WeThreeKings
Habs cup - its in the BAG
I appreciate the work and the thought but I had to drop every online league with one because of the way the PA gets distributed to player archetypes.
Eastside Hockey Manager - Part VIII
- Thread starter hangman005
- Start date
- Status
I appreciate the work and the thought but I had to drop every online league with one because of the way the PA gets distributed to player archetypes.
Savi
Registered User
I hate it when my team only shows up for one period 
WreckingCrew
Registered User
- Feb 4, 2015
- 12,325
- 37,961
Hi all,
Lots of questions about the prospect generator. I'm easier found on Discord, but I got paged here by ECK.
TL;DR:
1. History of PG and apology for the upsets it has caused
2. Aim of the PG
3. Recent Changes to PG to address concerns
4. How I'm moving forward with PG to get it to a final place I'm happy with.
5. Last notes on what the PG doesn't (and can't) do.
6. PA Distribution that the PG does
1. Firstly, I apologise to those who've found the early versions of the prospect generator has not helped their experience. Especially the leagues like FDL, INHL, 5WHL (the original incarnation) etc. who used these. Unfortunately, development of stuff at this scale takes multiple iterations and every lesson that is learnt goes back into the development. Even worse, it often takes 7-10 seasons to see the results, which can be 18-24 months of online league time. As a result, The Prospect Generator is now at v11.2. The aforementioned league were all using pre-v1. The current TBL 12.4.3 rosters use v11.0. NTHL uses v 10.3, XHL uses v11.1. Overall, there's been almost 100 releases of the prospect generator since I first did it in late 2017. Main structural changes shift the header number, smaller tweaks the sequence numbers.
2. The whole aim of the prospect generator is to try and make sure that beyond the existing players, you get a realistic talent distribution in the game and it looks as close to the TBL rosters as possible. Originally, I had a lot of feedback that people wanted to see "sons" of real life players, because quite often that's what happens in real life. I wrote special formulae that copied the attributes and characteristics of the fathers to their sons to try and make them seem similar (Fake Crosby plays similar to Sidney Crosby etc.). The switch from v11.1 to v11.2 removes that (v11.0 to v11.1 drastically reduced the PA's they were generating at, which is important). The fakes retain the same player role as the real players, as well as some non-essential stats like aggression, dirtiness etc. but almost every visible attribute is now with the new generation formulae (one of the main changes from v10.3 to v11.0) that controls every other prospect generated.
3. The current generating formulae are almost there. I'm 95% happy with it. The bits that still need work are I think playmaking forwards need a little increase to their shooting stats when they get to the NHL level. I think smaller players need tweaking to reduce their balance and strength at the NHL level, and perhaps increase their agility too. Some work needs done on player reputations as well, because reputation is not rising fast enough and you see guys with 150CA at age 22 sign ELC extensions at crazy things like 2.5m x 5y. But that's not just a generator problems that affects all players.
4. In a few weeks (timescale dependent), I'm going to open up a wider fun sim sign-up game come beta test thing focussed on nailing down any final generation issues before a move to the future v12 and (I hope) final release before v1.5 comes out and changes everything about prospect development. I have big plans with ECK about incorporating parts of the prospect generator into his ECK databases but before I do anything, I want everything perfect. It's a huge level of responsibility and it needs that level of review.
5. What the prospect generator can't do (so please don't judge me on it).
- I cannot make NHL-ready players out of the draft without completely screwing over the junior levels. I did that before and got pretty heavy abuse for it by parts of the community. EHM just does not do CA acceleration like what happens nowadays. I expect v1.5 might change this a little, but we'll see.
- I cannot currently stop the game from generating other top-end prospects and the eventual talent escalation that comes with it. With that said, even though CA goes up in the later saves 5-10 seasons in, what you see doesn't necessarily change because a guy who grows from 50CA to 150CA will look like a 130CA guy in the TBL Databases. I include for this when I do the PA distributions.
- Once a game is started, there isn't a cure-all for the player archetype issues that I'm aware of. All I can do is apologise and try to make things better. I do try, but I'm just one guy with an hour or so a week - feedback always helps.
6.
Assuming a 10-year average NHL career for a 2nd liner and up, and an average 5-year career for a 3rd liner and down (In a 32-team NHL to allow for not all players make their PA):
= 32*10 = 320 top-end players every 10 years = 32 players per draft average (Starter G, top 3 D, top 6 F)
= 32*13 = 416 bottom-end players every 5 years = 83.2 players per draft average (Backup G, 4-6 D, 7-12 F)
Total = 115.2 players per year
End up with a distribution like this:
[TBODY]
[/TBODY]So it means generating the following per year:
Lots of questions about the prospect generator. I'm easier found on Discord, but I got paged here by ECK.
TL;DR:
1. History of PG and apology for the upsets it has caused
2. Aim of the PG
3. Recent Changes to PG to address concerns
4. How I'm moving forward with PG to get it to a final place I'm happy with.
5. Last notes on what the PG doesn't (and can't) do.
6. PA Distribution that the PG does
1. Firstly, I apologise to those who've found the early versions of the prospect generator has not helped their experience. Especially the leagues like FDL, INHL, 5WHL (the original incarnation) etc. who used these. Unfortunately, development of stuff at this scale takes multiple iterations and every lesson that is learnt goes back into the development. Even worse, it often takes 7-10 seasons to see the results, which can be 18-24 months of online league time. As a result, The Prospect Generator is now at v11.2. The aforementioned league were all using pre-v1. The current TBL 12.4.3 rosters use v11.0. NTHL uses v 10.3, XHL uses v11.1. Overall, there's been almost 100 releases of the prospect generator since I first did it in late 2017. Main structural changes shift the header number, smaller tweaks the sequence numbers.
2. The whole aim of the prospect generator is to try and make sure that beyond the existing players, you get a realistic talent distribution in the game and it looks as close to the TBL rosters as possible. Originally, I had a lot of feedback that people wanted to see "sons" of real life players, because quite often that's what happens in real life. I wrote special formulae that copied the attributes and characteristics of the fathers to their sons to try and make them seem similar (Fake Crosby plays similar to Sidney Crosby etc.). The switch from v11.1 to v11.2 removes that (v11.0 to v11.1 drastically reduced the PA's they were generating at, which is important). The fakes retain the same player role as the real players, as well as some non-essential stats like aggression, dirtiness etc. but almost every visible attribute is now with the new generation formulae (one of the main changes from v10.3 to v11.0) that controls every other prospect generated.
3. The current generating formulae are almost there. I'm 95% happy with it. The bits that still need work are I think playmaking forwards need a little increase to their shooting stats when they get to the NHL level. I think smaller players need tweaking to reduce their balance and strength at the NHL level, and perhaps increase their agility too. Some work needs done on player reputations as well, because reputation is not rising fast enough and you see guys with 150CA at age 22 sign ELC extensions at crazy things like 2.5m x 5y. But that's not just a generator problems that affects all players.
4. In a few weeks (timescale dependent), I'm going to open up a wider fun sim sign-up game come beta test thing focussed on nailing down any final generation issues before a move to the future v12 and (I hope) final release before v1.5 comes out and changes everything about prospect development. I have big plans with ECK about incorporating parts of the prospect generator into his ECK databases but before I do anything, I want everything perfect. It's a huge level of responsibility and it needs that level of review.
5. What the prospect generator can't do (so please don't judge me on it).
- I cannot make NHL-ready players out of the draft without completely screwing over the junior levels. I did that before and got pretty heavy abuse for it by parts of the community. EHM just does not do CA acceleration like what happens nowadays. I expect v1.5 might change this a little, but we'll see.
- I cannot currently stop the game from generating other top-end prospects and the eventual talent escalation that comes with it. With that said, even though CA goes up in the later saves 5-10 seasons in, what you see doesn't necessarily change because a guy who grows from 50CA to 150CA will look like a 130CA guy in the TBL Databases. I include for this when I do the PA distributions.
- Once a game is started, there isn't a cure-all for the player archetype issues that I'm aware of. All I can do is apologise and try to make things better. I do try, but I'm just one guy with an hour or so a week - feedback always helps.
6.
Assuming a 10-year average NHL career for a 2nd liner and up, and an average 5-year career for a 3rd liner and down (In a 32-team NHL to allow for not all players make their PA):
= 32*10 = 320 top-end players every 10 years = 32 players per draft average (Starter G, top 3 D, top 6 F)
= 32*13 = 416 bottom-end players every 5 years = 83.2 players per draft average (Backup G, 4-6 D, 7-12 F)
Total = 115.2 players per year
End up with a distribution like this:
| Draft Requirements | PA Low | PA High | G | LD | RD | LW | RW | C | TOTAL | CUMULATIVE | -13 | -14 | -15 | -7 | -8 | -9 | -10 | PA Count Per Draft |
| Generational | 190 | 200 | 0.05 | 0.025 | 0.025 | 0.04 | 0.04 | 0.1 | 0.3 | 0.3 | 1.2% | 35.5% | 1 x -10 | |||||
| Elite | 180 | 189 | 0.3 | 0.2 | 0.2 | 0.2 | 0.2 | 0.3 | 1.4 | 1.7 | 12.3% | 3.2% | 32.3% | 1 x -10, 8 x -15 | ||||
| Franchise | 170 | 179 | 0.7 | 0.5 | 0.5 | 0.5 | 0.5 | 0.7 | 3.4 | 5.1 | 12.3% | 32.3% | 32.3% | 1 x -10, 8 x -15, 6 x -9 | ||||
| 1st Line | 160 | 169 | 1.4 | 1.2 | 1.2 | 1.2 | 1.2 | 1.7 | 7.9 | 13.0 | 1.4% | 12.3% | 3.2% | 32.3% | 8 x -15, 6 x -9, 20 x -8, 60 x -14 | |||
| 2nd Line | 145 | 159 | 3.5 | 3 | 3 | 3 | 3 | 3.5 | 19.0 | 32.0 | 21.1% | 18.5% | 48.4% | 32.3% | 8 x -15, 6 x -9, 20 x -8, 60 x -14 | |||
| 3rd Line | 130 | 144 | 6 | 5 | 5 | 5 | 5 | 6 | 32.0 | 64.0 | 1.4% | 21.1% | 18.5% | 32.3% | 48.4% | 8 x -15, 20 x -8, 60 x -14, 30 x -7 | ||
| 4th Line | 115 | 129 | 10 | 8 | 8 | 8 | 8 | 10 | 52.0 | 116.0 | 21.1% | 21.1% | 18.5% | 48.4% | 8 x -15, 60 x -14, 30 x -7, 125 x -13 |
- 1 -10PA (random position)
- 8 -15PA (2G, 2C, 1 everything else)
- 6 -9PA (1 every position)
- 20 -8PA (4G, 4C, 3 every other position)
- 60 -14PA (9G, 9C, 7 every other position)
- 30 -7PA (5G, 5C, 4 every other position)
- 125 -13PA (25C, 20 every other position)
Thank you for this.
This explain why the draft goes how it goes with the top 45 having a bunch of NHLer and the rest being pretty barren. You have 30-35 top6 fwd/top4 D/ starting goalie going in the top 45 and then the rest is either bust or bottom 6/bottom pair/ backup goalie.
I just find it weird that I don't really see any steals in later round or UDFA as I would in some older DB. I have no idea how you would make some prospects "fall" to the 4-7th round to add the excitement factor of drafting a steal.
This explain why the draft goes how it goes with the top 45 having a bunch of NHLer and the rest being pretty barren. You have 30-35 top6 fwd/top4 D/ starting goalie going in the top 45 and then the rest is either bust or bottom 6/bottom pair/ backup goalie.
I just find it weird that I don't really see any steals in later round or UDFA as I would in some older DB. I have no idea how you would make some prospects "fall" to the 4-7th round to add the excitement factor of drafting a steal.
In my most recent test sim this guy was picked at 104th overall. He had a 62 goals in one season once he made the NHL.
https://media.discordapp.net/attachments/682046485038104721/723694405973835816/unknown.png
He went low because he was still playing AA hockey at draft year but a PA of 188 still gets you a highish rank). In his D+1 he played the SJHL. He didn't break the NHL until D+5. He started at 21CA, so maybe you can start a guy at 1CA and 1 reputation and there's a chance he won't get up to 50-60 CA at draft time. It's another thing on my "to research" list.
There was a 95th overall that got 64 pts in his first season. He was drafted out of CzeJr, so reputation was probably low as well.
I call this one: we win? I guess...
doing a test where I release all of the players from 1 team into free agency to see what happens... this was my "all free agents" arizona coyotes in there first game of the pre season against vegas...
there was a line brawl that included the goalies!
but vegas forfeited... the game doesnt explain why, although Lehner got thrown out with a misconduct and it did say "fleury was hurt on the play but continued on"
maybe the injury was more serious? but after the game he didnt get injured for any significant time.
doing a test where I release all of the players from 1 team into free agency to see what happens... this was my "all free agents" arizona coyotes in there first game of the pre season against vegas...
there was a line brawl that included the goalies!
but vegas forfeited... the game doesnt explain why, although Lehner got thrown out with a misconduct and it did say "fleury was hurt on the play but continued on"
maybe the injury was more serious? but after the game he didnt get injured for any significant time.
Attachments
WeThreeKings
Habs cup - its in the BAG
If you wanna set up a discord and a server to test with some online players. Let me know. I was very excited for the prospect gen because of issues and helped run FDL so it definitely hurt. Appreciate the work and the concept. So I'm good to aid in any way to get it to a better place.
In my Detroit save (Draft only, no bundling mediocre players for better picks, only trade for picks if player can bring 1st or 2nd), I have won my 5th President trophy (2031-2032 to 2035-2036) in a row with a 132 pts 63-13-6 record. I'm surprised as my last top 10 pick season was 2025 (except a #8 in 2032 I got in a trade and he's still stuck in the AHL) and I traded Seider in January as he was asking for 7.2 millions for multiple years at 34 and I have younger guys.
Cholowski and Zadina (the last original Wings) physical are taking a big drop (Zadina will need to test FA to lower his salary as I am not paying 9 millions for a 36 years old with decaying stats). They maybe have a year or two left with me. I already let Larkin go expecting him to retire at 39, but he decided to have one last hurrah with the Panthers posting 58 points (34 points the year before with me). Mantha retired a Red Wings in 2034.
We'll see if this team can 5-peat the President Trophy/Stanley Cup combo.
Cholowski and Zadina (the last original Wings) physical are taking a big drop (Zadina will need to test FA to lower his salary as I am not paying 9 millions for a 36 years old with decaying stats). They maybe have a year or two left with me. I already let Larkin go expecting him to retire at 39, but he decided to have one last hurrah with the Panthers posting 58 points (34 points the year before with me). Mantha retired a Red Wings in 2034.
We'll see if this team can 5-peat the President Trophy/Stanley Cup combo.
abo9
Registered User
- Jun 25, 2017
- 9,091
- 7,184
Just thought I'd share something of my most recent save:
For some reason, the Penguins built a dynasty at the same time as I managed to build up Seattle. For about 7-8 years, we've exchanged the Stanley Cup wins. But legit: Seatle - Penguins - Seatle - Penguins ... for 8 seasons
I've never seen the AI build such strong teams by themselves. Colorado somehow managed to keep together a core of Mack - Rantanen - Lafreniere - Byram - Makar - Girard for years too (and I'm probably forgetting a forwar or two). It was insane.
For some reason, the Penguins built a dynasty at the same time as I managed to build up Seattle. For about 7-8 years, we've exchanged the Stanley Cup wins. But legit: Seatle - Penguins - Seatle - Penguins ... for 8 seasons
I've never seen the AI build such strong teams by themselves. Colorado somehow managed to keep together a core of Mack - Rantanen - Lafreniere - Byram - Makar - Girard for years too (and I'm probably forgetting a forwar or two). It was insane.
Just thought I'd share something of my most recent save:
For some reason, the Penguins built a dynasty at the same time as I managed to build up Seattle. For about 7-8 years, we've exchanged the Stanley Cup wins. But legit: Seatle - Penguins - Seatle - Penguins ... for 8 seasons
I've never seen the AI build such strong teams by themselves. Colorado somehow managed to keep together a core of Mack - Rantanen - Lafreniere - Byram - Makar - Girard for years too (and I'm probably forgetting a forwar or two). It was insane.
The AI can definitely do it in certain circumstances. The financial figures vs. the cap is the key one. The other key parts are player ambition and loyalty as well as club success of course.
Updated the 06-07 DB to have prominent players up to Bedard/Kiviharju doing some test runs now to see how it plays out then going to try and get the Sedins a much deserved cup before they retire
alko
Registered User
Really bummed we aren't starting NHL UFA next week but I've been able to add a lot of detail into the DB over this downtime, really excited for the first release...which we are still tracking to come alongside 1.5.
tealhockey
@overtheboards
Does anyone use the editor? I haven't been able to get it to load my current saved games (nothing shows up but it acts like it is loading it up). It works when I want to see databases
- Status
