Don't know if what we have is fancy or not.

My group uses a simple grid system. Each player gets some standard graph paper, marked with zones boundaries at \$x\$ = −15/+15 and \$y\$ = −15/+15. In between those are Neutral zones. Going up the \$y\$ axis is good; down is evil. Right on the \$x\$ axis is lawful; left is chaotic.

During play as we progress and perform actions the DM may call for a change in alignment. We then move out the dot of our current alignment the number of points he said. This gives us the ability to track our characters' alignments pretty well.

If we disagree with his decision this is the only time in the game we stop the game to calmly and enjoyable debate with the disagreeing person, getting 5 minutes to convince everyone vs. 5 minutes for DM to plead his case.

After each has spoken the group votes. This vote supersedes the DM's ruling.

We do this because we decided that none among us can honestly say that we have a perfect moral compass. So the decision is left in the end to a group morality decision.

You could incorporate the graph system into your game for each person and keep it hidden, but as others have said I think hiding it is a bad idea. Someone wanting to hold to their alignment needs to know when it might change, especially if that change may have negative consequences on his class.

So, you are looking for a mechanical way of judging the outcomes of moral agência? Good for you: you are joining a quest that humanity has been engaged in for at least 4,000 years covering fields as diverse as philosophy, religion, sociology, politics and law - there is no end point in sight.

What is evil anyway?

If you actually read the description of D&D 5e's alignment on p. 122 of the PHB it says:

Alignment is a combination of two factors: one identifies morality (good, evil, or neutral), and the other describes attitudes toward society and order (lawful, chaotic, or neutral).

So herein lies a problem: after several thousand years of thinking about this stuff, there is no consensus of what "good" is: is it altruism, utilitarianism, liberalism, egalitarianism, something else? Similarly with "law", do we need: democracy, communism, feudalism, socialism, something else?

If you read the actual descriptions of each alignment combination the commonalities are that "good" appears to be synonymous with "altruistic" while "evil" means "selfish". Similarly, "lawful" creatures put society ahead of the individual while "chaotic" ones do the reverse.

Dê uma olhada em Como você decide o alinhamento das ações de um PC? for some thoughts on this.

If you adopt this as your definition of good/evil, law/chaos; notwithstanding that it models actual moral agency the way a light globe models the sun, it at least gives you a metric that can be applied to alignment.

Is evil in thought, deed or outcome?

So now that you can qualitatively assess alignment you need to decide if you will judge based on:

1. thought processes i.e. if the players contemplate an Evil (or Good/Chaotic/Lawful) course of action does that move them on the alignment scale even if it is rejected?
2. intentions i.e. if the characters do a evil thing for good reasons is that evil or good?
3. the outcome i.e. if the characters' actions cause evil to occur in spite of their intentions are they morally culpable? Does this change if they were negligent? Or reckless?
4. A combination of all three.

The currency of alignment

So, are you going to quantify alignment? If you are considering this think about:

1. How many orphans do I have to feed to equal 1 drowned puppy?
2. Is it more good to donate 10,000gp to build a temple to Torm or to feed 10,000 people in Torm's name?
3. Is spending 50gp when your net worth is 100gp in a good cause worth more or less than spending 100gp when your net worth is 200gp? Or 300 gp? Or 150gp?
4. If I break the law by stealing 10gp, is my account equaled by catching another thief who also stole 10gp? If not, what if they stole 1,000gp?
5. If I save the world from horrors beyond the edge of sanity, can I go around kicking beggars for the rest of my life?

Putting a metric on this stuff is hard.

Tensão

What is a Lawful Evil character to do when following the law will damage him personally? Or a Chaotic Good when it is clear that by bowing to the will of others many people will benefit (essentially the central issue of Huxley's Admirável Mundo Novo).

Complicações

There are lots. It will suffice for you to consider your and your player's reaction to the Trolly Problem

Who decides?

Are you going to agree with your players what is and is not a good/evil/lawful/chaotic action or are you going to impose your beliefs on them?

Will they know in advance or as each action comes up?

Status quo

If you do nothing at all, does your alignment stay where it is or does it drift towards True Neutral?

That is, to stay Lawful Good must you meet a quota of Lawful and Good acts? If so, how many?

Sigilo

Secrecy about a player's character is almost universally ruim.

The character is the apenas thing the player has control of and they know all about them: hit points, Intelligence, Armor Class, Traits, Flaws etc. Keeping nada about their character secret from the player is não fair. As a DM you have the whole world to keep secret: let the player's know the consequences of their actions.

That means they should know what alignment they are and what alignment they are heading towards.

Who are the heroes?

Quoting from my answer to Como faço para que meus PCs não sejam um bando de cretinos assassinos?

As for Role-Playing; what is the role that the players fulfill in the world? If they are seeking classic heroic fantasy (whatever the genre) then it is incumbent upon them to behave heroically and it should be clear what heroic behavior is before play commences.

If heroic means Sir Galahad then

1. This is a high bar to set, and
2. It is not fair of a GM to put in place ethical dilemmas that cannot be unambiguously resolved
3. It is also not fair to put the players in a position where ethical behavior is obviously suicidal.

If heroic means Wyatt Earp then

1. This person has been delegated a state's "monopoly on violence"
2. There must be unambiguous bad guys
3. These bad guys must be brought to justice
4. If they can't be brought to justice its OK to kill them
5. Everyone else must be protected

If heroic means James Bond then

1. This person has also been delegated a state's "monopoly on violence"
2. There must be unambiguous bad guys
3. It is clearly OK to kill people simply because they work for the bad guy.
4. Everyone else must be protected

If heroic means Batman then

1. The heroes are self-appointed vigilantes
2. Ambiguous bad guys are OK (e.g. law enforcement that don't understand what's really going on)
3. The bad guy must die or be captured
4. Anyone who gets in your way is fair game
5. Collateral damage should be minimized but that is a secondary consideration.

If heroic means Torquemada then

1. You have a divine mission
2. People are damning themselves for all eternity
3. It is merciful of you to give them pain now in order top save them from eternal agony.

If heroic means Tamerlane

1. Everything and everyone in the world belongs to you
2. If they resist they should be butchered and have their skulls made into a pyramid.

A Solução

So, having decided how hard it is, here's a simple way to do it.

1. The normal rough and tumble of what passes for adventuring in your campaign has no effect on alignment.
2. There should be example acts defined and known to the players for "Good", "Evil", "Lawful" and "Chaotic". Committing acts like these move you to that alignment - do not stop at Neutral, do not collect $200.
3. There should be edge case examples and known to the players where this is há pouco "Good", "Evil", "Lawful" or "Chaotic". A sustained pattern of behavior like this will move you one alignment in that direction.
   1. At the end of each session the player's should be told of their "drift".
   2. Good acts don't cancel Evil nor do Lawful cancel good.
   3. Old acts fade away - say after 10 sessions.
   4. 7 sessions of the last 10 with a given "drift" moves you in that direction and resets the Good/Evil or Law/Chaos "clock".

Take inspiration from the WoD Humanity/Path system.

The only alignment-tracking system that I've used that both allows for shifting alignments like you specify and also has a decent method of bookkeeping is the Humanity (and Path) system from the World of Darkness games. There's a decent rundown of the specifics aqui. While there are a number of specifics unique to Vampire: The Masquerade in that article, the way I run it in a D&D-style game is this:

The Good/Evil and Law/Chaos alignments work on a sliding scale of 9 points. If you have a 7-9 Good rating, then you are Good. 4-6 is Neutral, 1-3 is Evil. The same applies to Law and Chaos.

There is a list of "sins" against Law and Good, nine entries long. Each entry describes the worst act that someone of that Law or Good rating would normally feel bad about doing. For example, a Good 7 character might sometimes steal things, but only if there's a very good reason, where a Good 6 character wouldn't feel bad about stealing a thing that they wanted. A Good 3 character would hurt someone to get their way, and not feel bad about it. A Good 1 character wouldn't even be bothered by cold-blooded murder.

If a character commits a "sin" below their current Law or Good rating, then they get a black mark against them in that rating. If the "sin" is lower than 2 points below their current rating, then they get 2 marks instead. If they get a specific number of black marks (I use 3, but you can change that number to make alignment shifts more or less likely), their rating drops 1 point.

On the other end, if someone does an act that is significantly Lawful or Good, they get a white mark on that rating. A white mark cancels out a black mark, and if the character accumulates 3 white marks, then their alignment rating raises by 1 point.

The advantage of this system is that you have a list of specific acts that tell you if someone is breaking alignment, rather than a few wishy-washy sentences from the PHB. Your players might not know what their alignment is, but if you show them the list, then they can at least know what kinds of things they should avoid. I also find that it doesn't involve much in-the-moment bookkeeping. My usual practice is to note down whenever a player does something morally questionable, and then think about whether or not it counts as a sin later on, after the session is over.

The disadvantage of this system is the same disadvantage as every alignment system: very nearly every person has a different idea of what "good" and "evil" mean. If you have a system more granular than "do good and your alignment will be Good", then you will have arguments and disagreements about exactly where each act sits on the chart. I sidestep this issue a bit by saying that the definitions of Good and Evil are decided by the in-game gods, who are imperfect beings and occasionally make bad decisions.

In the end, I find that all alignment systems of any sort are more trouble than they are worth, but I've used this system in the past to track the more granular changes that you're looking for.

Designate Touchstone Events, Evaluate Intent, and Involve Your Players

Touchstone Events

These are events which are designed to call into question a specific Alignment Axis (Good/Evil or Law/Chaos). They should be placed in an adventure and should generally be unavoidable. They may or may not target multiple characters.

These events should be predefined by the DM with a sample of potential PC reactions and the alignment correspondence of each. How a player reacts based on these guidelines determines how you mark their score for the event.

Example: The players see a green dragon flying toward a nearby town. It looks angry.

• Good: “We should hurry to the town and do whatever we can to help regardless of the danger!”
  Rationale: This character has placed the well-being of others above his own.

• Neutral: “There’s nothing we can do against a dragon that size. We should warn the city guard in the capital city.”
  Rationale: This character recognizes the need for action but does not place himself in harm’s way.

• Evil: “This means the dragon’s hoard is unguarded! We can grab the treasure while the beast slaughters the townfolk!”
  Rationale: This character has disregarded all other needs beyond his own.

• Lawful: “We should rally the guardposts and ensure the local clerics are prepared for the influx of wounded.”
  Rationale: This character looks to laws, structure, or ruling powers for solutions.

• Neutral: “What is the best way we can deal with this situation?”
  Rationale: This character evaluates the situation using no guidelines but the existing circumstances.

• Chaotic: “We don’t have time for guards, here! Let’s just get into the action!”
  Rationale: This character disregards laws, structure, or ruling powers completely.

Evaluating Intent

This is a less specific way of gauging where a character’s alignment may lie. Consider this as a guideline: “The actions a character suggests and advocates for are the actions the character would take in a vacuum, and thus are indicative of his alignment.”

A Lawful Good Paladin devoted to protecting the weak and fighting evil in all its forms, for instance, should never even suggest that the dragon can be kept busy eating townsfolk while the party gets a better tactical position. Players of such characters, likewise, should not present such options or at least preface them with “This is strictly out of character, but…”

Avoid penalizing here for players who are “overruled” by the group. If the paladin suggests charging in to help but the other four players disagree, it may not be appropriate to penalize the character for “going along” with a different plan (unless that plan violates a larger code).

Involve Your Players

This is particularly important with Evaluating Intent. A Chaotic Evil character, for instance, might strongly urge the group to charge toward the green dragon and help the town. On the surface this appears to be a Good act, but the player may see this as a way to gain new loot or dispose of troublesome allies (if it’s that kind of game).

When a player comes up with a plan that seems “at odds” with their chosen alignment, talk to them about their rationale. This can be done in public or in private and a request from a player for a private conversation should be honored if possible.

Potential Concerns

Beware the “Neutral” alignments, as they are often difficult to roleplay consistently (Lawful Neutral: “There’s nothing we can do about the lich stealing children from the orphanage due to Code 16-32B of the local adoption laws stating that…”). Some, like Chaotic Neutral, can even be interpreted as “I do what I feel like doing all the time, so everything I do is Chaotic Neutral.”

For classes where Alignment can be important (such as 3E druids and paladins), remember that part of their training prior to first level is “how to act.” Consider giving warnings about a specific course of action that it goes against their teachings, since your players may have different concepts about what is “appropriate” compared to your own interpretations of an action.

Consider a “point system” (perhaps utilizing the grid mentioned in other posted answers) where an Intent Evaluation is worth 1 point and a Touchstone Event is worth some multiple of X. The two may be roughly equivalent- for instance, you should expect to see 3 Intent Evaluations for each Touchstone Event if a TE is worth 3 points. Alternatively, change this value if actions should speak significantly louder than words.

Source of Framework

This is not a system I have utilized in a specific attempt to keep players unaware of character alignment, but rather the method I have used over 20+ years of GMing D&D-based games to evaluate character actions against a stated alignment.

Pathfinder has an alignment tracking/change set of rules completely compatible with 5e. You can read the basics here on the SRD, but there's more color around them in the actual Ultimate Campaign book.

Every character has a 9-point scale for the lawful-chaotic alignment axis, with 1, 2, and 3 representing lawful, 7, 8, and 9 representing chaotic, and the rest representing neutral. Each character has a similar scale for the good-evil alignment axis, with 1, 2, and 3 representing good and 7, 8, and 9 representing evil.

The player decides where the character's alignment is on the alignment track. (...) When a character performs an action that is out of character for his listed alignment, the GM decides whether the action is enough to shift the character's alignment on the appropriate alignment track, and if so by how much...

The system has a lot of advice on adjudicating alignment and alignment change and covers how lot of common cases (using the atonement spell, helms of opposite alignment, etc.) interact with the system.

Há também a system for deliberately changing alignment to good and examples of what kind of penances etc. are needed to do so. (Changing to evil's pretty easy...) This system is part of the Wrath of the Righteous Adventure Path so has been used a lot by players.

I have had a few games where alignment was rather important, and I've always used the below logic to handle the situation:

I made a note of each player's alignment in a notebook. Whenever, during conversation or by action, their character did something pretty contrary to their alignment (DM's discretion as to what that means, but I always took the circumstances into account, so something like the trolley problem wouldn't factor in one way or another), I would make a tally next to their name and the alignment they were "drifting" towards. Acting consistently in their alignment for a session would erase a mark.

Once the player received a significant number of marks (I ranged this from 5-7, depending on the seriousness of the playing group), I would pose them with an event that would "test" their alignment. Basically I would set up a situation similar to the ones that had caused the marks and was appropriate for their character. However they acted during the test determined their alignment change, but never more than one "step." If it was a class where the change would remove abilities, I might give them an in-game warning or omen about their actions (e.g. a god talking to them in their dreams, or a vision of the world should they continue down their current path).

To provide an example, say a Paladin (LG) regularly executes "bad people" like thieves or anyone he doesn't trust not to harm others. From my DM perspective, this goes against lawful good, and probably tends towards Lawful Evil. After a number of incidents like this, I send the player a warning in a dream, the image of his character cutting down young children (or some other symbol of innocence). In the next session, the players are given a similar situation: they end up with some prisoners, who they think have done some evil deeds. The paladin tries to execute one, only to have it turn into a visage of his god and smite him with an alignment change, making it very clear why its happening.

I've also always given players a chance to atone for their "sins" through question and other things. Even going so far for a paladin similar to the one above to be temporarily regranted his abilities, as long as he was on his quest for atonement. I rarely try to be this strict as it doesn't make for very fun gameplay, and most of my games go towards the lack of absolutes in good or evil.

When I need to make an alignment decision in those kinds of games, I just go for the WWJD position: I determine goodness or evilness based on the viewpoint of the being that created the spell or effect. A nature goddess might see "evil" as destroying nature, while a civilization god might see the creation of cities as "good" (even if it destroys some natural land).

This is the (somewhat simplified) system I use for tracking players alignment changes. I have play-tested it a couple of times on D&D 5e with my friends and they all seem to approve.

I will be bypassing the aspect of which choices lead to various results - as I believe that is up to the DM to decide. An example of this might be that killing for some is a wholly evil act - where as others may deem it to be an act of mercy/good. I think this is to open a topic to be properly discussed.

Anyways, here are the steps to my system.

1. Players decide their alignment at creation. I do not normally judge them unless they have an elaborate backstory that counters what they are saying (e.g. I am not going to let the rogue be Lawful good if he consistently breaks the law in his backstory).
2. Every major choice by the player that may effect the alignment is accounted for, typically what I do is a 'tick' on my DM sheet (usually a + or -). This indicated a good or evil choice the character made.
3. With enough 'ticks' I will ask the player to update their alignment - generally cancelling all 'ticks' when the alignment is changed.
4. Typically 3 'ticks' to leave a regular alignment in either direction. If the alignment is chaotic I usually give up to 5 ticks.
5. Any act that is deemed a major change of character (Such as the holy cleric suddenly murdering in cold blood for an evil reason) I ignore the tick system and do a complete change of alignment on the spot. Some acts are seen as too severe to do anything but an immediate change.
6. Finally, if the action is in line with the current alignment, no tick is added. This helps people from accidentally swapping between similar alignments (such as Chaotic Good to Lawful Good). It has to be a concentrated effort to move to similar alignments.

The reason that Chaotic is treated as special is because generally it can swing either way. A player - playing for self interest might act lawfully for quite some time before breaking the law. This does not mean the character should have his alignment changed, simply that he was being utilitarian and only acting when needed. Since chaotic is harder to pin down, without additional space you will end up swapping alignments a lot.

I would suggest something along the lines of a sort of scale. With benchmarks at each point. For example you have your Good/Neutral/Evil scale out of 100. Good sits at 75, Neutral at 50 and Evil at 25. Whenever a character does something that you determine to be good, depending on the size of the deed, the scale goes up X. Viceversa for a evil deed. Repeat this scale with Lawful/Neutral/Evil.

However for this system to work, in my opinion, it would be imperative that the players not know where they are on these two scales. Otherwise it'll defeat the purpose. Additionally it would be beneficial that the players not know at all how you are tracking their alignments. And how they managed to transition from one alignment to another.

However consider this. My DM I am currently playing with has a similar but slightly different way of dealing with alignments. He lets your character pick hhis/er alignment. Then waits a few sessions until we come across a pivotal moment in your characters development and the development of the story. And depending on the actions of your character, he will then take you aside and ask you to defend your actions if he felt it was something that definitely wasn't an action in your alignment. If he feels your argument reaffirmed your alignment you stay in that alignment. However if not, or if you do say, "Yea my character definitely is not the alignment on the paper", then he has you change it. He commonly says "Your alignment is not fluid, until it has been tested".

For example I was playing a Chaotic/Neutral ranger in a game. My character met up with the group, who were on their way to rescue one members, Docklan, children who had been taken in the middle of the night by some army of the "Lord of the Mountain". We come across a town on our way there, the towns people treated us well. We learned of a possible shipment of soldiers coming in to give supplies to the town, from the "Lord of the Mountain". We then waited until nightfall, where we tied up the militia of the town and waited for the cart to arrive. When it did my character investigated the materials inside. He found a crate of weapons and a note that implied that the town was to go armed and kill off the small encampment where Docklan had his children stolen from. Later on that night we hear from a shed a crying, I go investigate and we find one of Docklan's kid, all beaten and bruised and burnt in the shed. He runs to his father and it is here where my character loses it and kills most of the town in their sleep.

The DM took me aside and I explained that over their journey he bagan to respect Docklan and a combination of the weapons and note and in his opinion knowing that the town was guilty was enough to send him off the edge. He said that, that made a lot of sense and so I stayed Chaotic/Neutral. Instead of Chaotic/Evil.