Hello community,
I know this subject has already been mentioned, but it's extremely painful to be "afraid" of touching the settings or making changes for fear of getting a goal.
To back up what I'm saying, I've played my last 3 games against big guns. I'd like to make it clear that my defeat in these games was logical, even if I could have hoped for a different outcome in one of them.
1- #match?mid=6764987 : I take a goal in my changes
2- #match?mid=6794108 : I take a goal when I consult my opponent's 'style stats'.
3- #match?mid=6795352 : I take a goal 7mn after my changes

Do I have to play AFK like a big shot? If every time I touch a parameter or make a change, I take a goal, I might as well go and play tennis. At least I won't be wasting my time.
It's weird to lose stats when you're consulting, and when you're making changes, you bring in freshness and sometimes strength, and that's fatal.

Player changes that lower the stats... such a logical setting... It's as if the substitutes don't train with the starters every day... Unless they have a memory problem and it takes 10 minutes to recognise their team-mates... Anyway, it's exciting to be active in matches 🤣

Every now and then, an opponent's change leads to a goal, and then, through bad luck, it's fallen to you, 3 times in a row.

Looking at your opponents, yes, it could have happened to anyone.

Don't be fooled, a change leads to a drop in stats for you and your opponent after a while and the 'compensation' comes in between.

Don't denigrate your matches with subjects like that, you'll see that you'll do the same thing in some of your next matches 😉

Respectively 6th, 10th and 13th in VF...

This drop in stats may be understandable in some cases. In others it is not.

A coach changes everything, which makes sense in terms of the potential negative impact.

In the final minutes, when you have to go all out to win or come back from a goal down, the drop in stats does the opposite...

I'm not questioning the quality of my opponents, who deserve their victory. However, I don't understand the circumstances that made me concede.
With the same strength, I don't have this problem (or a lot less), but here I've played 3 games when I was far inferior on paper (the fact that I played away 3 times didn't help either). It was tactically nice to put up some resistance, but it's a shame that the goals came in a strange way, each time at the bottom of my stats... We're not allowed to touch anything any more when we're playing big managers?

Basically, your complaint is understandable.

As for the form, you said it yourself: 3 very big opponents, and what's more, playing away from home, a change that leads to a smaller drop in stats and boom, you pick up 1.

It's not a question of moving or not moving, amha, it's a question of knowing if you need to and if they move ;)

More and more matches give an impression of inconsistency. The game has always had an element of randomness, and that's normal. Nobody wants the favourite to win 100% of the time. But before, this element of randomness was generally acceptable. Today, it seems to have gone well beyond that limit.
We are seeing more and more results that are very difficult to explain, even by looking at the statistics, the xGs or the balance of power between the teams. Where once a match might have produced a slightly lucky goal, it now seems that this type of situation is repeated far too often. Innocuous actions become decisive, major gaps between two teams are no longer sufficiently noticeable, and matches are becoming less and less reliable.
In my opinion, the problem lies mainly with the balancing of the match engine. There were probably two ways of responding to criticism of the lack of goals in big games, particularly in the Champions League:
Either increase the gap between small, medium, good and very big chances.
In this case, the best teams would have been rewarded more, but the gaps between the highly optimised clubs and the others would also have been more visible.
In other words, to increase the probability of conversion more generally, without necessarily accentuating the differences between the different types of chances.
This second option seems to have been chosen. The problem is that increasing the probabilities without sufficiently reinforcing the hierarchy between chances and teams mechanically increases the weight of chance.
The result is matches that seem much less readable, much less controllable, and sometimes even frustrating for those who are really interested in the competition.
This is not a complaint after a bad run or a few poor results. Losing is part of the game. So is playing a bad match. But stringing together a string of very surprising results, match after match, in all the agreements and all the competitions, gets tiring. Many of the players involved have noticed.
The CPs have already done a lot of work analysing VF's results, but there hasn't really been a clear answer. Yet there is plenty to discuss. When teams with an average of 250 points face off against teams with 200 or 210, the difference is not insignificant. On paper, it represents a huge difference. And yet, sometimes that difference seems to matter a lot less than it used to.
Of course, a strong team on paper is not enough. You also have to optimise your match statistics, your tactics, your physical strength and your choices. But still: at some point, differences in level, preparation and investment have to have some real weight in the results.
The aim of this message is not to attack the MDJ or to denigrate the work done on the game. The aim is to say that a significant proportion of competitive players today feel a loss of consistency. And if this feeling is shared by so many people, it at least deserves to be heard.
VF needs randomness, yes.
But above all it needs acceptable, legible and consistent randomness.
Today, many people feel that this balance has been upset. And it's precisely this point that we need to be able to discuss seriously with the MDJ.

Drosko: Je ne remets pas en cause la qualité de mes adversaires qui méritent leur victoire. Néanmoins, je ne comprends pas les circonstances qui me font encaisser.
A force égale, je n'ai pas ce problème (ou bcp moins), mais là j'ai joué 3 matchs en étant largement inférieur sur le papier (le fait de jouer 3 fois à l'extérieur ne m'a pas aidé non plus). C'était sympa tactiquement de faire de la résistance mais dommage que les buts arrivent franchement de manière bizarre, à chaque fois à la baisse de m...

But your opponents weren't moving?
If so, what's the problem? You were both under malus... and if one of you took the risk of not moving, despite the other's movements, his (risky) choice gave him a little extra stats. It's a choice you have to make! This is also a game of choices...

As for IRL parraleles, changing a player certainly brings fresh blood, but it also disrupts the symphony of a team. It takes a little time for the substitute to find the right rhythm for the match, apply the instructions and so on. The same goes for the rest of the players: if the instructions change, it's not surprising that they need time to adapt.

PS: I'm just providing some counter-arguments for the discussion.

@ Woodz, even if we can agree in part with the content of your "pavement", it's completely off topic!!!!

The subject is, in particular:
Why, for example, if I just want to visualise the impact of my opponent's instructions over the space of 5 seconds, does it cost me 100 stat points for a good 10 minutes?
After that, whether that translates more or less automatically into a goal as Drosko suggests, you could say that it's an unfortunate coincidence of circumstances.

But the fact is that we can understand his annoyance because during the match he quotes
#match?mid=6795352
I was his opponent and we were exchanging PMs and he told me that he was afraid of getting a goal by making changes.

Apart from the impact settings, which need to be reviewed, at the very least the impact of the opponent's instructions needs to be displayed.
Alternatively, you might as well display nothing, but setting up an invisible display that's only visible if you sacrifice 100 points of statistics is the height of illogic.

Woodz: De plus en plus de matchs donnent une impression d’incohérence. Le jeu a toujours eu une part d’aléatoire, et c’est normal. Personne ne demande que le favori gagne 100 % du temps. Mais avant, cette part d’aléatoire restait globalement acceptable. Aujourd’hui, elle semble avoir largement dépassé cette limite.
On voit de plus en plus de résultats très difficiles à expliquer, même en regardant les statistiques, les xG ou les rapports de force entre les équipes. Là où un match pouvait parfois pro...

In the end, the analysis showed that the main change from last season was to allow bigger xG per act for strong dominations, which was a request from many people here. But that's about the only visible change that's been made.

There's no less or more consistency than before. Not all opportunities are equal in the sense that greater dominance will give you more xG (more opportunities/goals).

There was one other thing: it was shown that below 0.25 xG generated (whether 0.2; 0.1; 0.001...), the chances of scoring are blocked at 0.25.

Example: if two opponents both finish a match at 1.2 xG. So with the SAME xG

• opponent 1 had 12 chances at 0.1 xG
• opponent 2 had 5 chances at 0.24 xG
  ... then the most likely score is ... 3 - 1!

But that's still not the point. 😅

Misha: Y avait quand-même un autre truc: on a montré qu’en-dessous de 0,25 xG générés (que ce soit 0,2; 0,1; 0,001…), les chances de but sont bloquées à 0,25.

Exemple: si deux adversaires finissent tous les deux un match à 1,2 xG. Donc avec la MÊME xG

• que l’adversaire 1 a eu 12 occasions à 0,1 xG
• que l’adversaire 2 a eu 5 occasions à 0,24 xG
  … et bien le score le plus probable est … 3 - 1 !

Mais en effet, c’est toujours pas le sujet. 😅

Chances of scoring?

At 0.01 xG you have a 1% chance of scoring, not 25%.

@ Misha, Galywat
You too are out of line.

It's a topic but not the one Drosko is talking about! :) :) :)

myforsans: @ Misha, Galywat
Vous aussi vous êtes hors sujet.

C'est un sujet mais pas celui qu'évoque Drosko ! :) :) :)

Yeah my bad, on the app I only saw Woodz's reply and replied to it without bothering with the original topic.

On this subject, I agree with Magpie.

Galywat: Au final l'analyse a surtout montre que le changement principal de la saison passée c'était de permettre de plus grosses xG par actu pour de fortes dominations, ce qui était une requête de beaucoup ici. Mais c'est a peu près le seul changement visible qui a été fait. (Et une tentative début de saison passée sur le même principe mais avec une limite haute mal calibrée ?)

Il y a pas moins ou plus de cohérence qu'avant. Toutes les occasions ne se valent pas dans le sens où une plus forte domina...

I don't understand how you can be so adamant about this.

What the analysis of the CPs shows is precisely that there doesn't seem to be a clear enough hierarchy between chances with a very low probability of scoring and much clearer chances.

In my opinion, there has been a fundamental change to increase the number of goals. But from what I've observed, this increase hasn't come from the top down, by giving more value to the really big chances. Instead, it seems to have come 'from below', by making situations that should remain relatively innocuous much more dangerous.

And for me, the main problem is not the loss of statistics linked to tactical choices. Most of the time, when you adjust your positioning to improve it, your opponent does the same thing on his side. So the two teams often take a comparable penalty.

The real issue isn't just the loss of stats. Above all, it's about the consistency between the real quality of a chance and its likelihood of ending up as a goal.

Woodz: Je ne comprends pas comment tu peux être aussi catégorique sur ce point.

Ce que montre l’analyse des CP, c’est justement qu’il ne semble pas y avoir de hiérarchie suffisamment claire entre des occasions avec une probabilité de but très faible et des occasions beaucoup plus franches.

À mon sens, il y a bien eu une modification de fond pour augmenter le nombre de buts. Mais d’après ce que j’observe, cette hausse ne s’est pas faite “par le haut”, en valorisant davantage les très grosses occasi...

(sorry for the out of order)

Because, like Misha, you've got the problem backwards. Who cares about goal-scoring opportunities?

Your stats and those of your opponent will allow a roll of the dice to determine who has the initiative on an action and on which side it takes place, and the associated xGs.

Based on these xGs :

• xG = % chance of scoring

• % chance of having an opportunity = xG x 3 (obviously above 0.25 it decreases because you can't have more than a 100% chance of having an opportunity or a goal).

• % chance of having no action on the ball = 100% - xG *4 (obviously if you go over 0.25 then you're sure to have either a chance or a goal)

But it's not an opportunity that determines the xGs, but the xGs that give you the % chance of having a goal (and therefore an opportunity by ricochet, but you could almost do without it, it's just there for decoration, technically).

Drosko: Hello la communauté,
Je sais que le sujet a déjà été évoqué, mais c'est ultra pénible d'avoir "peur" de toucher aux paramètres ou faire des changements de peur de se prendre un but.
Pour étayer mes propos mes 3 derniers matchs contre des gros bras. Je précise que ma défaite sur ces rencontres reste logique même si j'aurai pu espérer un autre sort sur l'un des matchs.
1- #match?mid=6764987 : je prend un but dans mes changements
2- #match?mi...

we need a little help from the mdm when we put our children to bed 🤷‍♂️

On a more serious note, the question of the drop in stats is always a subject. There's a logic to it that I don't totally disagree with, but at the same time the strength of the drop seems a bit excessive.
It's all the more penalising against stronger teams, but personally, I see a consistent logic in the way I follow my games: when I'm on, I win more often than when I'm not.
I also don't see goals conceded 'often' when I make a change, even though I know that's a reaction that some managers have on a regular basis.
After that, I'm lucky enough to have a team whose stats are more often than not superior to my opponents over the length of the match, so maybe that reduces this type of misfortune?

Galywat: (désolé pour le HS du coup)

Parce que, comme Misha, tu prends le problème à l'envers. On s'en fout des occasions de buts et du ratio but/occasion, c'est une simple loi mathématique qui lie les buts et occasions.

Tes stats et celles de ton adversaire vont permettre un jet de dés qui détermine qui a l'initiative sur une actu et de quel côté cela a lieu et des xG associés.

à partir des ces xG, grosso modo c'est :

• xG = % de chance de marquer
• % de chance d'avoir une occasion = xG x 3 (fo...

I think you're oversimplifying the problem by considering that "it's just a mathematical law" and that, therefore, everything is necessarily coherent.

The problem isn't denying that the game works with probabilities. Of course it's an algorithm, of course there are rolls of the dice, of course the xGs are used to determine the probability of a goal. On that, we agree.

But saying "it's mathematical" doesn't mean "it's well balanced".

An algorithm can be perfectly logical in its internal workings, while producing inconsistent results if the values used are badly calibrated.

It's like a recipe:
if you add twice as much salt, the recipe remains 'logical'. You've followed the steps, you've stuck to a method. But the end result is a failure because the dosage is wrong.

For me, that's where the current problem lies: it's not the principle of calculation that's necessarily wrong, it's the proportioning of probabilities.

When you say that chances are 'decoration', that's precisely the problem. In a football game, a weak, medium or very strong chance should have a real difference in weight. If everything is reduced to an xG that mechanically gives a chance of a goal, then the hierarchy between the types of action needs to be extremely well regulated. Otherwise, you're creating a reading illusion.

The player sees an opportunity, a balance of power, statistics and xG, but has the impression that the real quality of the action doesn't count for enough.

Another simple parallel:
if you have a 100-sided die, the problem isn't whether the die exists. The problem is knowing how many sides you assign to each result.

If a small action has 3 winning sides and a big opportunity has 8, the difference exists, but it may be too small. And if you increase all the chances of scoring without widening the gap between the small and large chances, then you're mainly increasing the weight of the 'easy' goals, and therefore the weight of chance.

That's what many people are criticising at the moment.

You don't say: "I'm losing, so the engine is bad".
We're saying: "The gaps between weak, average and strong situations no longer seem wide enough to produce a coherent reading of the matches."

Your example of 12 chances / 1 xG versus 3 chances / 1 xG is interesting. Yes, mathematically, if both teams finish with 1 xG, the expected score is close. But the question is: how do we get to that 1 xG? Should 12 small situations really be put on the same level as a small number of more straightforward actions? Are the construction of the match, the dominance, the difference in stats and the quality of the actions correctly reflected in this total?

That's where the debate lies.

Saying "xGs determine everything" doesn't answer the problem. It simply displaces it. The real question becomes: is the way xGs are generated consistent with the actual balance of power between the teams?

Because if the algorithm assigns xGs too easily to weak situations, or if it gives too many chances of goal to actions that are not very dangerous, then the final result may be mathematically valid, but sportively inconsistent.

Nor should we lose sight of another important point: the match engine does not appear to analyse the 90 minutes continuously. It only takes a certain number of samples, often between 14 and 30 really decisive actions or updates in a match, whereas a match lasts 90 minutes.

And that changes a lot of things.

If the engine only really 'sees' a limited part of the match, then there's bound to be an information deficit. The minutes that are not taken into account do not have a direct impact on the way the match is read. In other words, a team can dominate, position itself well and control many phases, but if this domination does not fall at the right moment in the samples actually used by the engine, it may be rewarded very little.

It's like judging an entire film by looking at just 15 or 20 random scenes.
You may come across some important scenes, but you can also miss out on everything that gives meaning to the story.

Or like assessing a pupil over a whole year by correcting only a few exercises chosen at random. The result may give an indication, but it doesn't necessarily reflect all the regularity or all the real mastery.

That's why the problem isn't just the xG/goal/opportunity ratio. You also have to look at how these xGs are generated, at what times they are generated, and how many samples the engine uses to construct its result.

The smaller the sample, the greater the weight of randomness.
And if, in addition, the conversion probabilities have been increased, then each sample becomes even more decisive.

This is where we can have a snowball effect: few actions really taken into account, higher probabilities, a lack of hierarchy between small and large opportunities, and ultimately results that become much less legible.

It's exactly like a GPS:
if it calculates the route perfectly from bad data, it can still send you into a wall. The calculation is right, but the starting point is wrong.

So no, the debate is not "are there probabilities?
Of course there are.

The debate is: are these probabilities correctly balanced so that the game remains readable, coherent and acceptable?

And today, many players feel that this is no longer the case.

Woodz: Je pense justement que tu simplifies trop le problème en considérant que “c’est juste une loi mathématique” et que, donc, tout est forcément cohérent.

Le souci, ce n’est pas de nier que le jeu fonctionne avec des probabilités. Bien sûr que c’est un algorithme, bien sûr qu’il y a des jets de dés, bien sûr que les xG servent à déterminer une probabilité de but. Là-dessus, on est d’accord.

Mais dire “c’est mathématique” ne veut pas dire “c’est bien équilibré”.

Un algorithme peut être parfaite...

What you write is factually incorrect. There is precisely a difference in the weight of each action, it's just not directly visible. The only tangible thing is that you can only see it through the prism of xGs and not chances or goals.

If you download the "Match Actions" file from #matchs?export you'll be able to see that for each visible action (opportunity/goal), the xGs are roughly correlated to the stats ratio of the sector concerned. Example for a given action:

You have the initiative in the game and the game has determined that you will play on the left (globally according to probabilistic laws based on all the stats for a given news item in the 9 sectors of the game).

If your stat is 60 and your opponent's is 40, your xG for this event will be around 0.21-0.24.
If your stat is 40 and your opponent's is 40, your xG for this event will be around 0.1.

There's a direct link here and I invite you to look at this file if you want to see for yourself.

Woodz: Je pense justement que tu simplifies trop le problème en considérant que “c’est juste une loi mathématique” et que, donc, tout est forcément cohérent.

Le souci, ce n’est pas de nier que le jeu fonctionne avec des probabilités. Bien sûr que c’est un algorithme, bien sûr qu’il y a des jets de dés, bien sûr que les xG servent à déterminer une probabilité de but. Là-dessus, on est d’accord.

Mais dire “c’est mathématique” ne veut pas dire “c’est bien équilibré”.

Un algorithme peut être parfaite...

Make us a little Youtube video steuplaît 😁

Galywat: Ce que tu écris est factuellement faux. Il y a justement une différence sur le poids de chaque action, elle n'est juste pas visible directement. La seule chose tangible, c'est que tu ne peux le voir que par le prisme des xG (et donc pas pendant le match) et pas par les occasions ou buts.

Si tu télécharges le fichier "Actions de matchs" de #matchs?export tu pourras voir que pour chaque action visible (occasion/but), les xG sont grosso modo corrélés au rapport de st...

But when there are situations like that throughout the match, what conclusion do you want to draw?
Over-domination ++ throughout the match but never a chance generated, so weak xG
I don't mind if you don't get the roll several times, but given that there are between 14 and 30 samples per match, that's a lot of bad luck!

image](https://i.imgur.com/dcv9iVS.png)

Galywat: Ce que tu écris est factuellement faux. Il y a justement une différence sur le poids de chaque action, elle n'est juste pas visible directement. La seule chose tangible, c'est que tu ne peux le voir que par le prisme des xG (et donc pas pendant le match) et pas par les occasions ou buts.

Si tu télécharges le fichier "Actions de matchs" de #matchs?export tu pourras voir que pour chaque action visible (occasion/but), les xG sont grosso modo corrélés au rapport de st...

Yes, I understand what you mean, and I don't deny that the xGs are correlated to the stats reports in the action file.

But my point isn't that there's no link between stats and xG. My point is that this link alone does not guarantee sufficient consistency in the final result.

Even if an action at 60 to 40 gives more xG than an action at 40 to 40, the real question remains: is the gap big enough to properly reflect dominance?

Because an algorithm can very well incorporate a difference without that difference being sufficiently discriminating in practice.

A simple example: if one student has 18/20 and another has 10/20, but the system transforms that into 55% success rate compared with 45%, there is a difference. It does exist. But it may not reflect the real difference in levels.

So yes, the opportunities may be text. Yes, xGs are the main indicator. But that doesn't fully answer the problem: is the translation of stats reports into xG strong enough, legible enough and consistent enough to ensure that domination is genuinely rewarded?

That's where my doubt lies. Not about the existence of the mathematical link, but about its balance.

Magpie: Fais nous une petite vidéo Youtube steuplaît 😁

Get your priorities straight, we're not making any headway as it is, so if you drag the debate down we won't make it.

Personally, I think it's great as it is at the moment. The more you dominate, the more xG you generate (up to around 0.32 if it hasn't changed recently). Conversely, the more you dominate defensively, the less xG you concede.

For me, the problem, if there is one, isn't the generation of xG. It's getting the mdm where you want him to be. That's a lot more opaque.

Many people think that because you overdo it on one side, the news will happen there, when there's always the possibility that it could happen elsewhere. (The same goes for the defence).

It's a bit clichéd, but VF is an optimisation game. You try to optimise your stats according to your objectives.

Personally, I'd like it if we could always know who took the initiative and on which side, even if it means adding another type of action in the match description. It would be so much clearer for everyone.

Ced90: Mais du coup quand il y a des situations comme ça sur l'ensemble du match tu veux tirer quoi comme conclusion?
Une surdomination ++ tout le match mais jamais une occasion générée, donc xG faible
Je veux bien que plusieurs fois tu n'aies pas le roll, mais vu qu'il y a entre 14 et 30 échantillons par match apparemment, ça fait beaucoup de malchance !

It's hard to say on just this one capture, you'd have to see when the stats in question were in the news. After that, maybe it's just that the rolls aren't on the right side - that can happen, yes. It's very rare for it to happen over an entire match, even if it is possible.

Galywat: Sur juste cette capture c'est difficile à dire, faudrait voir quand il y a eu les actu les stats en question. Après peut-être juste pas les rolls du bon côté ça peut arriver oui. C'est très rare que ça arrive sur tout un match même si c'est possible. Déjà ton adversaire, si la possession est restée la même, avait 30% de chance d'avoir l'initiative à chaque actu.

I can tell you that I saw the whole match in great detail, it's the most rigged match I've ever seen ^^.

That, added to the law of fish which inevitably tends towards: xGs = goals

There are many more very open matches with big scores, and.... On the other hand, to balance things out, there are some incomprehensible matches

Ced90: Je peux te dire que j'ai vu tout le match bien en détail, c'est le plus truqué que j'ai jamais vu ^^. Et toutes les actus c'était boulevard

Ça ajouté à la loi de poisson qui, fait forcément tendre vers : xGs = buts

Il y a beaucoup + de matchs très ouverts avec de gros scores, et.... A l'inverse pour équilibrer, des matchs incompréhensibles ou il ne se passe rien (pour justement ne pas créer d'xG)

A boulevard. Well, it wasn't 10 stats at 0 in each sector. That means that the news didn't necessarily happen where you wanted it to (hence my previous post, if we could have this notion systematically, the game would no doubt be easier to read).

Fish law has nothing to do with it. There's no catching up on VF, once again you only have to look at the file used by the CPs to see that. Firstly, y xG = y % chance of scoring, that's mathematical. Secondly, there's a direct correlation between the stats in a given sector (the opportunity/goal sector, since that's all we can see) and the xG generated.

If there were a 'balance' between games, you'd see huge disparities.

Yes, so when there are 200 vs 50 in at least (for example) left midfield + left attack 10x in a row it's "just" bad luck if nothing happens. It's not punitive enough, that's all.

At some point, if almost all of VF's top 20 think the same way (yes, they play competitively every day), then there's something weird going on.

And of course there's catching up to do, otherwise it would be impossible for xG = goal in the end.
If a match is won 6/2 with a total of 1.8xg produced, there's bound to be a catch-up somewhere else... Pure logic

Galywat: Perso je trouve ça très bien tel quel actuellement. Plus tu domines plus tu générés de xG (jusqu'à environ 0.32 si ça a pas changé récemment). A l'inverse plus tu domines défensivement moins tu encaisses de xG.

Pour moi le problème, s'il y en a un, c'est pas la génération de xG. C'est de pouvoir bien amener le mdm là où on veut avoir notre action. Ça c'est beaucoup plus opaque.

Beaucoup pensent que parce que tu surdomines un côté l'actu se passera là, alors qu'il y a toujours la possibilité...

You said that what I wrote was "factually wrong", but in the end your response qualifies the subject rather than contradicting it.

We agree on one thing: the more you dominate, the more xG you generate, and the more you dominate defensively, the less xG you're expected to concede. I'm not denying this link. What I'm questioning is its balance and readability.

And the best players in the game know how to manage these probabilities very well. In fact, that's why they're among the best: they know how to optimise their stats, steer the engine, reduce the risks and maximise their chances. So when some of these players feel a loss of consistency, it's not because they've discovered that VF is a probabilistic game. On the contrary, it's because they are very familiar with these mechanisms.

Where I agree with you is on the opacity of the MDM: knowing who has the initiative, in which sector, and why the action takes place here rather than elsewhere, would already be a huge step forward in terms of legibility.

But for me, that doesn't close the debate on xG. Even if generation exists and follows a logic, the real question remains: is this logic sufficiently discriminating to reward optimisation and domination properly?

So yes, VF is an optimisation game. But that's precisely where the problem lies: if optimisation becomes less legible or less rewarded.

Ced90: Oui donc quand il y a des 200 vs 50 dans à minima (par exemple) milieu gauche + attaque gauche 10x d'affilée c'est "juste" de la malchance s'il ne se passe rien. Ce n'est pas assez punitif c'est tout

A un moment donné si quasi tout le top 20 de VF pense pareil (oui ça joue compétition tout les jours), c'est qu'il y a quelque chose de bizarre

Et bien entendu qu'il y a du rattrapage sinon il serait impossible qu'à la fin xG = but
Si un match est gagné 6/2 avec au total 1.8xg produit, il y a f...

Yes, there was undoubtedly a certain amount of bad luck. It happens very, very rarely. Most of the time it's a question of a lack of realism.

"If a match is won 6/2 with a total of 1.8xg produced, there's bound to be a catch-up somewhere else... Pure logic"

Absolutely not. It's just statistics. You roll your die 3 times and 3 times you get a 6, that doesn't mean you won't get a 6 over the next 18 moves. It's just that the more you roll the die, the more the average of that roll will tend towards 1/6. It's statistical, but it's not a mechanism for your die to catch up.

Woodz: Tu disais que ce que j’écrivais était “factuellement faux”, mais au final ta réponse nuance plutôt le sujet qu’elle ne le contredit.

On est d’accord sur une chose : plus tu domines, plus tu génères de xG, et plus tu domines défensivement, moins tu es censé en concéder. Je ne nie pas ce lien. Ce que je questionne, c’est son équilibrage et sa lisibilité.

Et justement, les meilleurs joueurs du jeu savent très bien gérer ces probabilités. C’est même pour ça qu’ils sont parmi les meilleurs : ils...

Well then, propose a different law to the one currently in force. No one has proposed anything specific, even though the stats transcription of a sector --> xG is quite legible.

Personally, I think it's fine as it stands and that you're mistaken about the nature of your misunderstandings.

The stats --> xG conversion is very good. The more you dominate the more you're rewarded, more or less squared. The other way round, the more you dominate the defensive sector in question, the less likely you are to get a goal, it's also more or less squared.

Galywat: Bah dans ce cas, proposez une loi différente de celle régit actuellement. Personne n'a proposé quoique ce soit de précis alors que pour le coup, la transcription stats d'un secteur --> xG est assez lisible.

Perso je trouve que c'est très bien tel qu'actuellement et que vous vous trompez sur la nature de vos incompréhensions.

La conversion stats --> xG est très bien. Plus tu domines le secteur donné de l'actu plus tu es récompensé, c'est plus ou moins une loi carré. Pareil dans l'autre sen...

I understand your point, but in fact a proposal has already been made: go back to the previous balancing while increasing the actual sampling of the match.

The idea is not to say that the old version was perfect. It wasn't. But it did have a certain consistency, particularly in terms of the acceptability of results and the overall reading of the balance of power.

Today, the problem may not just be the stats → xG formula. I'm quite happy to accept that it is readable and that it rewards strong dominance. But if this formula is applied to too few updates that are actually taken into account, then the randomness remains enormous.

For me, a more coherent approach would have been :

go back to a balance closer to the old one;

increase the number of real updates per match;

add or reinforce a calculation of complementary opportunities.

Mechanically, more samples means less variance. And with more actions actually calculated, we could have increased the number of goals without necessarily artificially inflating the conversion probabilities. Dominance would have been more likely to be expressed over the duration of the match, rather than being heavily dependent on a few decisive updates.

It's a bit like throwing a dice: on 10 throws, you can get a very unbalanced result. On 100 throws, the statistical logic comes out much better. The current problem, as I see it, is that we've increased the weight of certain throws without necessarily increasing the number of throws enough.

So no, I'm not necessarily saying that the whole stats → xG law is bad. I'm saying that the overall balance between probabilities, number of updates and consistency of results seems less good than before.

After that, yes, increasing the number of real updates would surely require a bit more server resources. But given that the number of active users doesn't seem to be increasing, I imagine that technically it would at least be a feasible option.

The aim isn't to break the engine, but to find a healthier compromise: more goals, yes, but with more consistency and fewer results that give the impression that a few isolated throws overwhelm all the optimisation work.

Woodz: Je comprends ton point, mais justement, une proposition a déjà été faite : revenir à l’équilibrage précédent tout en augmentant l’échantillonnage réel du match.

L’idée n’est pas de dire que l’ancienne version était parfaite. Elle ne l’était pas. Mais elle avait une certaine cohérence, notamment dans l’acceptabilité des résultats et dans la lecture globale des rapports de force.

Aujourd’hui, le problème ne vient peut-être pas uniquement de la formule stats → xG. Je veux bien entendre qu’elle...

I agree with you, but increasing the number of news items per match would also be a good idea.

Because the balancing didn't change too much in the end. I also remember that the variance was ok from Aymeric's point of view (from memory he compared it with that of Ligue 1), so I don't think he'll be keen on that.

But I really think that for a good understanding of the MDM by everyone, having additional data even when a news item doesn't generate actions such as a goal or an opportunity, such as an indication of the sector where it takes place, would be a good development.

Whether you use the term "statistics" or "catching up", they all mean the same thing.

A lot of goals for few XG on the one hand and a lot of XG for few goals on the other.
If goals increase without a proportional increase in XGs, in a system where XGs and goals come together, there will automatically be matches with a lot of XGs but no goals.

Incidentally, there was talk of a proposal above, but there was one (by Tomas, from memory) that suggested reviewing the maximum XG limit for an occasion.

Marcus Aurelius: Que ce soit utiliser le terme statistique ou le terme rattrapage; c'est au final désigner la même chose.

Beaucoup de buts pour peu d'XG d'un côté pour beaucoup d'XG pour peu de buts de l'autre.
Si les buts augmentent sans augmenter proportionnellement les XG ; dans un système où XG et buts se rejoignent, il arrive mécaniquement des matchs avec beaucoup d'XG sans but.

D'ailleurs, ça parlait de proposition au-dessus ; mais il y en avait une (de Tomas de mémoire) qui proposait de revoir le bri...

Yes, but we know that when the term "catch-up" is used, it's to support the belief that an 8-0 friendly defeat will penalise you for your league match the following day. When in fact the two matches are not at all connected.

Yes, increasing the maximum xG per action is an idea, but Aymeric already did that last season.

And once you've got that down, you don't need to look any further to understand the increase in inconsistencies. Even if the system as a whole gets back on its feet and tends towards coherence.

As for the increase in XG; IRL the max for an action is much higher than vf. Between 0.76 and 0.79 for a penalty (depending on where you find the info) and obviously a shot on target 0.9 (I think that's low, but they must have over-counted Choupo-Moting's shots).

Galywat: Ca je te rejoins, augmenter le nombre d'actus par match serait aussi une bonne piste.

Parce que l'équilibrage n'a pas trop varié au final. Je me souviens aussi que la variance était ok du point de vue d'aymeric (de mémoire il avait comparé avec celle de la ligue 1), je pense pas du coup qu'il sera chaud pour ça.

Mais bon, je pense vraiment que pour la bonne compréhension du MDM par tous, avoir une donnée supplémentaire même quand une actu ne génère pas d'actions tels que but ou occasion, ...

And the first indicator that wouldn't be difficult to put in place would be the xg, which would be displayed continuously during matches, as it was before. This would allow us to see 'live' whether we're creating chances that have a good chance of generating a goal.

We're even prepared to do some teaching to explain to those who will complain when they see the xg increase hundredth by hundredth that, of course, it doesn't correspond to reality, but that it's just a display ...

where the simplest thing is to play without worrying, whether the results are good or bad, it's all good. It's a little everyday pleasure.

Demi-cerveau: Et, un premier indicateur pas compliqué à mettre en place pour avoir une meilleure lecture, ce serait la xg qui s'affiche en continu pendant les matchs, comme c'était le cas avant. Histoire de voir "en direct" si nous sommes en train de créer des occasions ayant une bonne chance de générer un but.

On est même prêt à faire de la pédagogie pour expliquer à ceux qui râleront en voyant augmenter la xg centième par centième que certes, ça ne correspond pas à la réalité, mais que c'est juste un af...

Except that the xGs increased for both at the same time, so it was a bit tricky.

Blagoje Vidinic: Sauf que les xG augmentaient pour les deux en même temps, c'était bancal malgré tout.

I don't see why xg shouldn't increase for both at the same time. Fortunately, you have the right to create a situation for yourself on an update at the same time as your opponent. Then, depending on the parameters, the game creates an opportunity for one or the other, for both or for neither.

The game translates the situation statistically in terms of xg. But it's not binary, all for one, or all for the other,.on each update.

Woodz: Revois tes priorités, on avance déjà pas alors si tu tir le débat vers le bas on s'en sortira pas.

The question is above all: should the subject move forward WITH you or let the people who have done the statistics and in-depth studies lead the debate?

aloisio: La question est surtout : est ce que le sujet doit avancer AVEC toi ou laisser mener les débats avec les personnes qui ont fait des stats et mener des études approfondies?

Easy answer, do without me.

Demi-cerveau: Je ne vois pas pourquoi les xg n'augmenteraient pas pour les deux en même temps. Heureusement que tu as le droit de te créer une situation sur une actualisation en même temps que l'adversaire. Après en fonction des paramètres, le jeu crée une occasion pour l'un, ou pour l'autre, ou pour les deux ou pour aucun des deux.

Le jeu traduit la situation au niveau des statistiques que nous voyons en terme de xg. Mais ce n'est pas binaire, tout pour l'un, ou tout pour l'autre, sur chaque actualisation.

Of course we have the right, I'm just saying that as an indicator, it's not necessarily relevant in the sense that it's not clearly legible.

I recently watched an 'after' programme on Canal where they used an indicator called 'momentum' to show the highlights and their intensity for both teams.

Demi-cerveau: Je ne vois pas pourquoi les xg n'augmenteraient pas pour les deux en même temps. Heureusement que tu as le droit de te créer une situation sur une actualisation en même temps que l'adversaire. Après en fonction des paramètres, le jeu crée une occasion pour l'un, ou pour l'autre, ou pour les deux ou pour aucun des deux.

Le jeu traduit la situation au niveau des statistiques que nous voyons en terme de xg. Mais ce n'est pas binaire, tout pour l'un, ou tout pour l'autre, sur chaque actualisation.

Personally, I think that if it's binary as it stands

Either one or the other, and I'm 90% sure of that

Blagoje Vidinic: Évidemment qu'on a le droit, je dis juste que comme indicateur, ce n'est pas nécessairement pertinent dans le sens où la lisibilité n'est pas claire.

J'ai récemment regarder un "after" sur canal où ils utilisaient un indicateur appelé "momentum" pour indiquer les temps forts et leur intensité pour les deux équipes.

Yes, well, when both increase at the same time, and one increases 5 times faster than the other, I personally think the message is pretty clear. Even if it's an indicator that can be improved, I'll grant you that. I was looking for something very easy to implement. And I know that this one is easy, because it was in place until recently. The calculation still exists, it's just not displayed any more.

Ced90: Personnellement je pense que si en l’état c’est binaire

Soit l’un, soit l’autre, et jen suis sûr à 90% (sur une actu)

I think your account is too new for you to have experienced this, but when I started in 2020, you could see the xg evolving, live, minute by minute, and it was very common for both to evolve at the same time.