For the Nth time - LOWER TV Diff - Even optionaly

Hi,

I am sure this has been chewed to death, but thought it was worth another try. Apologies if I missed a thread about just this topic.

I like playing competitive, and I am getting sick of almost every good team I have ending its run by running into +350-500TV teams - especially bashy ones.

It also gives unfair advantage to players who can put in MANY games in the first few games of the season - so they can run into less higher TV teams (as there are none around).

PLEASE reduce TV diff to something like 250-200 (at least in competitive play).

I would be totally fine looking longer for games or NOT playing at all instead of playing a high dif game. I have a feeling this would be the preference for most players.
If some players don't care - maybe offer this as an option.

I think currently this is the most broken aspect of the game and needs priority in fixing.

Thanks.

BB2 Champion Ladder Admin Team

Reducing TV differences across the board would result in smaller numbers of viable matches, which will in turn result in fewer matches played, and therefore longer wait times. It would also be far worse for higher TV teams where the population is more sparse.

A better solution would be to have a rising increase with games played. Start at, e.g., 300 TV max difference for new teams rising to 500k after 10 matches. Since the majority of teams are low TV it would have a far smaller impact while protecting the newer teams from developed teams.

last edited by dode74

@dode74 said in For the Nth time - LOWER TV Diff - Even optionaly:

A better solution would be to have a rising increase with games played. Start at, e.g., 300 TV max difference for new teams rising to 500k after 10 matches. Since the majority of teams are low TV it would have a far smaller impact while protecting the newer teams from developed teams.

What's the metric for "better" here? Is any of the "new team protection" theory based on numbers or is this just a feelsies thing?

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

What's the metric for "better" here? Is any of the "new team protection" theory based on numbers or is this just a feelsies thing?

Would an acceptable metric be a higher winrate (i.e. closer to 50%) for new teams against developed teams?

Where does the 'TV-difference means higher win-chance for overdog' get significant?

BB2 Champion Ladder Admin Team

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

What's the metric for "better" here? Is any of the "new team protection" theory based on numbers or is this just a feelsies thing?

More games played which would not have been played, particularly for experienced teams.

Using S10 data, there were 19021 matches played. Of those 19021 matches a straight limit of 300TV would have prevented 2314 matches (12%) from happening. By increasing the TV limits based on games played from 300 to 500 by 20TV per game and using the lower limit of the two teams that 2314 drops to 1431. That's 883 more matches made.

Experienced teams (>5 matches) played 4759 matches against each other (i.e. both teams were experienced). Of those matches, 893 (18.%) would be prevented by a straight TV limit of 300TV. With the sliding scale that drops to 318.

Of the 883 more matches made under the sliding scale system, 575 of them are between teams with >5 games.

last edited by dode74

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

Would an acceptable metric be a higher winrate (i.e. closer to 50%) for new teams against developed teams?

Well, my main question is: what is it that makes us think there's a problem that needs to be solved - what metric are we looking at to say there's actually an issue? It's peachy to say that new teams often face higher TV opponents - that's almost a given since new teams tend to be among the lowest TV teams and well below the average TV - but what are we looking at that tells us they need special matching rules?

@dode74 said in For the Nth time - LOWER TV Diff - Even optionaly:

More games played which would not have been played, particularly for experienced teams.

Ok, but that's a metric to explain why we don't want to simply lower the general cap from 500... I'm asking why we think there's any need for "new team protection" such that there will be a notable benefit to adding in a scaling cap rather than leaving things the way they are.

Anything that reduces the cap, even temporarily, will result in fewer matches made... so there really should be some demonstrable benefit to doing so... or at least some extant issue we can see in the numbers that gives us a reason to think that the loss of those potential matches is worthwhile to change those particular numbers.

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

Well, my main question is: what is it that makes us think there's a problem that needs to be solved - what metric are we looking at to say there's actually an issue?

The metric here would be the number/regularity of posts that seem to see an issue with it, i.e. the feelsies-rate seems to be very high 😉

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

but what are we looking at that tells us they need special matching rules?

That's why I asked the question whether the TV-difference has the same predictive power for low TV of the underdog (i.e. fresh teams) than for higher TV. Fresh teams in general have fewer dice-tempering abilities which can change dramatically once they get a few level-ups.

If it's the same on all levels, I'd say there really is no issue, as well. If it's higher for low-TV underdogs, it would be an argument for 'protection' as these matchups are more unfair and discouraging new coaches with especially hard matchups harms the playerbase which also leads to fewer matches being played.

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

The metric here would be the number/regularity of posts that seem to see an issue with it, i.e. the feelsies-rate seems to be very high 😉

Based on that metric we'd better focus all our attention on fixing the RNG instead, regardless of what the data says about it being fine.

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

That's why I asked the question whether the TV-difference has the same predictive power for low TV of the underdog (i.e. fresh teams) than for higher TV. Fresh teams in general have fewer dice-tempering abilities which can change dramatically once they get a few level-ups.

Well, lets take a look at CCL season 3's data (since I was working on that recently and already have relevant fields created). We'll compare "fresh" teams (new teams with less than 5 games played) with teams that have 5 or more games played and see how things are:

New Teams (less than 5 games played)

Correlation between TV advantage (their TV minus the other guy's TV) and match outcome: r = 0.133, p < 0.01
Correlation between cTVP advantage (their cTVPlus rating - other guys) and match outcome: r = 0.187, p < 0.01

TV advantage predicted 46.3% of outcomes for non-draw matches. cTVP advantage predicted 48.9% (ctvpf 86.6%) of outcomes for non-draw matches.


Other Teams (5 or more games played)

Correlation between TV advantage (their TV minus the other guy's TV) and match outcome: r = 0.166, p < 0.01
Correlation between cTVP advantage (their cTVPlus rating - other guys) and match outcome: r = 0.332, p < 0.01

TV advantage predicted 56.5% of outcomes for non-draw matches. cTVP advantage predicted 63.5 (ctvpf 77.4%) of outcomes for non-draw matches.


So, the relationship between TV advantage and match outcome is weak for new teams - weaker than it later becomes, but weak all around. In fact, based on TV advantage's ability to predict the winner in games that have a winner, TV difference is literally worse than flipping a coin at predicting the winner accurately for new teams.

But lets look at those new teams at the suggested breakpoint of 300 TV difference and see how things are 0-290 vs 300-500:

Below 300 TV difference

Correlation between TV advantage and match outcome: r = 0.075, p < 0.01
Correlation between cTVP advatnage and match outcome: r = 0.164, p < 0.01

TV advantage predicted 44.5% of outcomes for non-draw matches. cTVP advantage predicted 47.3% (ctvpf 81.1%) of outcomes for non-draw matches.


Beyond the 300 TV difference

Correlation between TV advantage and match outcome: r = 0.139, p < 0.01
Correlation between cTVP advatnage and match outcome: r = 0.288, p < 0.01

TV advantage predicted 68.4% of outcomes for non-draw matches. cTVP advantage predicted 69.3% (ctvpf 76.9%) of outcomes for non-draw matches.


We find that the overall relationship between TV differences and outcomes is still quite weak, especially in comparison to our cTVPlus rating differences, but that the ability of TV difference to predict the winner in non-draw games goes up quite a bit. Is that a case for TV protection? Well, we'll need one more comparison to decide that, unless we're trying to guarantee wins for new teams (which is where the pressure ends up being given the < 50% prediction rate for TV advantage, which suggests that for new teams playing less than 300 TV differences, a TV DISadvantage seems to be a better predictor of winning the game). We need to look at non-new (5 or more games) teams and how playing games over 300 TV difference looks:

Teams with 5 or more games, 300 or more TV difference

Correlation between TV advantage and match outcome: r = 0.287, p < 0.01
Correlation between cTVP advatnage and match outcome: r = 0.435, p < 0.01

TV advantage predicted 66.9% of outcomes for non-draw matches. cTVP advantage predicted 69.8% (ctvpf 78.1%) of outcomes for non-draw matches.


What that's showing us is that there's not much difference between 300+ TV difference outcomes for teams with less than 5 games played, or teams with more than 5 games played. If anything, the relationship between TV difference and outcome of matches gets stronger the more games a team has played, not weaker, which is what the whole "new team protection" concept is predicated on.

Given that, for these "new teams" the TV advantage had a mean of -42.6 and an SD of 136.4, we're seeing 95% or more of the games they play be in that under 300 TV range anyway, and they have a win rate of 56.3% in matches where they don't decide to concede, I think things are just fine the way they are.

last edited by VoodooMike

Ok, the really high non-draw prediction numbers for cTVP were really bugging me so I went back to the data and realized that it was cTVPF not cTVP, which is to say, their final cTVP rating, not their in-place cTVP rating. The proper metric was used for the correlations, but not for the prediction %.

I'm going to go back and recalculate those for each, and edit the post to have the correct numbers... but will leave the originally stated numbers in brackets.

Ultimately, the cTVP numbers aren't directly relevant to this discussion, but are included for comparison, since they're what's being used for matchmaking.... usually.

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

New Teams (less than 5 games played)

First of all, thanks for that rundown.

Several thoughts on this:

a) Even if the correlation between new teams and non-new teams for >300TV is comparable, wouldn't reducing that advantage reduce the number of losses for fresh teams and thus the number of teams being restarted?

b) Is it really prudent to ignore concessions for the winrate of fresh teams? Disconnections aside, we still have 20 to 25% concessions in CCL and (I assume) normally people concede for either not seeing how they can still win (i.e. acknowledging the loss) or because of bad attrition-avoidance (regardless if they would win) for future matches.

Toward that end, is concession correlated with anything what we're talking about? Team freshness or TV disadvantage?

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

a) Even if the correlation between new teams and non-new teams for >300TV is comparable, wouldn't reducing that advantage reduce the number of losses for fresh teams and thus the number of teams being restarted?

It actually ends up giving an advantage to new teams, which can just as easily be said to encourage people to restart teams early since the TV difference variance would go up over time. Keep in mind this is new teams not new coaches we're talking about. If you know things are just going to get harder for you, will you be more or less likely to keep a new team if it loses its first or second game?

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

b) Is it really prudent to ignore concessions for the winrate of fresh teams? Disconnections aside, we still have 20 to 25% concessions in CCL and (I assume) normally people concede for either not seeing how they can still win (i.e. acknowledging the loss) or because of bad attrition-avoidance (regardless if they would win) for future matches.

Their own concessions, yes. The point is that new teams that actually play out their games are producing a better than 50% win rate. We can't say why people "normally" concede, but we can say that when they don't, they're certainly not losing, on average, more than they're winning.

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

Toward that end, is concession correlated with anything what we're talking about? Team freshness or TV disadvantage?

I don't see that as relevant. Concessions are based on the conceding coach's perceptions, and those are not objective things. If they were we'd be busy figuring out how to fix that "broken" RNG that people complain about regularly.

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

It actually ends up giving an advantage to new teams, which can just as easily be said to encourage people to restart teams early since the TV difference variance would go up over time. Keep in mind this is new teams not new coaches we're talking about. If you know things are just going to get harder for you, will you be more or less likely to keep a new team if it loses its first or second game?

It's always more encouraging to restart if you lost than if you won games, of course, no matter how much harder it will get later. If you won the first games, there's no reason to restart and if you have played a lot of games, it's often too expensive to restart (due to lack of time to complete the necessary number of games).

If you have lost the majority of the first few games and they had a lot of TV gaps against you, you can hope that another fresh team will have more luck in match-making and meet more TV-balanced opponents in the first few matches. If that hope doesn't pan out, you either repeat that process or in the end choose the most promising team to continue (if you're rational, that is).

Of course, that is all only relevant for those who actually play to compete, but I guess the whole win-rate thing relates to them only anyway.

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

Their own concessions, yes. The point is that new teams that actually play out their games are producing a better than 50% win rate.

It could very well be that the win-rate of the non-conceders is so 'high' because the conceders concede those games that they would have lost with a very high probability. Of course, they will mostly be new coaches and they will do it only a number of times, but still, that is a number that shouldn't be just disregarded.

We can't say why people "normally" concede, but we can say that when they don't, they're certainly not losing, on average, more than they're winning.

Of course, we can only observe at what point in games people concede. Eyeballing this from the concessions I had against me, there are very clear patterns when it happens. Of course, the eyeballing just helps to give you an idea for what pattern to look for in the data, if you were really interested. Most of the people that I have met were even so polite to (at least) actually press the concede button and not simply disconnect.

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

Concessions are based on the conceding coach's perceptions, and those are not objective things.

It has to do with the skill of evaluating a game situation and seeing it as (mostly) hopeless or risky to continue. That is definitely something you can learn with experience over time. But, of course, you could only do that with an experiment where you'd need to somehow give an incentive to people to not concede even if they normally would have and then record whether their prediction was accurate or not.

For less skilled players that see that they are being outplayed by a more skilled opponent, it's probably even an accurate perception that they will lose the game most of the time.

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

It's always more encouraging to restart if you lost than if you won games, of course, no matter how much harder it will get later. If you won the first games, there's no reason to restart and if you have played a lot of games, it's often too expensive to restart (due to lack of time to complete the necessary number of games).

It honestly seems like you're saying nothing at all now. We can guess at what people's mindsets will be, but the numbers don't indicate that any significant differences exist between new teams or non-new teams related to TV differences. We could throw wins at new teams to pretend they deserve them, but this is a multiplayer game and match outcomes are a zero-sum thing, so... make it better for one person involves making it worse for someone else.

I read data, not minds. I'll leave that for you to imagine you're doing.

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

It could very well be that the win-rate of the non-conceders is so 'high' because the conceders concede those games that they would have lost with a very high probability.

It could also be because of sun spots, or taco tuesdays. Concession is based on perceptions. Period. How accurate or inaccurate those perceptions are is a different matter, but it's something we non-mind-readers cannot measure. What we can measure is what happens when people do NOT concede. Anything else is dross.

@ugh said in For the Nth time - LOWER TV Diff - Even optionaly:

Eyeballing this from the concessions I had against me....

Zzzzzzzzz.... sorry, I passed out at this point. Your eyeballing, anecdotes, and interpretations of such are worthless forum masturbation that nobody but you will ever base anything on. NEXT.

last edited by VoodooMike

Agree 100% Yogi! I would like a slider from 100-1000 to give player choice on the TV difference they are willing to suffer.

All this data about the less games that would be matched is all well and good. But ignoring people who aren't playing due to this terrible matchup system is just silly. And comparing these complaints to complaints about RNG is ludicrous.

Maybe as the steam numbers / CCL games drop, Cynanide will eventually have a re-think. That's what I hoping for! In the meantime, things will never change, especially while there are such strong "influential" opinions for TV+ "handicapping" system which requires the larger uncapped TV differences to work...

BB2 Champion Ladder Admin Team

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

ignoring people who aren't playing due to this terrible matchup system is just silly.

How many people are being ignored, exactly?

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

All this data about the less games that would be matched is all well and good. But ignoring people who aren't playing due to this terrible matchup system is just silly. And comparing these complaints to complaints about RNG is ludicrous.

It's silly to ignore numbers we don't have and, in fact, can't have? Well then, first lets address the number of people who aren't playing due to there being no poodle roster... and then lets address the number of people who aren't playing due to the fact that there aren't enough black bretonnians... and then...

You quite literally cannot pay attention to unobtainable data. People who tell you that you should are just trying to pretend their opinions are data driven... but driven by data nobody (but them) can ever see. woooOOooooOOoooo spooky magic BS!

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

Maybe as the steam numbers / CCL games drop, Cynanide will eventually have a re-think.

Rethink what, exactly? Other than waving your hands around in the air, are you able to link those decreasing numbers to any specific issue? Unless you can, you're just blowing smoke up people's asses and doing exactly what I mentioned in the previous statement: trying to BS in a link between real phenomenon and imagined causes. Do you think they'll be motivated to change things the way you randomly want them to because of that, or do you think they're more likely to change things based on real data-driven evidence?

No need to answer that - I know you think they'll just "come to their senses" and see things your way based on nothing.

@dode74 said in For the Nth time - LOWER TV Diff - Even optionaly:

How many people are being ignored, exactly?

I hear its about the same number as those who don't like TV+ in a competitive league. 🙂 Was glad to hear you say you are not married to TV+ on the AndyDavo stream though! I did note that TV+ went down like a lead balloon with that stream audience too though...(like it did on reddit).

Mike, I think the developers at Cyanide are smart enough to distinguish between a broken match up system where TV1000 Orcs can be matched against 1500TV cpomb Chorfs (for a potentially new player in their first ever online experience) versus other factors like lack of poodle teams and ethnic diversity.
You may also want to consider to switch analyzing data from Season 3 to something more recent. Chorfs didn't even exist in season 3...

BB2 Champion Ladder Admin Team

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

I hear its about the same number as those who don't like TV+ in a competitive league. 🙂

So you've no idea. Cool. Your narrow view on how competitions must necessarily work are noted, though 😉

Was glad to hear you say you are not married to TV+ on the AndyDavo stream though! I did note that TV+ went down like a lead balloon with that stream audience too though...(like it did on reddit).

Nobody with any honesty should be married to any system, just what the data shows us. You may have been watching a different stream, though: I saw a split opinion among the audience. Certainly there were some vocally against, but loud != correct.

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

I hear its about the same number as those who don't like TV+ in a competitive league. 🙂

Really? How many is that? How many of them are aware of what the data says about the resulting differences in MM under TV matchmaking and TVPlus matchmaking versus their random assumptions? I'm betting precious few. You, for example, probably don't have the foggiest...

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

Mike, I think the developers at Cyanide are smart enough to distinguish between a broken match up system where TV1000 Orcs can be matched against 1500TV cpomb Chorfs (for a potentially new player in their first ever online experience) versus other factors like lack of poodle teams and ethnic diversity.

Beg your points much? Again, this is another case of your feelsies not matching the data - while such matches are possible they are unlikely and indeed rarely happen. Ironically, such matches are far more balanced under TVPlus matchmaking where the effects of TV difference are basically halved as compared to TV matchmaking. But of course... facts aren't important... lets talk about what you like.

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

You may also want to consider to switch analyzing data from Season 3 to something more recent. Chorfs didn't even exist in season 3...

Do you seriously believe that is going to change those relationships? I used season 3 because I'd already calculated all the relevant fields for it. Feel free to hit season 10 and see if you find something different... oh wait... you're not about data, you're about feelings.. maybe wave a crystal over season 10's data and see how it affects its aura? The more bro-crystals you get involved the more accurate the auras will be, I'm sure.

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

Do you seriously believe that is going to change those relationships?

No, of course not. The hatred for TV differences is not contained within the game data. But in your season 3 data, the unbalanced chorf match ups are definitely not going to be happening...

How rare an event is, is different from the perceived risk of it happening. If someone told me a I could stroke a tiger with a 5% chance of being mauled to death (entering a fresh team into CCL) or stroking a tabby cat (table top BB with equal TV or other board games). The latter is becoming more appealing, particular with only having time to stroke one cat per night. Why risk getting mauled, when other more fun options are available? I really struggle to understand why you cannot see that point of view. Maybe its because you have never played in CCL and had it happen to you?

This is a suggestion thread, and my suggestion remains:

"ADD A TV + OR - SLIDER TO MATCH MAKING SO A COACH CAN CHOOSE THE MAXIMUM TV DIFFERENCE TO SUFFER".
People at high TV can make it + or - 1000 to improve their matching chances for those coaches who don't mind the challenge...

At the end of the day, its up to Cyanide, not you...

BB2 Champion Ladder Admin Team

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

This is a suggestion thread, and my suggestion remains:

"ADD A TV + OR - SLIDER TO MATCH MAKING SO A COACH CAN CHOOSE THE MAXIMUM TV DIFFERENCE TO SUFFER".
People at high TV can make it + or - 1000 to improve their matching chances for those coaches who don't mind the challenge...

At the end of the day, its up to Cyanide, not you...

My objection to putting it in a competitive division remains.

But you're right: it is up to Cyanide. And they have an aversion to UI changes.

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

No, of course not. The hatred for TV differences is not contained within the game data. But in your season 3 data, the unbalanced chorf match ups are definitely not going to be happening...

How rare an event is, is different from the perceived risk of it happening.

Haha, so our concern isn't whether there's a problem or not, but with changing the game based on people's incorrect perceptions? Folks like you remind me of the old Dara O'Briain routine about homeopathy...

https://www.youtube.com/watch?v=DHVVKAKWXcg

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

Why risk getting mauled, when other more fun options are available? I really struggle to understand why you cannot see that point of view. Maybe its because you have never played in CCL and had it happen to you?

You risk getting mauled every time you play the game - that's just part of base Blood Bowl. I do play in CCL, I just don't play under this name... though it wouldn't much matter if I didn't - it wouldn't change the data.

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

At the end of the day, its up to Cyanide, not you...

Of course it's up to Cyanide, but I'm going to continue providing them with the facts so that their decisions can be better informed. So far you've provided zero evidence to support your idea... but who knows? Maybe they'll say "screw data and facts, we've just gotta get on board with @runningdragon!".

My own experience of tv differences is that it's not too bad being down tv in the first few games because:

a) A significant portion of that TV is often fan factor, which often isn't worth its value to the higher tv team.

b) Even with a TV advantage, the teams you are facing are maybe around 1400-1500 TV max and may have weaknesses that you can exploit with inducements.

c) In the worst case you can restart without losing huge progress.

In my experience the worst time to be down TV is when you are in the mid stage of development (1300-1600 for most teams).
If you are down TV at this range then you are likely facing a very tasty opponent with a well rounded team while you are still missing some important skills. Here the inducements are very weak at fixing your mechanical weaknesses, you are also likely around capped out on fan factor yourself, so the TV differential is all real skills. You also really don't want to chuck away your progress.

Edit: The issue was resolved and it is working now (see following posts)

Just as a small follow up to our last discussion (mean tv-difference different to 0), I tried to reproduce your numbers, but I can't. Could you check, where the differences are or what I am missing?

For the sake of similarity, I will just look at this value now:

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

Beyond the 300 TV difference

Correlation between TV advantage and match outcome: r = 0.139, p < 0.01

The first steps I did:

  1. I downloaded the data base for CCL-S3 from goblinspy
  2. I exported the table teammatches as csv and read it into R (variable teammatches)

The dimensions of the variable are here:

dim(teammatches)
[1] 37834 85

I then filtered out draw games and games, with an absolute tvdifference smaller than 300

teammatches = teammatches[

teammatches[,"draw"] == 0 &
teammatches[,"tvdiffabs"] >= 300

,]
dim(teammatches)
[1] 4492 85

The correlation (pearson) between tv-difference and outcome (win) is then:

cor(teammatches[,"tvdiff"], teammatches[,"win"])
[1] 0.3545435

In this case every game is counted twice (both directions). To only include every game only once, I looked at every second row (2,4,6,...) in the table teammatches:

index = (1:(nrow(teammatches)/2))*2
sum(teammatches[index,"id"] != teammatches[index-1,"id"]) #(just a double check for this apporach)
[1] 0
cor(teammatches[index,"tvdiff"], teammatches[index,"win"])
[1] 0.2648016
mean(teammatches[index,"tvdiff"])
[1] -226.5093

I wanted to try out other things on how to select the directions. The next attempt was to only look at the perspective from the team with the higher TV

filter = index - 1*(teammatches[index,"tvdiff"] < 0)
cor(teammatches[filter,"tvdiff"], teammatches[filter,"win"])
[1] 0.07901165
mean(teammatches[filter,"tvdiff"])
[1] 389.3945

In the last attempt, I choose randomly which direction to pick

filter = index - 1*(runif(nrow(teammatches)/2)>0.5)
cor(teammatches[filter,"tvdiff"], teammatches[filter,"win"])
[1] 0.3544961
mean(teammatches[filter,"tvdiff"])
[1] 2.831701

-> The resulting correlation heavily depends on direction which was choosen. Based on our last discussion, I know that you only look at one direction and that you have somehow a systematic of choosing that direction (your mean tv-difference is different from zero). So my question would be: How do you choose the direction you look at? Otherwise: Do I have some mistakes/errors or something else in my script?

last edited by Arne

@arne said in For the Nth time - LOWER TV Diff - Even optionaly:

I then filtered out draw games and games, with an absolute tvdifference smaller than 300

There's your problem. There's no reason to remove draw games in this case - we only need to do that when we're checking how many games we accurately predict the winners of, since draw games have no winner and thus don't really tell us anything about how well a metric predicted that match. The metrics don't try to predict draws.

@arne said in For the Nth time - LOWER TV Diff - Even optionaly:

In this case every game is counted twice (both directions). To only include every game only once, I looked at every second row (2,4,6,...) in the table teammatches:

Also a problem. You should leave the matches in. Yes, it potentially counts each match twice, but only if both teams counted as being in that category (less than 5 games, more than 5 games). It seems like you missed the part where I point out we're talking about "new teams" only when dealing with the "beyond the 300 TV difference".

Dropping every second row is a bad way to do things in either case.

@arne said in For the Nth time - LOWER TV Diff - Even optionaly:

Based on our last discussion, I know that you only look at one direction and that you have somehow a systematic of choosing that direction (your mean tv-difference is different from zero). So my question would be: How do you choose the direction you look at?

I don't know what you mean by that. I'm generally NOT using only one datapoint per match unless we're talking about match effects that are isolated from their relationship to team specifics - there are two teams in each match, and each team has its own outcomes and data. Not considering both outcomes is ignoring half the data.

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

@arne said in For the Nth time - LOWER TV Diff - Even optionaly:

I then filtered out draw games and games, with an absolute tvdifference smaller than 300

There's your problem. There's no reason to remove draw games in this case - we only need to do that when we're checking how many games we accurately predict the winners of, since draw games have no winner and thus don't really tell us anything about how well a metric predicted that match. The metrics don't try to predict draws.

Fair enough, how did you include draws? did you do it like this:
loss: 0
draw: 0.5
win: 1
?

I still can't reproduce your results if I include the draw games like this.

@voodoomike said in For the Nth time - LOWER TV Diff - Even optionaly:

I don't know what you mean by that. I'm generally NOT using only one datapoint per match unless we're talking about match effects that are isolated from their relationship to team specifics - there are two teams in each match, and each team has its own outcomes and data. Not considering both outcomes is ignoring half the data.

We have some serious communication issues, I could have sworn, that you critized me last time for using 2 data points per match. Anyway, I am happy, that it is an non-issue then.

edit: did you use the pearson correlation?

edit2: silly me, I overlooked something, forget what I wrote first.

last edited by Arne

Remark: I was able to reproduce it now, thanks for your clarification.

@arne said in For the Nth time - LOWER TV Diff - Even optionaly:

We have some serious communication issues, I could have sworn, that you critized me last time for using 2 data points per match.

No, I criticized you for using both copies of the match to calculate mean rating advantage, since that meant every match contributed a net variance of ZERO... if one team played at +300 TV, the other necessarily played at -300 TV, which means that taking the average of those numbers always resulted in 0. That isn't a problem that extends to all things, only for taking mean values of mirrored metrics.

@arne said in For the Nth time - LOWER TV Diff - Even optionaly:

Remark: I was able to reproduce it now, thanks for your clarification.

Glad to hear it.

@voodoomike
lol. That video was particularly funny since fear of (dirty player) zombies is actually relevant to my point.
I'm not sure what homeopathy has to do with the FACT that many people do not want 300-500 TV down matchups, nor the FACT that some people have stopped playing the game due to this. But that brings us back to your poodle argument, so I'll stop there...

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

That video was particularly funny since fear of (dirty player) zombies is actually relevant to my point.

It shouldn't be. You're talking about people's fear of something that there's no evidence exists... the comedy routine makes fun of people who counter evidence with people's feelings to the contrary.

@runningdragon said in For the Nth time - LOWER TV Diff - Even optionaly:

I'm not sure what homeopathy has to do with the FACT that many people do not want 300-500 TV down matchups, nor the FACT that some people have stopped playing the game due to this.

How many people don't want it? How many people stopped playing due to it? Without metrics you're just waving your hands around in the hopes people will be confused by all the smoke and mirrors and think you're more than one guy.

Looks like your connection to Focus Home Interactive - Official Forums was lost, please wait while we try to reconnect.