Is it better to balance your lines or go top-heavy? I ran some tests

Argonaut · 07-30-2020, 04:47 PM

TOP-HEAVY OR BALANCED?

I'm back, with more of a positive test today. I've seen it asked a few times whether it's better to stack your top line, or spread your best players throughout your lineup.

This is a real question that NHL teams have grappled with, and different approaches have worked for different teams at different times. My team, the Blackhawks have almost always separated Toews and Kane at even strength. This approach can often be beneficial for working the matchup game. But there's something to be said for stacking your best guns together.

The team I'm going to be looking at is the 1995-96 Penguins. I decided against the Penguins from the earlier 90s, because those teams were a bit more stacked and had enough quality depth to basically make every line an awesome one. Also, in 95-96 the Penguins mostly ran with Lemieux-Francis-Jagr on the top line, the most top-heavy setup imaginable. But what if they went with a more balanced approach?

Injuries and suspensions off of course, and I'm sure I'm forgetting some other things I did to limit variables.

THE TEST

For the 1st test I ran "The Great Line" as it was called, and basically went with the next best players in descending order for lines 2-4. The big guys also played together on the power play. For the 2nd test I kept everything the same with the defense pairings, and PP/PK. But Lemieux, Jagr, and Francis each had their own line at even strength. I made sure each of them had at least a decent player or two on their line. I also spread the minutes around more evenly.

THE RESULTS

To my surprise, the balanced approach was pretty definitively the better option in my simulations. The Penguins scored more, allowed fewer goals, and had more team points when splitting up Lemieux, Jagr, and Francis. Perhaps the spread around skill was tough for opposing matchups? I had a quick glance at the numbers in one balanced sim, and Jagr and Francis had an absurd percentage of their points on the power play though (like two-third of their points). So I don't know if that means something.

CONCLUSION

It seems like a pretty good idea to go with a balanced approach for your even strength lines in FHM. It may depend on your personnel available I suppose. And load up the power play though of course.

Sidenote: I believe Lemieux led the league in points in every single sim. Mario Lemieux with injuries turned off is a very scary thing!

EDIT: Also, this is my #999 post and I've given out 444 thanks. Nice.

ZootMurph · 07-30-2020, 09:57 PM

I'd like to see this test with a less loaded team. 3 quality players with not so much talent around them. I think that would really have different results.

As I'm just starting out playing and learning the game, I'd don't have the knowledge or time to do the testing myself, otherwise I would.

DFyvie · 07-31-2020, 06:31 AM

I wonder if you did the experiment with the Present Day Edmonton Oilers. Splitting up versus combining Draissaitl and McJesus. It would be a more binary test so a bit simpler. The Oilers are a bit of a flawed team in many ways so I'm not sure if it would make the results more of less drastic though.

#4BobbyOrr · 08-03-2020, 04:50 AM

Quote:

Originally Posted by DFyvie

I wonder if you did the experiment with the Present Day Edmonton Oilers. Splitting up versus combining Draissaitl and McJesus. It would be a more binary test so a bit simpler. The Oilers are a bit of a flawed team in many ways so I'm not sure if it would make the results more of less drastic though.

I've done this but didn't take down any numbers or anything from it, just observations. In the first year or two, it doesn't work particularly well as there isn't enough talent on the roster for either to thrive with their own line and I don't think either had a top 5 point finish in the first two years (19-20 and 20-21) when they were separated. In year three I put them together and their numbers both exploded with McDavid winning the Art Ross and Draisaitl coming in 3rd despite missing games.

As I started drafting and developing other top talents and as a few already drafted guys like Yamamoto and Lavoie started panning out I was able to spread the talent better but did put them together some years.

For what it's worth over a about a 15 year save I never won a cup with them together but won 3 with them on separate lines.

TheZomBee · 08-11-2020, 02:02 PM

I decided to test this myself after reading the post, since it’s one of the most popular debates in team-building and I’ve never found any evidence one way or the other.

So, I simulated ten individual seasons with the Bruins, all of them under the same conditions, using the same teammates and tactics (auto generated, along with player roles), no injuries, development, transactions, etc.

The first five seasons I played Marchand, Bergeron and Pastrnak together, and the lines looked like this:

After the five seasons, these were their numbers:

Seasons 6-10, I separated all three forwards, leaving everything else as similar as possible, and the lines looked like this:

Things to consider:

- I put Pastrnak on the first line, Bergeron on the second and Marchand on the third based on their Point totals from the first part of the test. Changing them around could show different results.
- I could’ve put DeBrusk on the LW first line, Ritche on the second, etc., but this way I made as few moves as possible. This possibly has an impact on the results as well.
- Special teams remained the same.
- Tukka Rask played every game of all ten seasons. That’s probably a reason why the Point totals are so high, since other teams were rotating Goalies, but not the Bruins. But I wanted the Goalies to have as little impact as possible, and this way it was always the same for all ten seasons.

And here are the numbers for seasons 6-10:

Conclusions:

- My main conclusion is that this is anything but conclusive. The variance between seasons is too high to find any clear evidence, and I would need a lot more than ten seasons for the averages to matter.
- It does look like the team scored more Goals when the players were separated, but they also allowed more. Again, it’s hard to tell why since so many other factors need to be considered and the sample size is too small in this case.
- Overall, the team had very similar success, both in regular season and Playoffs. Point totals are very similar, and each time the Cup was won twice.
- Individually, the players will do much better when they play together, for obvious reasons, but they were also a lot more consistent each season. When they play together you can expect a level of performace from them that is much more difficult to predict when they are separated.
- Playing them separated means depth players will get a bigger share of the Points (players like DeBrusk or Coyle greatly benefit from this), which should result in them asking for more money, so in terms of team-building it could be worse. But you could also argue that this way you can play an average player like Ritchie (with his average salary) on the first line and have the same overall results, so you could save money that way. Again, really hard to lean one way or the other.
- Special teams are a big part of the game, and the variance on Power Play performances has also a big impact on the final results.
- There’s a reason why every team in real life adopts a different strategy when it comes to this topic, and I can’t see a strategy being better than the other in the game, either.

Argonaut · 08-11-2020, 03:16 PM

Thanks for doing your own tests, that's what I like to see in my threads... additional testing and data more so than armchair critiquing!

I agree that it depends on the team and you'll probably get different results with different teams, eras, and methods. It's too bad there's no simulation module like in OOTP, otherwise I could have tens of thousands of simulations at a time.

The historical database sims pretty fast though. What ends up taking up time is CTRL-ALT-DEL closing the game and rebooting it to run the test again. That dastardly autosave on exit otherwise...

One thing I notice is that Rask is hot in the 1st picture and cold in the 2nd picture. I don't know if that had some effect on your goals against? I've seen that goalie hotness has a huge impact on how well your team does. I wish there was an option to turn it off.

07-30-2020, 04:47 PM	#1
Argonaut All Star Starter Join Date: Apr 2019 Posts: 1,085	Is it better to balance your lines or go top-heavy? I ran some tests TOP-HEAVY OR BALANCED? I'm back, with more of a positive test today. I've seen it asked a few times whether it's better to stack your top line, or spread your best players throughout your lineup. This is a real question that NHL teams have grappled with, and different approaches have worked for different teams at different times. My team, the Blackhawks have almost always separated Toews and Kane at even strength. This approach can often be beneficial for working the matchup game. But there's something to be said for stacking your best guns together. The team I'm going to be looking at is the 1995-96 Penguins. I decided against the Penguins from the earlier 90s, because those teams were a bit more stacked and had enough quality depth to basically make every line an awesome one. Also, in 95-96 the Penguins mostly ran with Lemieux-Francis-Jagr on the top line, the most top-heavy setup imaginable. But what if they went with a more balanced approach? Injuries and suspensions off of course, and I'm sure I'm forgetting some other things I did to limit variables. THE TEST For the 1st test I ran "The Great Line" as it was called, and basically went with the next best players in descending order for lines 2-4. The big guys also played together on the power play. For the 2nd test I kept everything the same with the defense pairings, and PP/PK. But Lemieux, Jagr, and Francis each had their own line at even strength. I made sure each of them had at least a decent player or two on their line. I also spread the minutes around more evenly. THE RESULTS To my surprise, the balanced approach was pretty definitively the better option in my simulations. The Penguins scored more, allowed fewer goals, and had more team points when splitting up Lemieux, Jagr, and Francis. Perhaps the spread around skill was tough for opposing matchups? I had a quick glance at the numbers in one balanced sim, and Jagr and Francis had an absurd percentage of their points on the power play though (like two-third of their points). So I don't know if that means something. CONCLUSION It seems like a pretty good idea to go with a balanced approach for your even strength lines in FHM. It may depend on your personnel available I suppose. And load up the power play though of course. Sidenote: I believe Lemieux led the league in points in every single sim. Mario Lemieux with injuries turned off is a very scary thing! EDIT: Also, this is my #999 post and I've given out 444 thanks. Nice. Attached Images __________________ Texas Rangers Dynasty (Est. 1972) Hartford Whalers Dynasty: Brass Bonanza Last edited by Argonaut; 07-30-2020 at 04:52 PM.

08-11-2020, 03:16 PM	#6
Argonaut All Star Starter Join Date: Apr 2019 Posts: 1,085	Thanks for doing your own tests, that's what I like to see in my threads... additional testing and data more so than armchair critiquing! I agree that it depends on the team and you'll probably get different results with different teams, eras, and methods. It's too bad there's no simulation module like in OOTP, otherwise I could have tens of thousands of simulations at a time. The historical database sims pretty fast though. What ends up taking up time is CTRL-ALT-DEL closing the game and rebooting it to run the test again. That dastardly autosave on exit otherwise... One thing I notice is that Rask is hot in the 1st picture and cold in the 2nd picture. I don't know if that had some effect on your goals against? I've seen that goalie hotness has a huge impact on how well your team does. I wish there was an option to turn it off. __________________ Texas Rangers Dynasty (Est. 1972) Hartford Whalers Dynasty: Brass Bonanza

07-30-2020, 09:57 PM	#2
ZootMurph Minors (Double A) Join Date: Jun 2015 Posts: 124	I'd like to see this test with a less loaded team. 3 quality players with not so much talent around them. I think that would really have different results. As I'm just starting out playing and learning the game, I'd don't have the knowledge or time to do the testing myself, otherwise I would.

07-31-2020, 06:31 AM	#3
DFyvie Major Leagues Join Date: May 2014 Location: Cambridge, UK Posts: 389	I wonder if you did the experiment with the Present Day Edmonton Oilers. Splitting up versus combining Draissaitl and McJesus. It would be a more binary test so a bit simpler. The Oilers are a bit of a flawed team in many ways so I'm not sure if it would make the results more of less drastic though.

08-11-2020, 02:02 PM	#5
TheZomBee Bat Boy Join Date: May 2013 Posts: 15	I decided to test this myself after reading the post, since it’s one of the most popular debates in team-building and I’ve never found any evidence one way or the other. So, I simulated ten individual seasons with the Bruins, all of them under the same conditions, using the same teammates and tactics (auto generated, along with player roles), no injuries, development, transactions, etc. The first five seasons I played Marchand, Bergeron and Pastrnak together, and the lines looked like this: After the five seasons, these were their numbers: Seasons 6-10, I separated all three forwards, leaving everything else as similar as possible, and the lines looked like this: Things to consider: - I put Pastrnak on the first line, Bergeron on the second and Marchand on the third based on their Point totals from the first part of the test. Changing them around could show different results. - I could’ve put DeBrusk on the LW first line, Ritche on the second, etc., but this way I made as few moves as possible. This possibly has an impact on the results as well. - Special teams remained the same. - Tukka Rask played every game of all ten seasons. That’s probably a reason why the Point totals are so high, since other teams were rotating Goalies, but not the Bruins. But I wanted the Goalies to have as little impact as possible, and this way it was always the same for all ten seasons. And here are the numbers for seasons 6-10: Conclusions: - My main conclusion is that this is anything but conclusive. The variance between seasons is too high to find any clear evidence, and I would need a lot more than ten seasons for the averages to matter. - It does look like the team scored more Goals when the players were separated, but they also allowed more. Again, it’s hard to tell why since so many other factors need to be considered and the sample size is too small in this case. - Overall, the team had very similar success, both in regular season and Playoffs. Point totals are very similar, and each time the Cup was won twice. - Individually, the players will do much better when they play together, for obvious reasons, but they were also a lot more consistent each season. When they play together you can expect a level of performace from them that is much more difficult to predict when they are separated. - Playing them separated means depth players will get a bigger share of the Points (players like DeBrusk or Coyle greatly benefit from this), which should result in them asking for more money, so in terms of team-building it could be worse. But you could also argue that this way you can play an average player like Ritchie (with his average salary) on the first line and have the same overall results, so you could save money that way. Again, really hard to lean one way or the other. - Special teams are a big part of the game, and the variance on Power Play performances has also a big impact on the final results. - There’s a reason why every team in real life adopts a different strategy when it comes to this topic, and I can’t see a strategy being better than the other in the game, either.