smozoma

Home/Away Advantages Analysis

22 posts in this topic

(Not quite sure where to put this, I went with the Player Ratings area since it is (or rather will be) mainly about how home/away advantages affect the players.

Home/Away Advantages Analysis

Plabax's Plea for making the home/away advantages equal in future seasons of (Genesis) Classic League made me wonder, what does the data say about these advantages, and how do they actually work?

I'm ignoring PP/PK as that wasn't part of the discussion, but that would an interesting thing to check out as well, using a similar technique.


What are the Home/Away Advantages*?

*Not the proper name, as we shall learn shortly.

Each team has a pair of attributes that modifies their behaviour at home and on the road. These are shown in the game manual (PDF) for each team, starting on page 55. They are referred to as "Home Ice Advantage" and "Road Ice Disadvantage." In the game's code, these values are represented by numbers, typically 0 ("Weak"), 1 ("Average"), and 2 ("Strong"). Additionally, some expansion teams have a value of 3 for their Away Ice Disadvantage; the manual mistakenly calls this "Low", but I named it "Brutal" a few years ago when developing EARE.
To make the name of these attributes shorter, I'll call them the "Home/Away Modifiers". Or the HAMs for short.


What the Home/Away Modifiers Do

I looked at these in the code many years ago, and took some notes, but the notes were not very clear, so I can't say specifically how the modifiers work. I intend to investigate this again soon and will do a Part II of this post. Since this post does not explain what exactly the home/away modifiers do, there can be no real conclusion with regard to Plabax's Plea. However, I do believe / vaguely recall that the values tend to favour the home team more than the away team, which is what makes most teams perform better at home than on the road.

Classic League Data

I took the Home/Away records of all A or A/B leagues available on nhl94online.com, as well as 4 more seasons from 2006-2008 that Halifax somehow had stashed away (which chaos hunted down for me), plunked it into excel, did some excel-y stuff and came up with the following.

Team Results

NDmMXKr.png


The meanings of Columns 1-14 are pretty obvious. The "results" are columns 15 (H-A) and 16 (H/A). These columns both quantify how much better at home a team has performed than on the road.

  • "H-A" is "Home W% minus Away W%" (column 10 minus column 14).
  • "H/A" is "Home W% divided by Away W%" (column 10 divided by column 14)

Things to note (about interpreting the data):

  • I ignored ties completely in calculating Win Percentages
  • I only counted data from seasons where that team had a Pts% greater than .200 or less than .800, because, #1 I felt that you couldn't tell too much from really dominant or outclassed players and, #2 it smoothed out the data a bit (which encourages me that #1 is correct).
  • In column 5 (Total W+L), I have highlighted in red the teams that don't have a large number of games played, which means their results are less reliable than the other teams. These teams tend to be outliers with their groups (i.e., teams with the same HAMs), so tend to be at the top or bottom of the group (Leafs and Capitals in the 1/1 group, Lighting in the 0/0 group, Penguins and Blues in the 2/1 group). The teams with more games played are likely more representative of the true effects of the particular HAM values.
  • I split out the teams with fewer than 100 total W+L because there is not enough data to trust their values.
  • The table is sorted first by Total Advantage (column 4), then Home Advantage (column 2), then by H/A (column 16)
  • Probably a better way to look at the H/A values are to think of them as in the following table, as the home and away win percentages for an average team, or what their home and away records would be in a 40-game Classic season:
    • wgcaqpT.png
    • So for a team like the Flyers, who have a 135% H/A, it basically amounts to winning 1.5 more games at home than if there were no such thing as home ice advantage in the game.
    • As the HAMs may actually make some teams better on average than others, maybe the values should be shifted a bit more towards higher HW% and higher AW%. But that's for the followup post...

Observations & Conclusions

  • The HAMs do in fact have an effect on how much teams win at home or on the road:
    • There is a very high correlation between H/A and Total HAMs (column 4). The correlation function value is 0.736, which is very high/conclusive.
    • The correlation between H-A and Total HAMs is also very high (0.687)
    • However, because of the missing information about how the HAMs are actually used by the game, we can not say if teams are given large home boosts, or receive smaller boosts to their abilities at home and de-buffs to their abilities on the road.
  • 0/0
    • This means a team has Weak home advantage and Weak away disadvantage.
    • HAMs of 0/0 appear to pretty much neutralize home ice advantage, although the Stars' 113.2% is still quite high.
  • 1/1
    • This means a team has Average home advantage and Average away disadvantage.
    • These are the HAMs that GDL uses.
    • It also appears to neutralize home ice advantage, although not to the extent of 0/0. Considering the high effects of 2/1 and 1/2 HAMs we'll look at next, I'm surprised there isn't a more noticeable separation between 0/0 and 1/1. //(Maybe when I look into what the HAMs do in the code, the reason for this will become clearer)//
  • 1/2
    • This means a team has Average home advantage and Strong road disadvantage.
    • There are only 2 teams in this group, and neither has many games played, so the H/A values are not very reliable, but do suggest the value is around 120%, quite significant.
  • 2/1
    • This means a team has Strong home advantage and Strong road disadvantage.
    • These are the HAMs I used in Blitz.
    • We expect these teams to have a high disparity in home and away records, and this is backed up by the data, with these teams having around a 120+% H/A (or 130% for the teams with fewer games played).
  • 1/3
    • This means a team has Average home advantage and Brutal road disadvantage.
    • These are bad expansion teams, so there is not much data for them, although they do seem to be around 130% H/A
  • 2/2
    • This means a team has Strong home advantage and Strong road disadvantage.
    • Only the Flyers have this HAM. Their real-life home/away records in the 92-93 season were 23-14-5 / 13-23-6.
    • They have a huge 135.8% H/A, but with only 294 games played, there's a chance this value is a bit of an outlier, although seeing the pattern in the table it looks like it should be fairly accurate.

What is unclear at the moment is, if you have two identical teams, but with different HAMs, do certain HAMS make one of the teams actually better than the other? And how powerful is this effect?

Combined Results

To deal with the issue of teams with low amounts of games played, we can group the data, not by teams, but by HAMs:

RWW5N4Q.png


This grouping also includes the data from the teams with under 100 GP.

  • The data is pretty clean-looking now. 2/1, 1/1, and 0/0 each have over 2600 games played and I think should be considered accurate. 2/2, 1/3, and 1/2 still don't have a lot of games, but at least they fit the pattern well.
  • 0/0 and 1/1 teams have a slight home advantage of around 106%. This is equivalent to home records of around 10.3-9.7 for a .500 player.
  • 1/2 and 2/1 teams have a significant home advantage of around 120%. This is equivalent to home records of around 10.9-9.1 for a .500 player. In other words, for every 20 games played, they win one game that they would otherwise have lost.

Next Steps

  1. I will investigate how the game actually uses the HAMs by doing code and RAM traces with the hacking emulator http://forum.nhl94.com/index.php/topic/16742-tool-gens-rerecording-emulator-for-hacking/?p=147027.
  2. Something that is not covered in this post is how teams with different HAMs fare against each other. This could be analyzed in a similar manner but using schedule results (for example by going through all the pages like this http://nhl94online.com/html/coachpage.php?lg=31&sublg=GENS-A&team_ID=1165. Chaos gave me a copy of the database recently and I may some time try this.)

Images uploaded to the forum for safekeeping incase they disappear off imgur:

post-253-0-69884500-1469929850_thumb.png

post-253-0-70791700-1469929853_thumb.png

post-253-0-88648200-1469929855_thumb.png

1 person likes this

Share this post


Link to post
Share on other sites

(reserving follow up post 1)

1 person likes this

Share this post


Link to post
Share on other sites

(reserving follow up post 2)

1 person likes this

Share this post


Link to post
Share on other sites

Nice analysis!! There is also a link between stick handling and total HAM. The higher the total, the more likely a player with at least 4STH is likely to "toddle". That's why you see it much more often in Blitz (3TOT with 2/1) vs. GDL (2TOT with 1/1).

Share this post


Link to post
Share on other sites

Nice analysis!! There is also a link between stick handling and total HAM. The higher the total, the more likely a player with at least 4STH is likely to "toddle". That's why you see it much more often in Blitz (3TOT with 2/1) vs. GDL (2TOT with 1/1).

From my old notes that I don't quite understand, you are definitely correct. Shooting accuracy is also affected, as well as the awarenesses and a few others, but I may not get that stuff investigated until September :(

I updated my post to correct the part where I said Blitz uses 1/1, thanks

Share this post


Link to post
Share on other sites

good stuff as always. thanks for doing this!

Share this post


Link to post
Share on other sites

Big Ups!

Share this post


Link to post
Share on other sites

I haven't digested everything written above (and I may not even be able to, due to limited brain function) but I have to say Well Done! Really impressed with these "magazine quality" articles that people like Smozoma put together. Awesome stuff! :)

Share this post


Link to post
Share on other sites

Smoz is a big reason why this community is so amazing ....he could probably hack into Donald trumps emails but he be bored as f**k!

Share this post


Link to post
Share on other sites

talk about being thorough....great stuff dude

Edited by Uncle Seth

Share this post


Link to post
Share on other sites

WOW!

Share this post


Link to post
Share on other sites

Smoz is a big reason why this community is so amazing....

f'real. I wish there was a way to convince him to take on any more SNES hacks that seem to come so naturally to him. when he got so annoyed by unpolished attempts to have defense control default to active, it seemed to take him zero minutes to find the simplest fix. I'm SURE that the SNES dude could employ so many of the awesome hacks that the Genesis folks take for granted if we could just flail and fail dramatically first to trick him into swooping in like a caped hero.

Share this post


Link to post
Share on other sites

This is friggin' phenomenal. Love this approach, great way of looking at this and the results are quite strong! Great work Smoz!

Share this post


Link to post
Share on other sites

It does not seem much in a regular season, but that swing game can turn an even playoff series.

I Have different results with my teams.

Stl home 16-4 away 16-4

Mtl home 9-9-1 away 13-6

Bos home 14-6 away 16-4

Just take team advantages in to consideration when picking team if it is a huge deal to some. It does not show in my results with all teams being 2/1. :)

Edited by Depch

Share this post


Link to post
Share on other sites

Well, this ^ wasnt meant as a showstopper for this. But over a 40 game season you win possibly a couple of more wins than you are supposed to and the wings record shows that roster plays a way bigger role. It would be nice to know what the effects truly are for it does exist and I would support eliminating the effects on a modified rom league like gdl, I dont think the effects are gone by just adjusting the sliders or if then all effects should be "weak" afaik brutal was not applicable to all.

As for my own record I have had the hunch/feel that my home games feel "heavier" to play with and have experienced more sluggish games there. Seth knows what I am talking about when I once noticed his players skating on a porridge (opponent can basically skate around you then, but that bad stuff only happens against some players, but there are games where the connection/controls are not equal while more rare). So my record can somewhat be explained with that maybe? But I have always enjoyed playing down more than up as well.

Edited by Depch

Share this post


Link to post
Share on other sites

Yes, the effect is not that large. I showed that the Flyers, with the largest home/away disparity would have a 11.5-8.5 record at home instead of 10-10 due to their advantages.

You are correct that setting the modifiers to 1-1 or 0-0 does not remove the effects. I will be investigating the actual effects of the bonuses in the next few weeks and how to eliminate them.

Share this post


Link to post
Share on other sites

As for my own record I have had the hunch/feel that my home games feel "heavier" to play with and have experienced more sluggish games there. Seth knows what I am talking about when I once noticed his players skating on a porridge (opponent can basically skate around you then, but that bad stuff only happens against some players, but there are games where the connection/controls are not equal while more rare). So my record can somewhat be explained with that maybe? But I have always enjoyed playing down more than up as well.

Hot & Cold is the main culprit. In GDL, I know all my players' avg ratings, check them before every game, religiously. When players are very cold and you have several of them out there, their agility and speed are severely affected and noticeable. Especially when you're talking about guys in the 4 rating range and lower, the drop-off in ability is even greater, but so is the boost they get when they're red hot. I would know, I'm known for my love affair of of non-elite players in the average range. But it's a love-hate relationship. When they're very cold, their skate takes a big nosedive. Also remember, agility and speed are closely related so when they BOTH take a hit, you get the sluggish, skating-in-sand effect. I have no tolerance for it. If a guy is really cold, he isn't seeing the ice under for me. Those who don't notice these effects are simply not very perceptive enough/oblivious. Or their playing style of skating it primarily with their superstar means they just aren't paying attention to what their other guys are playing like....because those other skaters don't have the puck.

Smoz has showed that the home boost increases shot accuracy in classic. I don't notice any difference in performance in home/away in GDL in regards to skating though.

If you think guys are sluggish in a given game, check their ratings and compare it to their average ratings. You'll see it's no mystery why they feel like they're wearing cement boots.

Edited by Uncle Seth
1 person likes this

Share this post


Link to post
Share on other sites

Hot & Cold is the main culprit. In GDL, I know all my players' avg ratings, check them before every game, religiously. When players are very cold and you have several of them out there, their agility and speed are severely affected and noticeable. Especially when you're talking about guys in the 4 rating range and lower, the drop-off in ability is even greater, but so is the boost they get when they're red hot. I would know, I'm known for my love affair of of non-elite players in the average range. But it's a love-hate relationship. When they're very cold, their skate takes a big nosedive. Also remember, agility and speed are closely related so when they BOTH take a hit, you get the sluggish, skating-in-sand effect. I have no tolerance for it. If a guy is really cold, he isn't seeing the ice under for me. Those who don't notice these effects are simply not very perceptive enough/oblivious. Or their playing style of skating it primarily with their superstar means they just aren't paying attention to what their other guys are playing like....because those other skaters don't have the puck.

Smoz has showed that the home boost increases shot accuracy in classic. I don't notice any difference in performance in home/away in GDL in regards to skating though.

If you think guys are sluggish in a given game, check their ratings and compare it to their average ratings. You'll see it's no mystery why they feel like they're wearing cement boots.

No, this is not what I'm talking about, maybe you've forgotten the game where I especially approached you about the sluggishness you had, because it was so visible even to the opponent. This would be controversial to my home records in classic as well where you can't check the true values since they are masked (and remember my teams have had advantage 2 at home, it should affect the awarenesses at least as well), but this one game I was talking about was in gdl exi I think.

Edited by Depch
1 person likes this

Share this post


Link to post
Share on other sites

No, this is not what I'm talking about, maybe you've forgotten the game where I especially approached you about the sluggishness you had, because it was so visible even to the opponent. This would be controversial to my home records in classic as well where you can't check the true values since they are masked (and remember my teams have had advantage 2 at home, it should affect the awarenesses at least as well), but this one game I was talking about was in gdl exi I think.

I do remember that game. But I can't recall if it was classic or GDL. All I know is the games some of my guys are sluggish (in GDL), is when they are ice cold. And I've never had that happen why my players were only mildly cold, average, or hot. This observation of mine spans the entire period of time since GDL went to transparent ratings, and includes the thousands of exis I've played.

1 person likes this

Share this post


Link to post
Share on other sites
Quote
  • The data is pretty clean-looking now. 2/1, 1/1, and 0/0 each have over 2600 games played and I think should be considered accurate. 2/2, 1/3, and 1/2 still don't have a lot of games, but at least they fit the pattern well.
  • 0/0 and 1/1 teams have a slight home advantage of around 106%. This is equivalent to home records of around 10.3-9.7 for a .500 player.
  • 1/2 and 2/1 teams have a significant home advantage of around 120%. This is equivalent to home records of around 10.9-9.1 for a .500 player. In other words, for every 20 games played, they win one game that they would otherwise have lost.

For Brutz League, I currently have all teams set to 1/1 (same as GDL).

Blitz did 2/1 which I like better.

I think Plabs did 0/0.

Hope I am not opening a can of worms.

Brutus?

 

 

Edited by CoachMac

Share this post


Link to post
Share on other sites
1 hour ago, CoachMac said:

For Brutz League, I currently have all teams set to 1/1 (same as GDL).

Blitz did 2/1 which I like better.

I think Plabs did 0/0.

Hope I am not opening a can of worms.

Brutus?

 

 

I prefer NO home/away advantage, especially for Season 1. 

I will discuss it with some of the other coaches via Private chat and get this determined before end of tonight.

Share this post


Link to post
Share on other sites
1 hour ago, Brutus said:

I prefer NO home/away advantage, especially for Season 1. 

I will discuss it with some of the other coaches via Private chat and get this determined before end of tonight.

You can't get to none.

The current 1/1 or 0/0 minimize it the most per Smozoma's post. (106%)

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now