2.1 engine. The ratings have been around for a number of years now without a single complaint from the people who have bought them. In fact, everyone who bought them in the past returned to buy them again the following season. Would a product really have a 100% renewal rate with its customers if it didn’t deliver everything it claims to do?

The Process

While some producers of ratings focus on speed of release, these ratings are built with a very different focus in mind; realism. The objective is to always provide a product that delivers the most realistic simulation experience possible to its customers. A lot of people in the Sim-hockey community devote countless hours to their leagues and teams, only to end up with unrealistic simulations and/or seemingly random results. The main source of frustration, in my experience, is the fact that people still cling to the outdated and simply unbearable 1.5 engine. That was why I first decided to make these ratings available to others, an attempt to get the community to finally start migrating away from the 1.5 to the far superior simulation experience of the 2.1 engine. It’s an uphill struggle as preconceived notions and falsehoods are still being voiced by those who conclude that since they themselves haven’t been able to make the 2.1 function, nobody else could do it either. But with each passing year more and more people are making the jump which is really pleasing and keeps me motivated to continue developing and improving the ratings.

For me, the concept of realism in the simulation is a multilayered one. Breaking it down, we can identify five dimensions of realism which all need to be fulfilled in order to create a simulation experience worthy of release.
  • League level realism. Refers to how well the simulation corresponds to the real life hockey it is based on and supposed to mimic. Quantified by various stats such as goals per game, shots per game, average powerplay percentage and number of powerplays per game to name a few.
  • Positional realism. This is closely linked to the league level realism and quantified as the distribution of various stats (such as goals, assists, hits, blocked shots etc.) between forwards and defensemen.
  • Team level realism. All testing of these ratings is done using 100% accurate NHL rosters and line combinations (including ice-time distributions) in order to ensure an accurate level of competitiveness for the 30 NHL clubs.
  • Individual stylistic realism. One of the things that most annoy me with poorly made ratings is that there can be a complete lack of correlation between how a player plays the game in the real world, and how he plays it in the Sim. I’ve seen examples where Nicklas Backstrom would routinely register more shots (and goals) than Ovechkin! The horror. What the 2.1 engine provides for, in combination with specialized ratings, is an opportunity to mold each individual player so that his in-Sim performance will stylistically match his real world performance. Snipers will snipe, playmakers will pass and so on. Stylistic realism is absolutely essential to the enjoyment of a league experience, both in terms of the actual simulations but also, and perhaps even more so, when it comes to building a team. If players behave like they do in real life, building a team with specific roles in mind for each player becomes much easier to do.
  • Individual productivity realism. While the previous four levels are to various degrees largely ignored by other makers of ratings, they usually at least tend to pay some attention to this one. It simply means that players are expected to produce offensively (or in the case of goaltenders, as goaltenders) on par with what they did in real life.
Illustrations of all five dimensions of realism can be observed in the opening post of this old topic: http://sths.simont.info/Forum/viewtopic ... 11&t=12049 where there is also an in-depth look at the distribution of stats within a specific team, in that case the 2014/15 Columbus Blue Jackets.

While this philosophy of a multilayered concept of realism, and the attention to detail associated with it, forms the overall theme of these ratings, achieving it requires a meticulous approach to all phases of the construction.

How they are made

You need a few things to build ratings of this kind.

First, you need to have a strong understanding of the simulation engine you’re working with. What does each individual rating actually do? In what way will an adjustment to one rating effect the rest of that player’s game? What drives the balance of play forward? The reason there are no other acceptable rating packs available for the 2.1 engine is because they all fail at this very first hurdle.

Second, you need to have data to base your ratings on. I love data. You can never have too much of it so the more relevant data you can compile, the better. But just raw data will often not be enough, you need to be able to creatively use and enhance the data in several different ways to get the most out of it.

Thirdly: time. You don’t build a good set of ratings for the 2.1 engine in a day, or even a week. It takes a lot of time, at least a month of dedicated work, usually much more than that. Creating the first draft of the ratings in itself once you have compiled all the data doesn’t take very long by comparison. But running hundreds of test seasons tweaking and adjusting everything down to a level of detail where you end up running 20+ seasons just to get one or two individual players to perform the way they should, that takes time. And that commitment and attention to detail is what makes these ratings so much better than its rivals.

That attention to detail is also why we’ve never had to release a massive upgrade to fix a long list of mistakes and errors in our initial release. We take our time and we get it right.

As an example of how these ratings go above and beyond all others let’s take a closer look at the Penalty shot rating for skaters. This is obviously a very specific rating, quantifying a very specialized singular skill in a hockey player.

Since the NHL introduced the shootout for the 2005/06 season, 1786 games have gone to a shootout. No active player has taken more shootout attempts than Radim Vrbata who has scored 45 goals on 104 attempts. No player in NHL history has scored a better shootout goal than Marek Malik. Daniel Alfredsson scored the first ever shootout goal on the first ever shootout attempt (against Eddie Belfour). Overall though, there’s just not a lot of data to build a large enough sample size to create a very good base from which to quantify each player’s ability in the shootout using only actual shootout stats. You can take overall skill level into account, you can look at various stats for how players perform in pressure situations, but you can also go off the board and create something new, something unique. This is where dedication, creativity and attention to detail once again come into play.

By going through all the NHL shootout games that have been played (all 1786 of them) we’ve generated a database which lists how many NHL games going to a shootout each individual player has participated in. We have then listed how many times the player was selected to take a shot as one of the team’s designated three original shooters (also including games where the player was selected to shoot fourth, fifth, sixth etc. on a sliding scale). We then have what we need to produce a Shootout Selection Ratio and a Shootout Non-selection Ratio, unique for each and every player, to take into account when generating the Penalty shot rating. These stats don’t exist in any other set of ratings you’ll find here or anywhere else for that matter.

This is just one small example of the kind of thinking that typifies the level of detail and dedication behind these ratings. It also exemplifies the amount of time and effort that go into this process which in turn validates the higher selling price (which if you break it down to dollars per week of production puts us at less than half the price of other producers).

The purchase

The ratings are typically available around the first of August each year.

The ratings are typically based on the previous two NHL seasons (with a few exceptions I’ll outline below), weighted 67% to 33% in favor of the most recent season.

For Experience and Leadership we use our own database where we've tracked the complete careers for each player dating back to minor midget or equivalent play in junior leagues in Europe to make sure everything the player has done in his career is included. Captaincy, individual and team awards, participation in international events, all of it is included to impact the Experience and Leadership of every player.

For Durability we include the complete injury history of every individual player.

Fighting takes into account each player's complete fight card at all levels of play.

Discipline takes into account past suspensions for undisciplined play to some extent.

Penalty shot, as mentioned above, is another category where the rating is based on career figures rather than just the two most previous seasons.

As the ratings are intended to be used in a particular way, use of the Simulators own re-rating feature is not taken into consideration and therefore the Potential rating is not quantified.

The ratings include:
  • All players who appeared in at least one NHL regular season or playoff game over the previous two seasons.
  • All coaches who coached in at least one NHL regular season or playoff game over the previous two seasons.
  • All on-ice officials who appeared in at least one NHL regular season or playoff game over the previous two seasons.
These ratings are made for serious customers looking to have a league that goes beyond the standard 1.5 leagues. They were made for leagues driven by the same dedication to, and passion for, Sim-hockey as its creator.

The ratings come in two price categories:
  • Standard package: $75 US dollars.
  • DeLuxe package: $100 US dollars.
With the DeLuxe package the customer will receive full support and detailed instructions as far as preferred Sim settings and Overall formulas to get the most out of the ratings. Attached will also be a brief description of the 2.1 engine and some of the ways in which it differs from the 1.5 version in an attempt to make sense of why these ratings are unique. Customers who opt for the cheaper package will not receive these services, only the ratings.

DeLuxe customers who will want to purchase the DeLuxe package again the following year will be able to do so at the Standard package price, meaning the extra $25 dollars spent on the initial purchase will be a nice investment for future seasons.

New feature coming?

It's possible that there will be an AHL add-on available for the first time this season. There's no guarantee that one will be made, but data is being compiled and evaluated and if time can be found to develop an AHL product of high enough caliber (and if there is enough expressed interest in such an add-on from the community) it would most likely become available for purchase.
Image
august
The Crazy / Le Fou
Posts: 440
Joined: Mon Oct 23, 2006 7:00 am

Re: Dedicated rating for the 2.1 engine

Post by august »

We use it for a couple of years and they are by far the best ratings we ever had.
senshockey
The Accomplished One / L'Accompli
Posts: 71
Joined: Thu Aug 06, 2015 9:38 pm

Re: Dedicated rating for the 2.1 engine

Post by senshockey »

As great as this sounds, waiting until August is difficult for some leagues to operate. I'm currently 'shopping' for a new ratings provider and don't mind paying but typically in my league, we like having the new ratings prior to free agency which we aim for mid-July. Waiting additional weeks for these ratings could potentially delay our season. \

That being said, I'd love to see what the ratings look like and a sample output to back what you are claiming. Is there a lifetime subscription available or is it always $75/100 depending on the package desired?
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Dedicated rating for the 2.1 engine

Post by 36Henry »

senshockey wrote: Mon Jun 19, 2017 9:16 pm As great as this sounds, waiting until August is difficult for some leagues to operate. I'm currently 'shopping' for a new ratings provider and don't mind paying but typically in my league, we like having the new ratings prior to free agency which we aim for mid-July. Waiting additional weeks for these ratings could potentially delay our season. \

That being said, I'd love to see what the ratings look like and a sample output to back what you are claiming. Is there a lifetime subscription available or is it always $75/100 depending on the package desired?
Thank you.

I have sent you a pm addressing your points in some detail along with some samples of stats and ratings from last season.
Image
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Dedicated rating for the 2.1 engine

Post by 36Henry »

As I shared this with senshockey earlier, I thought I'd put it in the thread as well in case anyone else wanted a more recent sample of some stats generated by the ratings.

Here's something I compiled in November 2016 for another user who was interested in seeing some more details of the ratings. The examples shows a single test season compared to the NHL stats from the 2015/16 season.

First off, the league stats. This shows how well the simulation matches the NHL for various stats (what I referred to in the opening post as league level realism) as well as the distribution of several stats between defensemen and forwards (positional realism) and also the league average save percentage.

Image

Fighting is down league wide in the simulation compared to the NHL, but there's a reason for that. In the test season every team was 100% healthy with their best rosters dressed. That means by default that most of the fighters in the league were sidelined as healthy scratches, thus driving fighting down somewhat.

The second illustration shows the stats of one complete team, in this case I chose to show the Ottawa Senators based on senshockey's username.

Image

The red columns show the NHL stats for each player, prorated to a full 82 game season. A couple of players stand out here: Bobby Ryan had a very strong season in the Sim and the NHL stat-line for Clarke MacArthur is based on just the 4 NHL games he played that season where he failed to register a point. The others are all performing roughly on par with what one might expect from a well balanced set of ratings based on their NHL output.

Craig Anderson played all 82 games in the test season, registering a 41-33-6 record with a 0.917 save percentage. Prorating his NHL season to a full 82 game schedule (as with the skaters above) he would have had a 42-31-7 record with a 0.916 save percentage, not far off the Sim numbers.

In the final illustration we've then exemplified team level realism based on the very similar records for the team with Anderson in net and individual productivity realism with how well the players match their NHL based production across the board.
Image
senshockey
The Accomplished One / L'Accompli
Posts: 71
Joined: Thu Aug 06, 2015 9:38 pm

Re: Dedicated rating for the 2.1 engine

Post by senshockey »

Thanks for elaborating. I do appreciate it. Especially when you're the highest priced ratings package I've ever seen.

Would love to see a complete output from a season before committing more money to another ratings provider who may let me down again.
JiriHrdina
New in Town / Le Ptit Nouveau
Posts: 12
Joined: Sun Sep 20, 2009 3:39 pm

Re: Dedicated rating for the 2.1 engine

Post by JiriHrdina »

I'm interested in this as well but the timeline is problematic for the same reasons someone else mentioned. Can we get a more specific idea on when the ratings package will be released. And is there any way to pull that timeline in? I have to think that it is costing you some orders as most leagues need ratings in July at the latest.
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Dedicated rating for the 2.1 engine

Post by 36Henry »

JiriHrdina wrote: Tue Jun 20, 2017 12:38 pm I'm interested in this as well but the timeline is problematic for the same reasons someone else mentioned. Can we get a more specific idea on when the ratings package will be released. And is there any way to pull that timeline in? I have to think that it is costing you some orders as most leagues need ratings in July at the latest.
I hear you, and completely understand the concern. Spending any more time on it than I already do would be very difficult to square with real world matters that also (sadly) require some attention. The target for each year is August 1st. Some years they're done a few days earlier, some years it's been a couple of days later. I'll do what I can to get them done as soon as possible obviously and the hope is that this year I should be done before August 1st, but I can't make any guarantees (unexpected things happen from time to time after all).

Like I told senshockey last night, by July I'll be well into the testing phase with the ratings and would have no problem supplying customers with an early draft of the ratings at that stage. Since testing will still happen continually beyond that point, obviously those ratings would be subject to minor changes and tweaks as testing goes on, but they should still provide your leagues with a very good idea of what each player will look like in the finished product.

Simply put, if the player exceeds real life performance in testing, his ratings are likely to be turned down (from the early draft) and vice versa until everything meets the standards I expect from my product.

Hope that can be of some help to you :)
Image
SkyTeamAces
New in Town / Le Ptit Nouveau
Posts: 4
Joined: Thu Jun 22, 2017 6:19 pm

Re: Dedicated rating for the 2.1 engine

Post by SkyTeamAces »

If I do decide to re-build a new league, I'd be happy to pay for the Deluxe as I want as realistic ratings as possible, and I wasn't envying having to put in the work over the next month to start players from scratch (since most ratings packs are not that good, to be honest).

It's probably going to be a pay league, with sim-based rerates. Old school approach. So it would likely be a one-time purchase for me. Can't understand how no one does sim-based anymore.
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Dedicated rating for the 2.1 engine

Post by 36Henry »

In a few days the 2017/18 version of these ratings will be finished and available for purchase. The final tweaks are being made and the quality of the results is the same as in previous seasons. With practically all other released ratings being optimized for the 1.5 simulation engine, these remain the very best option for those who prefer to run the far more realistic 2.1 engine with all the extra features it brings to the table.

Before giving a few sneak peaks let me just discuss briefly the basis of the ratings. The general idea is to base the ratings on the two most recent NHL seasons, weighted towards the most recent one. In reality this means the ratings are based on (2017/18 x 2 + 2016/17 x 1). This is true for most categories.

However, a few categories are treated differently. Experience and Leadership are obviously based on career numbers, including all levels of play and not just the NHL. For example; Darcy Kuemper being named CHL Goaltender of the Year in 2011 will naturally benefit him in these categories.

Other categories like penalty shots are also based on career numbers due to the fact that in any given season players will only take a low number of shots, making the sample size too small to do anything useful with. However, unlike EX and LD the shootout stat is based only on NHL level shots.

The same goes for DU as the complete injury history at the NHL level for every player is taken into account (both number of injuries and number of games missed).

Fighting is another category where sadly there just isn't enough data to make sensible ratings from anymore due to the lack of fighting in hockey these days. So the FG-rating is based on career numbers on all levels of play.

With that out of the way, here's a quick preview of the Ottawa Senators. The testing procedure consists of me calculating a performance target for each player based on their NHL performance per 20 minutes of ice-time. These numbers are then weighted against the ice-time the player receives in the Simulation where proper NHL lines are used for each team. Stats in the black field represent the NHL Performance Targets, while the red field shows the results of the Simulation. The Simulation stats are an average performance based on five test seasons running the NHL 2017/18 schedule.

Image

And here we see the Vancouver Canucks.

Image

To end this post I just wanted to bring up something I had noticed over the past couple of weeks. Something I like to call "The curious case of Frans Nielsen".

Nielsen is a veteran of 764 NHL games. Generally considered a reliable two-way center with good defensive skills, Nielsen is actually also one of the best shootout performers in the NHL. Over his career he has participated in 99 games that have gone to a shootout. In 96 of them he has been selected to shoot! 62 times has he been selected to shoot first for his team, putting him behind only Jonathan Toews (80) and Zach Parise (66) among active NHLers. On top of that, he has been one of his teams three initial shooters in 93 of the 99 shootout games he's been a part of. He also leads all active NHLers with 22 gamewinning goals in the shootout (two more than T.J. Oshie).

Despite these phenomenal achivements, Nielsen is considered pedestrian at best among some of the free ratings available. I've seen him ranked tied for 211th (behind, among others, defenseman Paul LaDue who has never taken a shootout attempt in his life at either the NHL or AHL level) in one set and tied for 223rd in another. Both rankings making him the equivalent of a player selected to shoot in the seventh round.

You'd think after 700+ games in the NHL he'd deserve a bit more respect than that.
Image
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Dedicated rating for the 2.1 engine

Post by 36Henry »

The 2018/19 ratings are hereby completed and available for purchase. The poster boy for the ratings this year is one Taylor Hall.

Hall was named MVP of the NHL after a season in which he completely dominated the league. Prorated to a full 82 game campaign, Hall scored on pace for 42 goals and 100 points, not to mention leading his team in scoring by 42 points.

That dominance carries over into these ratings. Over five test seasons using the final version of these ratings, Hall averaged 42 goals and 54 assists for 96 points, leading his team in scoring by 36 points, by far the largest margin in the game.

As described in the post above, when evaluating my ratings I use performance targets for all the players. In Hall's case those performance targets call for him to score 34 goals and 52 assists. While his goals are slightly above those targets I feel he's close enough to let it slide. And the rest of the team falls well into line with their performance targets as well.

Image
Image
Blais
New in Town / Le Ptit Nouveau
Posts: 13
Joined: Tue Jul 14, 2015 10:39 pm

Re: Dedicated rating for the 2.1 engine

Post by Blais »

Can i see the results of à league that uses your rating please?

Ifs the only thing left To see before we buy your ratings.

You can PM me with 1 or more league website !

Thx
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Dedicated rating for the 2.1 engine

Post by 36Henry »

PM sent :)
Image
Blais
New in Town / Le Ptit Nouveau
Posts: 13
Joined: Tue Jul 14, 2015 10:39 pm

Re: Dedicated rating for the 2.1 engine

Post by Blais »

Thx for thé fast answer :)
Savard
The Accomplished One / L'Accompli
Posts: 119
Joined: Fri May 24, 2019 1:07 am

Re: Dedicated rating for the 2.1 engine

Post by Savard »

I would also love to see the standings and some team stats for that league, please. Thanks! Also: will there be new ratings this Summer?
Post Reply