Initial testing

This temporary section in English only contains idea / feedback / suggestion / problem related to the simulation engine beta.
Post Reply
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Initial testing

Post by 36Henry »

Since there seems to be so little interest in running tests with the new engine I figured I might as well give it the old go.

In these tests the ratings from my league have been used. As far as I know these ratings aren't used anywhere else so they are exclusive to our league. The tool used is the Automatic Test command on a 30 team league where each team plays 330 games per season. The target for the tests is to get all of the different statistical outcomes as close to the numbers listed for the NHL as possible, primarily by adjusting the sliders in the Sim and then by the Mass Edition option for further tweaks that go beyond what the sliders are capable of doing with the ratings as they are in their original state.

So lets proceed.

The tests began with the following settings: Shots - 50, Goals - 75, Penalties - 55, Hits - 45, Fighting 10

That produced a season where we got the following results: (NHL targets in brackets)

Goals per game: 3.15 (5.68)
Powerplay goals per game: 0.67 (1.58)
Powerplay efficiency: 9.49% (18.95%)
Powerplays per game: 7.11 (8.32)
Shorthanded goals per game: 0.15 (0.19)
Even strength goals per game: 2.33 (3.92)
Shots per game: 58.55 (60.40)
Penalty minutes per game: 22.79 (27.52)
Hits per game: 43.30 (41.92)
Blocked shots per game: 19.39 (26.28)
Fights per game: 0.37 (0.53) * the test rosters had very few fighters on them which skews this stat
Average Save Percentage: 0.946 (0.906)

Given these initial stats it became clear that scoring was way too low, as was the efficiency on the powerplay and shotblocking. The other factors were fairly accurate for a first run.

After a few more seasons had been run with variations of the sliders we ended up on season 4 where the following settings were used: Shots - 53, Goals - 99, Penalties - 62, Hits - 46, Fighting 40

These settings produced the following:

Goals per game: 4.28 (5.68)
Powerplay goals per game: 0.84 (1.58)
Powerplay efficiency: 9.98% (18.95%)
Powerplays per game: 8.39 (8.32)
Shorthanded goals per game: 0.24 (0.19)
Even strength goals per game: 3.20 (3.92)
Shots per game: 60.46 (60.40)
Penalty minutes per game: 27.17 (27.52)
Hits per game: 41.89 (41.92)
Blocked shots per game: 20.09 (26.28)
Fights per game: 0.45 (0.53)
Average Save Percentage: 0.929 (0.906)

As you can see there are a lot of categories that are pretty much right where you want them to be. But still the scoring, powerplay and shotblocking was nowhere in place and with the scoring now up to 99 there was nothing more the sliders could do to up the scoring. So we turned to Mass Edition.

Given that the csv-files the Testing tool provides has a percentage of scoring done by defensemen I could tell that defensemen were scoring 12.14% of all goals, compared to 14.6% in the NHL. So the first thing done in Mass Edition was to increase the scoring rating for defensemen by 10%.

That was the only ratings edited in a season run with these settings: Shooting - 53, Goals - 99, Penalties - 63, Hits - 46, Fighting - 45

The results were as follows:

Goals per game: 3.00 (5.68)
Powerplay goals per game: 0.55 (1.58)
Powerplay efficiency: 6.49% (18.95%)
Powerplays per game: 8.51 (8.32)
Shorthanded goals per game: 0.17 (0.19)
Even strength goals per game: 2.28 (3.92)
Shots per game: 61.55 (60.40)
Penalty minutes per game: 27.74 (27.52)
Hits per game: 42.61 (41.92)
Blocked shots per game: 20.46 (26.28)
Fights per game: 0.47 (0.53)
Average Save Percentage: 0.951 (0.906)

Not being a seasoned veteran at this Sim I didn't see that coming. By increasing Scoring for defensemen, goals per game dropped from 4.28 to 3.00 :shock: You can imagine my surprise. To make it even more surprising, the percentage of total goals scored by defensemen actually increased from 12.14% to 13.05%, still far from the 14.6% of the NHL but still remarkable. By increasing the Scoring rating for defensemen, forwards scored a LOT less despite the fact that their ratings had not been touched. Peculiar.

So obviously there was only one thing to do, run a season where Scoring for defensemen was reduced by 10% instead of increased by 10%. Settings used in this season was identical to the one above and produced the following:

Goals per game: 6.24 (5.68)
Powerplay goals per game: 1.28 (1.58)
Powerplay efficiency: 14.68% (18.95%)
Powerplays per game: 8.74 (8.32)
Shorthanded goals per game: 0.34 (0.19)
Even strength goals per game: 4.61 (3.92)
Shots per game: 59.57 (60.40)
Penalty minutes per game: 29.15 (27.52)
Hits per game: 41.50 (41.92)
Blocked shots per game: 19.82 (26.28)
Fights per game: 0.54 (0.53)
Average Save Percentage: 0.895 (0.906)

Goals per game rocket to 6.24 per game and we see the powerplay become a lot more effective as well, things we needed to see but didn't expect to see by reducing scoring. The effect of playing around with Scoring for defensemen is outlined below:

Original rating: 4.28 goals per game given settings of Shooting - 53, Goals - 99, 12.14% of all goals are scored by defensemen.
Defensemen with 10% bonus on Scoring: 3.00 goals per game given above settings, 13.05% of all goals are scored by defensemen.
Defensemen with 10% decrease in Scoring: 6.24 goals per game given above settings, 12.06% of all goals are scored by defensemen.
Defensemen with 5% decrease in Scoring: 5.15 goals per game given above settings, 12.08% of all goals are scored by defensemen.

The next step was to see if reducing Scoring for forwards would have the same effect so in the next season Scoring was reduced by 5% for both defensemen and forwards. Settings used for the season were: Shooting - 53, Goals - 99, Penalties - 62, Hits - 46, Fighting - 42

Goals per game: 7.45 (5.68)
Powerplay goals per game: 1.58 (1.58)
Powerplay efficiency: 17.37% (18.95%)
Powerplays per game: 9.07 (8.32)
Shorthanded goals per game: 0.43 (0.19)
Even strength goals per game: 5.44 (3.92)
Shots per game: 58.89 (60.40)
Penalty minutes per game: 29.24 (27.52)
Hits per game: 41.48 (41.92)
Blocked shots per game: 19.53 (26.28)
Fights per game: 0.49 (0.53)
Average Save Percentage: 0.874 (0.906)

By now it's pretty obvious that to increase scoring in the league, the Scoring ratings should be lowered. Quite surprising for me personally, but maybe that's just me.

With this season we also suddenly find ourselves having a powerplay that is exactly as effective as it should be based on the NHL standard of 1.58 powerplay goals per game. And the rate of efficiency with the man advantage reaches an all time high for these tests with a success rate of 17.37%.

There are downsides though as the goalies, who had been superhuman until now, suddenly are mediocre when the skaters they face are reduced for Scoring.

Another major problem is the offensive efficiency of the penalty killers. In the NHL they score 0.19 shorthanded goals per game. In this latest test they score 0.43, which is 126% higher than in the NHL. I believe this might be an area where the new engine needs to be tweaked as this number is consistently too high.

The next step was then to lower the slider for Scoring from the value of 99 used up until now. Over the course of the next three seasons the following settings were used: Shooting - 53 53 53, Goals - 90 80 72, Penalties - 61 61 61, Hits - 47 47 47, Fighting - 43 43 43

During these three seasons, result shifted as follows:

Goals per game: 6.69, 6.07, 5,52 (5.68)
Powerplay goals per game: 1.45, 1.47, 1.39 (1.58)
Powerplay efficiency: 16.51%, 16.55%, 15.68% (18.95%)
Powerplays per game: 8.80, 8.87, 8.85 (8.32)
Shorthanded goals per game: 0.36, 0.33, 0.31 (0.19)
Even strength goals per game: 4.88, 4.27, 3.83 (3.92)
Shots per game: 58.85, 58.96, 58.87 (60.40)
Penalty minutes per game: 29.60, 31.07, 32.07 (27.52)
Hits per game: 42.24, 42.33, 42.33 (41.92)
Blocked shots per game: 19.61, 19.67, 19.70 (26.28)
Fights per game: 0.57, 0.66, 0.74 (0.53)
Average Save Percentage: 0.886, 0.897, 0.906 (0.906)

Starting with the positives we now have a fairly accurate amount of goals being scored and the powerplay has come to life while still keeping a good amount of shots on goal. We've also managed to turn the goaltenders into just the right amount of awesome that we were looking for.

Now if lowering the Scoring rating had a surprising effect in that it actually raised the amount of goals being scored, it's even more strange to notice another effect here. When we lower the slider for Goals and keep everything else constant, we see fighting increase from 0.49 fights per game to 0.74. Remember that this increase in fighting comes without anything being done to either the Fighting or Discipline rating or from a change to any other slider but scoring being lowered from 99 to 72. Puzzling.

With this test done I tweaked the sliders to run for a season with the following: Shooting - 54, Goals - 73, Penalties - 61, Hits - 46, Fighting - 40

Goals per game: 5.71 (5.68)
Powerplay goals per game: 1.42 (1.58)
Powerplay efficiency: 16.09% (18.95%)
Powerplays per game: 8.81 (8.32)
Shorthanded goals per game: 0.31 (0.19)
Even strength goals per game: 3.98 (3.92)
Shots per game: 59.85 (60.40)
Penalty minutes per game: 31.06 (27.52)
Hits per game: 41.90 (41.92)
Blocked shots per game: 19.85 (26.28)
Fights per game: 0.68 (0.53)
Average Save Percentage: 0.905 (0.906)

For now this is where it stands. More testing will be done tomorrow.

It would be interesting to hear from Simon and jwr on a few of the things above. Especially interested in hearing if in your tests the PK is as offensively effective as it has been during these tests and if you have ideas on how to increase shotblocking which is on my list to do for tomorrow. I'm guessing the DF rating plays a part in that but will play around with different things to get it right.

I'd also like to hear from you on the matter of scoring actually going up by lowering the SC-ratings and how lowering the slider for Goals leads to an increase in fighting. Is this something you've seen in your tests as well? Now granted I haven't run multiple seasons to test these findings yet as I want to find the proper range of settings and ratings before I do so, but the effects on scoring and fighting that these changes have had is so dramatic that it can't be down to randomness of having too small of a sample size I believe.

Anyway, hell of a long thread of messing about from a complete tosser so it's time to turn it off.

Alright?
Image
Kramden23
The Addict / Le Drogué
Posts: 763
Joined: Tue May 13, 2008 8:33 pm

Re: Initial testing

Post by Kramden23 »

Wow, great information! I really wish I had the time to contribute, but I'm sure the community (I know I do) appreciates all your work!
Less than 1% of americans can speak french
More than 70% of canadians can speak english

Want more answers, why limit yourself? Post in english!
QSHL
The Crazy / Le Fou
Posts: 307
Joined: Mon Jun 02, 2008 5:31 pm

Re: Initial testing

Post by QSHL »

I wish I had time to test like you, but I read your text and thought it was pretty interesting.


Increasing D's SC means the defenseman will try to shoot more. Which is probably why it was a bad thing for scoring. More shots from farther away. Your Ds will prefer to shoot than to pass.

When you reduce everybody's SC, it means less chances players shoot from anywhere and pass more for a better opportunity. This, knowing more now how Simon codes his stuff, still makes sense in a way.


What puzzles me though is the high increase in SH goals and, especially, fighting. I've struggled badly to find a correct formula for fighting. I've mass-edited most of my league to get all my non-fighters under 30 FG. But guys who remain above that number still fight a lot. I cannot explain / guess why on Earth would fighting go up when you reduce the SC of players...



Thanks and keep us updated!
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Initial testing

Post by 36Henry »

I do have a theory about why shorthanded goals are so high and I've discussed it briefly with Simon regarding the 2.0 engine a short time ago. But with these tests I haven't gotten around to checking if that theory still holds true in 2.1.

Since the last update I've run 26 more seasons with various little tweaks to the sliders and by using the Mass Edition tool. At the end of all that I found a good balance and below are the average results of the last 3 seasons I ran. All of these three seasons were run with the following settings: Shots - 47, Goals - 62, Penalties - 63, Hits - 48, Fighting 42

Goals per game: 5.72 (5.68)
Powerplay goals per game: 1.57 (1.58)
Powerplay efficiency: 18.98% (18.95%)
Powerplays per game: 8.29 (8.32)
Shorthanded goals per game: 0.28 (0.19)
Even strength goals per game: 3.86 (3.92)
Shots per game: 60.38 (60.40)
Penalty minutes per game: 27.81 (27.52)
Hits per game: 42.15 (41.92)
Blocked shots per game: 20.30 (26.28)
Fights per game: 0.52 (0.53)
Average Save Percentage: 0.905 (0.906)

So like before the only real problems are shotblocking and shorthanded goals as I appear to be unable to tweak these without ruining other areas.

With that first phase finally done and having settings and ratings in order, it's time to look at other things than just these average stats. One of the things that has annoyed me (and many others I believe) about the 2.0 engine is the seemingly complete randomness of the whole thing. Simply having the best team didn't mean you couldn't finish 60 points behind a team with hardly a single player good enough to play on your team. So one key thing for me about the 2.1 version is: is this just as random as the previous version?

These early tests seem to indicate that it is not. Having run over 30 seasons with the Automatic Test tool, I find that there was very little variation in which teams made the playoffs. Usually teams would finish in the exact same spot year after year or maybe one position better or worse. Very rarely were there bigger fluctuations than that. And for the most part, the good teams finished consistently ahead of the bad teams. Which is a good thing.

However, it appears goaltending is the thing to have as the best goaltender in the league, playing for a pretty mediocre team, dominated the league every season. He was so good in fact that his fairly poor team finished top 3 in the standings with regularity. In one of my tests I then traded him off that club and sent him to the worst team in the Conference, replacing him with that clubs starting goalie. The results of this change were pretty dramatic as the star goalie now carried his new team, a perennial bottom feeder that always finished between 13th and 15th in the Conference, to the #4 seed on the heels of a 442 point season. Their average season without this new goalie was 243 points per season. So he improved them nearly 200 points. And his old club? How did they manage without him? They still made the playoffs with a 342 point season, claiming the #7 seed in the conference. However that was a drop of over 150 points from the average point totals they used to claim with the star goalie in net.

All of these tests were done with the Automatic Test tool. The next step now is to actually Sim a couple of seasons and see if these trends carry on in real play.
Image
SimonT
STHS Owner / Propriétaire du STHS
Posts: 14757
Joined: Sat Oct 08, 2005 4:18 pm
Location: Montreal, Canada
Contact:

Re: Initial testing

Post by SimonT »

I love the test for defenseman! It’s make sense totally for me! As you can see from here (http://sths.simont.info/Forum/viewtopic ... ing#p30633) and also in QSHL answer, player’s makes decision based on their skill. When you lower SC for DF, they make more passing decision to forward and forward get their shot block less often that defenseman.

The way you make your test is different from me. I start by find the correct number for fights, penalties, hits then I find the correct number for shots THEN I find the correct number for goals. I never look at goal stats if my scoring stats are not good.

For the more fights when lower goals, I think you are just unlucky. The fight formula is the formula that I’m having a lot of problem getting consistent number. I saw a 50% difference in the # of fight between the same two simulations but I often see a 25% difference.

V2.1 is less random that V2.0. If my memory is correct, my test show randomness had drop by 33%.

One of complain in V2.0 was goalie randomness and again, if my memory is correct, my test show I drop it by 50%! This makes good goalie very good and bad goalies very bad. But again, in a real life scenario, your first goalie will not play all game like they do in automatic simulation because of fatigue.
36Henry wrote:With that first phase finally done and having settings and ratings in order, it's time to look at other things than just these average stats. One of the things that has annoyed me (and many others I believe) about the 2.0 engine is the seemingly complete randomness of the whole thing. Simply having the best team didn't mean you couldn't finish 60 points behind a team with hardly a single player good enough to play on your team. So one key thing for me about the 2.1 version is: is this just as random as the previous version?

These early tests seem to indicate that it is not. Having run over 30 seasons with the Automatic Test tool, I find that there was very little variation in which teams made the playoffs. Usually teams would finish in the exact same spot year after year or maybe one position better or worse. Very rarely were there bigger fluctuations than that. And for the most part, the good teams finished consistently ahead of the bad teams. Which is a good thing.
I'm VERY VERY HAPPY that you see that!!!! You just make my day! :D :mrgreen: 8)
-SimonT
Forum Administrator / Administrateur du Forum
STHS Owner / Propriètaire du STHS
English V2 & V3 Manual - Manuel V2 & V3 Français
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Initial testing

Post by 36Henry »

I'm glad I made your day Simon :lol:

And it's interesting to read that you've done such a big adjustment on the goalies because it's really apparent in my seasons that unless you have a great goalie you can't compete. In fact, I would say that possibly the improvement on goalies has been a little too strong.

Having just completed five seasons by Simming the normal way and not using the Automatic Test tool, that goalie I mentioned before has led his club to four President's Trophies and three Championships. On what can only be described as a comparatively weak club on every position except goaltending.

Allow me to get technical and positively nerdy.

The average rating of the 30 starting goalies are: 87, 68, 90, 81, 90, 89, 89, 91, 91, 89, 88, 77, 83, 84

The near invincible goalie is rated as follows: 89, 77, 99, 82, 95, 94, 95, 98, 97, 96, 94, 99, 99, 91 and suits up for Team 14

So time for some numbers. Here are the total standings of the five seasons:

Image

As we see, Team 14 is far and away the best team in the league. And to support my claims that this is mostly due to their goaltender I split up goalies, defensemen and forwards and tallied up the ratings for each group on each team. Then ranked them from 1-30. Perhaps not a scientific method but oh well.

Team 14 obviously rank #1 for goaltending in this comparison. Their defensemen on the other hand rank 26th while their forwards rank 18th.

The top 12 teams in the standings had the top 12 goaltenders. Among the 9 worst goalies in the league only one of them managed to guide their team above the bottom 9 positions.

So the standings after 24 750 games is more or less nothing more than a ranking of the goaltenders. While people have complained a lot about the randomness of 2.0, I believe that if these numbers are somewhat accurate there will be a lot of complaints over how unless you have the best goalie you can't win. So is it possible that your tweaking of the goaltenders has been a little too strong?
Image
SimonT
STHS Owner / Propriétaire du STHS
Posts: 14757
Joined: Sat Oct 08, 2005 4:18 pm
Location: Montreal, Canada
Contact:

Re: Initial testing

Post by SimonT »

1) Are you playing with goalie fatigue?
2) You have the stat for goalie?
-SimonT
Forum Administrator / Administrateur du Forum
STHS Owner / Propriètaire du STHS
English V2 & V3 Manual - Manuel V2 & V3 Français
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Initial testing

Post by 36Henry »

1. Not playing with goalie fatigue as I only have the free version and normally when I run with fatigue active I keep getting error messages on how lines are incomplete. And with 24.000+ games to run it gets quite tedious to edit the lines constantly :lol:

2. I do. Will provide a file for you in a PM shortly.
Image
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Initial testing

Post by 36Henry »

Right so who scores the goals? Well it's not really the players we'd expect to do the majority of the damage. Let's start with the forwards and begin by laying out the ratings of these fellas.

Highest SC, Average SC, Lowest SC = 92, 71, 36
Highest PA, Average PA, Lowest PA = 92, 68, 39

Combining the two gives some kind of Offensive ability as follows:

Highest OFF, Average OFF, Lowest OFF = 182, 139, 80

Putting all the forwards that played in these five seasons into a long line and ranking their offensive ability is easy. You can then calculate their effectiveness by computing a value for Points Scored per 20 minutes of ice-time and do a similar ranking based on this value. Then it's easy to compare the two rankings to see if offensive ability leads to better efficiency.

So does it? In a word, no. The top 11 players for Offensive ability (three-way tie for 9th so it's top 11 instead of a traditional top 10), all failed to be among the top 30 most effective offensive forwards in the league. And only one of them cracked the top 40, claiming 40th by a hair. Here's how all the top 11 finished. Remember that only 360 forwards are part of this study.

The #1 rated offensive player finished 109th for offensive effectiveness.
#2 finished 47th.
#3 finished 122nd.
#4 finished 97th.
#t.5 finished 140th.
#t.5 finished 40th.
#7 finished 79th.
#8 finished 134th.
#t.9 finished 149th.
#t.9 finished 58th.
#t.9 finished 111th.

Not exactly a dominant performance from the big boys. So who finished top 10 in offensive effectiveness then?

#71 finished 1st.
#107 finished 2nd.
#187 finished 3rd.
#61 finished 4th.
#67 finished 5th.
#28 finished 6th.
#99 finished 7th.
#155 finished 8th.
#20 finished 9th.
#170 finished 10th.

The 71st best rated offensive players dominates the league with ratings of 77 SC and 81 PA. Even more remarkable is the third place finisher who manages to be the third most effective player in the league despite ratings of just 79 SC and 63 PA. His PA rating in other words is 5 points below the league average, yet he's scoring at this incredible pace.

One way of possibly explaining the unlikely success of player 187 is the fact that he plays for Team 17, the club that finished 29th overall. What I've found is that average players on bad teams will outdo themselves if they are the main offensive weapon on the club. Meanwhile you can have a team loaded with offensively excellent players on a strong club and they will all perform below expectations despite a glorious supporting cast. It's as if having more than one good player will do more bad than good towards creating offense (much in the same way that lowering SC for forwards lead to increased scoring league-wide).

To illustrate the above, take Team 8 as a case study. The top 6 forwards on Team 8 rank 3rd, 4th, 9th, t.12th, t. 12th and 31st in the league in terms of Offensive ability (PA+SC). That is the equivalent of picking 6 of the top 30 scorers in the NHL and putting them on the same team. Think that team would be able to score some goals? Exactly, they can't. Here's how they placed in terms of productivity:

#3 placed 122nd
#4 placed 97th
#9 placed 111th
#t.12 placed 70th
#t.12th placed 137th
#31 placed 157th

One would expect players this talented, playing with players this talented, would be able to match any team in the league in terms of offensive productivity. Not so. Not by a long shot.

Now lets compare them to Team 9 whose numbers are as follows for their top 6 forwards:

#107 finished 31st
#113 finished 67th
#164 finished 45th
#170 finished 10th
#223 finished 64th
#241 finished 207th

So a team with 6 of the 31 most talented forwards in the league had their most productive player rank three spots behind the fifth most effective player on a team that didn't have even one of the 100 most talented players? What is wrong with this picture?

I find that we still have issues that bothered me in 2.0 when it comes to having the best players be the best players. Now I wouldn't mind at all if a player would have ups and downs over the course of five seasons as that is nothing but natural, but that's not what is happening here. Year in and year out the stats are pretty much the same for these players. Poor players will outscore good players on a regular basis which completely screws up the balance of a league where GMs have spent years building competitive clubs that end up having no chance of competing because something in the Sim wont allow them to play to their ability while poor teams overachieve by default.

I'd like to know if this is indeed related to the matter of how lowering the SC rating of forwards will increase scoring. If that is true, then there has to be some information to those who build ratings as to what level of SC should be used for the top goal scorers. Year in and year out in the NHL there are a handful of players competing for the Rocket Richard, not almost 200 players as was the case here. The most effective goalscorer was the illustrious player #187 whose ratings of 79 SC and 63 PA made him rank first in the league for Goals/20 minutes of ice-time. The average rating of the top 10 goalscorers was 80.6 so perhaps instead of going up in the 90's when making ratings we should just keep the top guys around the 80 mark? Either way, some useful feedback from the man would be very interesting here to help me understand just why relatively bad players can dominate like this.
Image
SimonT
STHS Owner / Propriétaire du STHS
Posts: 14757
Joined: Sat Oct 08, 2005 4:18 pm
Location: Montreal, Canada
Contact:

Re: Initial testing

Post by SimonT »

Hi.

1) The true is that you look at 2 stats only for your offensive effectiveness. Technically, you add at least the puck control and skating skill.
2) If you read this topic http://sths.simont.info/Forum/viewtopic ... ing#p30633, you’ll understand that it’s not a matter a having the best offensive effectiveness that make a players dominant.
3) Goalie still stops 90% of the shot.
4) In the NHL, there are 583 forwards. In the top 10, you have 17 goals of differences between the first one (Corry Perry at 50 goals) and the 10th (Eric Stall at 33 goals).I don’t think that happen in the STHS (You can confirm). Why? Let do some math. How much good is a player with 79 of SC versus a player of 99 of SC? Mathematically speaking, it’s only 21%. So if your best player hit 50 goals, mathematically, a player who only 21% less good in scoring should score 41 goals. Then add some random and byebye the math thinking.

My conclusion: In the NHL, goal scoring over a huge sample do a logarithm function (For a sample of 583 players and on a scale from 0 to 50 players, only 25 players have 25 or more goals) while in the STHS, the formula is more linear. I think that why you guy complain a lot on that. This will give me food for thought.
-SimonT
Forum Administrator / Administrateur du Forum
STHS Owner / Propriètaire du STHS
English V2 & V3 Manual - Manuel V2 & V3 Français
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Initial testing

Post by 36Henry »

As always I really appreciate the feedback.

In the seasons I ran the difference in goals between #1 and #10 (pro-rated to a single 82 game season) was 13 goals. Not the 17 of the NHL, but not far from it. However a more adequate way of measuring imo would be goals per game since a lot of NHLers have injuries during a year while in my seasons injuries and suspensions were turned off. Converting the top goalscorers in the NHL to this ratio one finds that Crosby, if healthy all year, would outscore Perry by 13 goals on his own. Take him out of the equation and the difference between Perry and Grabner at #11 is no more than 13 goals, exactly the same as in my seasons. With Crosby as an outstanding anomaly that I agree would be impossible to re-create in the Sim I would say the Sim produces realistic levels as far as the distribution of goals among the top scorers. It's just that the top scorers aren't the usual suspects. In the NHL you generally see the same names pop up year after year on the top 30 goalscorers list. The difficulties in the Sim that I believe is the main criticism is that it's near impossible to re-create these top players to be the top players in the Sim. However I think a large reason why may be the ratings used as I'll expand on below.

Thanks for pointing me to that post as it certainly contained valuable information I didn't know about. Can you confirm that what that post says about the SK-rating means that if you want to create a player that is more prone to shooting and/or passing, you should make his SK low in comparison to PA and SC? I think if that is the case that could be a main reason why seemingly "less talented" players are more effective in these tests as in our ratings the SK is generally higher for players considered "more talented", same with PH if that has a similar effect on lowering a players tendency to shoot or pass the puck.
Image
SimonT
STHS Owner / Propriétaire du STHS
Posts: 14757
Joined: Sat Oct 08, 2005 4:18 pm
Location: Montreal, Canada
Contact:

Re: Initial testing

Post by SimonT »

36Henry wrote:Can you confirm that what that post says about the SK-rating means that if you want to create a player that is more prone to shooting and/or passing, you should make his SK low in comparison to PA and SC?


Confirm!
36Henry wrote:Same with PH if that has a similar effect on lowering a players tendency to shoot or pass the puck.
Not True. The PH skill is simply use in fews offensive formula.
-SimonT
Forum Administrator / Administrateur du Forum
STHS Owner / Propriètaire du STHS
English V2 & V3 Manual - Manuel V2 & V3 Français
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Initial testing

Post by 36Henry »

Thanks.

And SK has absolutely nothing to do with the speed of the player then? It's just a players decision to basically not pass or shoot the puck but instead skate with the puck?
Image
SimonT
STHS Owner / Propriétaire du STHS
Posts: 14757
Joined: Sat Oct 08, 2005 4:18 pm
Location: Montreal, Canada
Contact:

Re: Initial testing

Post by SimonT »

No. The "speed" factor is used to know if a player will get hit or not. (Faster player get less hit)
-SimonT
Forum Administrator / Administrateur du Forum
STHS Owner / Propriètaire du STHS
English V2 & V3 Manual - Manuel V2 & V3 Français
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Initial testing

Post by 36Henry »

OK.

Could you run through all the ratings and explain what they all do? Because clearly the ratings we've been using haven't been created with regards to how the Sim actually works, which probably explains why we haven't been able to get "realistic" performances from the players. So if you could sort me out with the information I'd need to create proper ratings, I can run some more tests using proper players (based on the NHL 10/11 season).

Would be greatly appreciated.
Image
Post Reply