Frustrated that despite my experience with the game, I seemed to be missing several subtleties, I did what I couldn’t do back then (around 2000) — I searched online. I was amused to find that, even in 2023, there are still websites that look like they’re straight out of the early 2000s, yet form an impressive knowledge base on the series. On those sites, I found detailed Tetra Master rules, and sure enough, I was missing a lot of pieces to fully understand the game’s rules. During my research, I also came across many forum posts from players just as disillusioned as I was. Once I discovered the rules, another question naturally arose: "How did these sites figure out these rules?"

Since the cards are part of the game’s story world, it makes sense that there are multiple ways to collect them. After discussing the various methods for acquiring new cards, I will try to subjectively determine the most effective approach:

Each card has several attributes that determine its strength, so two cards with the same artwork can actually differ. A point that is both fun and frustrating is that these attributes aren’t set until you actually obtain the card. In other words, cards given by other characters or found in villages and dungeons have their characteristics determined upon discovery. It’s therefore possible — and recommended — to save your game before collecting a card and reload until you’re satisfied with its stats. This trick works for all cards, including rare ones.

Throughout this guide, I will frequently complain about randomness, which amplifies the game's non-determinism. The variability in the stats of a card found or received is the first clear example of this.

This is probably the most comprehensive part of the guide: indeed, some NPCs play cards. To challenge them, simply approach and interact by pressing the square button.

This section outlines the game flow in very broad terms, leaving out many details:

You and your opponent take turns placing 5 cards each on a 4x4 grid. Cards can be won from your opponent depending on where they are placed. Once all cards are placed, the player who owns the most cards wins.

From this, we understand that the goal is to capture as many of your opponent’s cards as possible. What follows are the different methods of capturing cards.

In the various diagrams, black cards represent the player’s cards and gray cards represent the opponent’s. The card marked with a circle is the one that was just placed, and the arrow between card patterns indicates the effect of that placement.

"You win an opponent’s card when an arrow on your card points toward the opponent’s card."

From this first (very incomplete) bit of information, we can deduce several things:

• Cards have arrows — since the opponent’s card in the example has none, we can assume that cards have a variable number of arrows, pointing in different directions;
• It’s reasonable to assume that cards must be adjacent for one to be captured;
• A captured card stays in its position but changes color to reflect the player who now owns it.

"If an arrow on your card faces an arrow on your opponent’s card, a battle occurs. The numbers shown on the cards are subtracted, and the card with the higher resulting value wins the other."

The battle mechanic, as shown in the diagram, reveals new elements about how the game works:

• Cards have stats;
• It confirms our earlier assumption that the arrows on cards vary.

"When you win a battle, all cards targeted by the arrows of the newly captured card are also captured."

The main idea illustrated by this diagram is that, in addition to the arrows, it’s possible to create a capture chain — a combo that spreads the success of a capture to all adjacent arrows. However, this section raises more questions than it answers:

• Do combos propagate recursively? (This can be answered quickly by playing a few matches)
• If a captured card — as part of a combo — is placed into a battle scenario, does that battle still occur?

As mentioned in the introduction, the manual states that cards have stats — something we also saw during the battle example. However, it explicitly says that it won’t explain how to interpret those stats, and as we’ll see, deciphering them is far from straightforward without help.

At the end of a match, the winner can choose one card from those they successfully captured from the opponent. However, if the winner manages to capture all of the opponent’s cards — achieving a Perfect Game — they get to take all of the opponent’s cards. This is quite a windfall when you win, but can be incredibly punishing if you lose.

We also learn that it’s possible to reclaim lost cards by rematching NPCs who have defeated you.

A small note tells us that you cannot own more than 100 cards and that you can get rid of cards from the card menu. This is a strange choice because, as the card menu reveals, there are 100 different cards in total. We can therefore assume that once you are advanced in your card collection, building a strategy that uses multiple copies of the same card becomes difficult. Moreover, consulting this menu anticipates the frustration you'll feel during your first games, as it displays information that cannot be linked to what you learned from reading the manual.

Since the game is played between two players, each starting with five cards, the two prerequisites are:

• Have at least 5 cards;
• Find an opponent.

Once these conditions are met, starting a match involves interacting with a character using, by default, the square button. You will be asked to choose the five cards you want to use. One might argue that ideally you should pick your five best cards; however, since you can lose either one or all five cards, you also need to weigh the risk and reward (as we’ll see later, combos can turn the tide).

When a match begins, a coin toss decides who places the first card — the player has a 50/50 chance to start. After that, each player takes turns placing one card at a time.

The game board is a 4x4 grid, consisting of 16 spaces. To add variety, the start of a match can randomly block between 0 and 6 spaces (once again adding randomness), where no cards can be placed. Since a maximum of 6 spaces can be blocked, there will always be enough room to play all five cards.

Here are three examples of possible boards. The first has no blocked spaces, the second has 3 blocked spaces, and the last has 6 blocked spaces.

Although I complained about the excessive randomness in the introduction, here I find the slight variations in the board layout to be a very good idea, allowing players to devise complementary strategies. For example, in the third example, the isolated spot with no adjacent free spaces lets you place a weaker card that can never be captured, preventing a perfect defeat and ensuring that, at worst, you lose only one card.

We saw in the manual that to capture a card, you simply place a card, and all cards pointed to by its arrows are captured. In this example, the black card captures three targeted cards.

However, for the arrows to be effective, the card must have just been placed. In this diagram, the grey card does not affect the newly played black card:

Because a played card is offensive, and once placed, it switches to defensive mode. For a card that has already been placed to capture another, a battle and/or a combo must occur — both of which we'll explain in the following sections.

At this point in the article, you might be thinking I’ve been exaggerating the game’s complexity — and that, even if the manual wasn’t thorough, a few matches should be enough to grasp the full mechanics! Rest assured, in this section, we’ll quickly see why playing Tetra Master without understanding key information tied to card stats — which only become apparent during battle situations — leads to constant low-grade anxiety. You’ll frequently find yourself unsure of why you’re winning... or worse, losing.

As a reminder, when placing a card adjacent to another card of a different color, several scenarios can occur:

• If the placed card points to one or more opponent cards, and none of those cards have arrows pointing back, it simply captures all targeted cards.
• If the placed card doesn’t point to any opposing card — even if it is pointed at by an already-placed card — no capture happens.
• If the placed card points to one or more opponent cards and one of those cards points back, a battle occurs. The placed card is the attacker, and the targeted card is the defender.

We also read that, in a battle situation, the card with the highest number wins. And what’s amazing is... before diving into the rules, the obscure document, and a few attempts at reverse engineering, I was completely unable to understand how that number was determined. Buckle up — I think this is the point where the designers of Tetra Master — and whoever signed off on the Game Design Document — completely lost it. Because from my perspective, even after playing thousands of matches, it’s impossible to figure out how this number is defined.

Although we haven’t yet fully broken down the anatomy of a card, we’ve speculated that a card can have multiple arrows, potentially pointing to several different cards. For example, in this situation:

In this situation, we can see that the played card is in a battle situation with both opposing cards. While the game usually decides randomly or applies a questionably explicit rule to determine battle order, the attacking player chooses which card to attack first. This allows for genuine strategic play, as we’ll explain further when discussing combos.

A card — which is generally more visually appealing than my simple diagrams — can have between 0 and 8 directional arrows, each representing an attack direction — and, as we’ve seen in battles, also defense directions. It might seem strange to have (and use) cards without any arrows, as they would be completely ineffective — only able to be captured and never able to capture others. However, such cards can be surprisingly useful to break combos or, as a last move, to simply add a +1 to your count of captured cards.

Next, each card has a sort of cryptic identifier, hard to decipher — especially on a CRT television. For example, here is the diagram of a card that can attack in all directions, with the identifier: 1P23:

I believe we have a fairly good grasp of the arrow mechanics and their role, so we can speculate that the identifier describes a card’s stats and is used to calculate the number that determines a battle. That’s exactly right, but we’ll see there are many implicit rules making the understanding of these numbers very challenging.

From my perspective, it’s probably impossible to deduce the relationship behind that infamous identifier (in our example, the notorious 1P23). So I’ll be very explicit: it provides four pieces of information about the card, in this format: PTDM, where:

• P is the card’s offensive power, expressed in hexadecimal, from 0 to F; in our example, the value is 1.
• T is the card type, which can be P, M, X, or A; in our example, the value is P.
• D is the card’s physical defense, also expressed in hexadecimal, from 0 to F; in our example, the value is 2.
• M is the card’s magical defense, also expressed in hexadecimal, from 0 to F; in our example, the value is 3.

I assure you that the fact that these digits are expressed in hexadecimal can be very confusing once you start progressing in the card adventure. Indeed, neither the game nor the manual ever explains how to read these statistics. Having been used to seeing at least three digits, suddenly seeing letters appear can cause a lot of confusion! Especially since the statistic AAAA is perfectly valid. So even if, like me, years after your first encounters with Tetra Master, you have a background in computer science and decide, naively, to give its baffling complexity another try — even if you are already familiar with the hexadecimal numbering system — chances are you won’t realize it’s actually hexadecimal, since the pattern of the statistics doesn’t look anything like a number expressed in any known numeric system.

Before diving into the concrete meaning of these hexadecimal values (because yes, it’s way more complicated than you might think), let’s look at the different types of cards:

• P (for physical) is a card that uses its offensive power to attack the physical defense of another card. For example, in a situation where 7P23 attacks 2M12, we roughly have 7 vs 1.

• M (for magical) is a card that uses its offensive power to attack, logically, the magical defense of another card. For example, in a situation where 2M12 attacks 7P23, we roughly have 2 vs 3 once again.

• X (for ... power (haha) or, according to some translations, flexible) is a card that uses its offensive power to attack the weaker of the two defenses of another card. So, in a situation where 7X12 attacks 2M32, the card targets the magical defense. Roughly speaking, once again, it’s 7 vs 2.

• A (for advanced or assault) is a card that uses its highest statistic (whether offensive or defensive) to attack the lowest relevant statistic (including offensive power) of another card. For example, in a situation where 2A8F attacks 1X2F, the card uses its magical defense to target the opponent's offensive power, roughly F vs 1.

This allows us to assert that FAFF is the best possible stat combination for a card. We’ll see later that cards can be upgraded, but one challenge at a time! First, let’s look at how to interpret the hexadecimal digits.

Since the statistics expressed in hexadecimal only cover 16 values (0 to F), ties would happen quite often. But don’t worry, the game has another surprise in store for us—one that is also completely absent from the game manual or any in-game rule explanation. Indeed, it would be far too simple if 9 simply meant 9 or F meant 15!

Now that we can read a card’s statistics, combined with understanding the arrows and card capture mechanics, we should have enough elements to figure out how to win every match. Although we still need to cover the combos, which depend on the card positions on the board and will be detailed in the next section, one might think we have enough information to win local battles. However, experienced players will quickly realize that the numbers displayed on the cards during battles fluctuate and, even worse, scenarios that seem inevitable sometimes don’t play out as expected!

For example, imagine a battle between the cards 3P10 (attacker) and 2M02 (defender). We’re pitting a physical 3 (with power between 48 and 63) against a 0 (defense between 0 and 15). If our interpretation of the stats is correct, there’s no scenario where the attacking card should lose. And yet, it happens. Indeed, there are certain random factors that allow a weaker card to win a battle. This is why, during a battle, you’ll see multiple numbers flashing on the card—reflecting the application of these random factors.

Let’s revisit our example: 3P10, which we’ll call X, attacking 2M02, called Y. As mentioned, we pit 3 against 0. Suppose the following:

• X's attack value is 50 (within the range 48 to 63);
• Y's defense value is 7 (within the range 0 to 15).

These are the initial numbers displayed (and thus the only way to know the position within the interval), even though, as we’ll see later, this isn’t reliable for long-term strategy since cards can evolve. Additionally, in certain situations — like when in a defensive position — it’s hard to know which statistic is prioritized. For example, with 1P22 against a card of type X, how can you tell if the displayed number targets magical or physical defense?

Next, the computer will generate a random number between 0 and the previously displayed number (so for X, between 0 and 50, and for Y, between 0 and 7). We’ll call these numbers the rolls. For this example, let’s assume the following:

• X rolled a 46;
• Y rolled a 1.

The rolls are subtracted from the actual values within the interval. Since each roll is between 0 and the actual interval value, the results can never be negative.

• X ends up with 50 - 46 = 4;
• Y ends up with 7 - 1 = 6.

And the card with the highest subtraction result wins. Here, against all expectations, it’s the gray card (which seems much weaker) that wins (because... logically 4 < 6). So, we end up fucked (sorry), and our black card is captured.

Calculating the probability of victory in a battle is fairly easy to simulate. Here is an approximation which, for our example with X = 50 and Y = 7, yields about ~92%:

let proba my_stat opponent_stat =
  let a, b, need_reverse =
    if my_stat > opponent_stat then (opponent_stat, my_stat, false)
    else (my_stat, opponent_stat, true)
  in
  let x = 2.0 *. (1.0 +. b) and y = 1.0 +. a in
  let result = 100.0 *. ((x -. y) /. x) in
  if need_reverse then 100.0 -. result else result

However, this approach relies on two fixed values that are generally unknown during the game: the exact positions within the intervals. An (acceptable) solution would be to be very pessimistic when calculating these probabilities by taking the lowest bound of my card’s interval and the highest bound of the opponent’s:

let pessimist_proba my_stat opponent_stat =
  let my = to_decimal my_stat *. 16.0
  and op = (to_decimal opponent_stat *. 16.0) +. 15.0 in
  proba my op

This gives us a fairly good idea of the probabilities of victory (a bit pessimistic, but in life, isn’t it better to be cautious).

The funny thing is that once you know:

• how to read the compressed representation of the statistics;
• the random factors during battles;
• the more vague points (the exact number within the interval)

The rules become relatively straightforward, and even though players can sometimes be caught off guard by the randomness, with all this information, it is, in my opinion, possible to enjoy and play Tetra Master more calmly. While I’m jumping ahead a bit, the main frustration lies in how difficult it is to discover the rules just by reading the manual or playing the game. I find it quite odd that there was a deliberate choice to be so vague about how to read the cards and where randomness comes into play. Without further ado, let’s move on to combos which, surprisingly, are very predictable and understandable even without reading the rules (yay)!

It must be acknowledged that the manual is quite thorough regarding the flow of combos. Indeed, this time there’s no randomness, no implicit rules—the manual explains: "When you win a battle, all cards targeted by the arrows of the newly won card are also won." We could rephrase this as: "In a battle situation, all cards pointed to by the losing card (i.e., the card just won in the battle) are also captured."

It is important to highlight several points:

• Even if the flipped card points to another card during a battle, the other card is taken without a battle;
• Combo effects do not propagate recursively (so if a card is taken because it is targeted by the loser of a battle, the cards it targets will not be taken).

Combos can completely turn the tide of a match because it’s possible to flip up to 8 cards in a single battle. For example, in this situation:

This situation (artificial, I admit) shows how a single card position can completely change the outcome of a game. Indeed, if the black card wins (which seems normal given that in the worst case, 224 vs. 15, it has a 96.44% chance of victory), it will flip its neighbor, which in turn will flip without any battle its 7 neighbors. In the current board setup, this means that regardless of the next move, the worst-case scenario for the black player would be a draw.

Combined with the multiple battles mechanic, combos introduce fun micro-strategic choices (and are quite rewarding when successful). For example, let’s revisit the example presented in the multiple battles section:

In this situation, we have the opportunity to choose whether to attack the card above, which has stats EXFF, or to target the card on the right, which has stats 2M12. To recap, the (pessimistic) probabilities of victory are:

• 4P23 vs EXFF with only a 12.70% chance of winning;
• 4P23 vs 2M12 with a 75.38% chance of winning.

Since the card on the left also points to the card above, defeating the left card means capturing the top card by default. This makes it a far more strategic move! Combined with the blocks we discussed when detailing the game grid, and with some cards having no arrows (thus blocking combo propagation), there are many strategies to explore in Tetra Master when focusing on combos. In my opinion, this is an excellent mechanic — well documented and easy to grasp through play (even if understanding it can be a bit obfuscated by the difficulty of reading stats and the game's unpredictable randomness).

We touched on this a bit earlier, but cards can evolve within a defined limit (specific to each card), and as with many things in this game, this is determined in a rather random way. Indeed, throughout battles, cards can evolve in two different ways:

• their attack and defense stats (magical or otherwise) can evolve up to a certain limit;
• the card’s type can change as the adventure progresses.

And what’s really fun is that these two types of evolution can happen completely independently. In other words, a card’s type can evolve without it having reached its attack and defense limits. That’s why we’ll detail these evolutions in two separate sections, then provide a list of cards with the maximum stats they can reach.

A card has one attack stat and two defense stats (physical defense and magical defense). Remember, the indices shown on the card are expressed in hexadecimal, ranging from 0 to F (15). These indices don’t represent the card’s exact values but rather an interval. While cards of the same design can have different numbers of arrows and different concrete stats (their position within the interval), when acquired, they are assigned attack and defense limits. For example, the goblin card has, by default, the stats set 0P00, meaning its actual attack and defense values can range from 0 to 15. However, through some obscure magic — because, naturally, in a game so silent about its workings, being explicit about a card’s maximum stats seems a bit overrated — we know a goblin can have at most 7 in attack (still within the 0 interval), 9 in physical defense, and 4 in magical defense. Yet, as long as a card hasn’t reached these limits, it can continue to evolve.

Indeed, after each won match, every card used during the game will have one point added to either an attack or defense stat, chosen at random. This means that after each victory, each stat has a one-in-three chance of being incremented. The only way to know which stat was increased is to play that card again in a subsequent match and check its actual battle score to see which value went up.

An amusing aspect of this random choice of which stat to increase is that it completely ignores whether the card has already reached its stat cap. So, if our goblin already maxed out at 7 in attack but has 6 in defense and 2 in magic defense, and luck decides to increment the attack stat... that point is simply lost... meh.

Once again, the obfuscation of this information — and its randomness — serve to complicate the understanding of the game’s rules! Because yes, even though the idea of card experience is, in my view, a great concept that motivates players to keep playing, without informing the player about it, this mechanic essentially makes understanding the stats (and the numbers shown on the card during battles) even harder to figure out.

In addition to being able to upgrade a card’s attack and defense stats, cards can also change type. Any card of type P or M can evolve into type X, and any card of type X can evolve into type A. These evolutions follow this hierarchy, which makes perfect sense since type X is objectively superior to types M or P, while still being inferior to type A:

Unlike the previous evolutions, which depend on winning an entire game, cards can change type each time they win a battle (whether they are on the offensive or defensive side). Specifically, when a card wins a battle, a P or M type card has a 1 in 32 chance to evolve into an X card, and an X card has a 1 in 256 chance to evolve into an A card.

Even though the chances of evolution (of card type) are relatively low, unlike the increases to offensive or defensive points, type changes are much easier to track (since, logically, the type is clearly displayed on the card). When we get to the final section of this article — the collector’s level — we’ll see that reaching level A with all cards is necessary, which makes it all the more important to clarify the method for getting a card to reach that level.

The rules of the game and the logic behind card evolution no longer hold any secrets for us—so it’s time to tackle the final mystery: the infamous Collector Level. When opening the card menu (from the main menu), you can view your entire card collection and see several statistics displayed.

• the number of games won
• the number of games lost
• the number of draws
• the level, shown in p

The first three pieces of information shouldn’t (in principle) require any clarification. The game keeps track of played matches and provides a summary through this menu. However, what really draws our attention is the level. Indeed, by observing how this level evolves over time, like many elements of Tetra Master, it’s very difficult to understand how this counter changes (once again, we’re faced with a rather convoluted rule).

Even more amusingly, you can quickly realize that the three summary elements (wins, losses, and draws) don’t actually affect this score. Indeed, there are several situations where the outcome of a match won’t change the level. And yes, as is often the case with this game, why settle for something simple and deducible when you can choose a metric that’s completely all over the place? To put it simply, the level is determined by the cards you own, according to a very specific mechanic that we’ll now detail without further ado.

Before explaining how to calculate points based on the cards you own, here is the list of different levels you can reach in the game. Remember, your card level has no impact on the actual gameplay.

Funny to have so many progression steps for a point system that's so hard to figure out, isn’t it? Once again, we can grumble about the game’s excess of implicit rules, where every design choice seems made to confuse the player. As we’ll see in the upcoming sections, the points calculation is far from trivial!

Before diving into the correlation between the cards you own and the points they yield, it’s important to clarify some terminology. Here are the key elements that define a card for the purpose of point calculation:

• its figure, for example Goblin or Fat Chocobo;
• its arrow pattern;
• its type (X or A).

Cards of type P or M do not provide any bonus points, and the card statistics (including their evolution) do not affect the point calculation.

So when we talk about a unique card, we mean a card with a specific figure that you own only once. When we talk about an additional card, we mean a card with a figure that you own multiple times.

Points are calculated based on the cards you own and how their arrows are arranged (ouch). Schematically, we can define the axes that earn points as:

• the diversity of cards you own. Simply put, you can only hold 100 cards at once, and there are exactly 100 card figures (like Goblin). So, to maximize points, you need to have only different figures;

• the diversity of arrow patterns on the cards you own. To maximize points, each card should have a unique arrow pattern (no, this is not a joke);

• the power level of the cards you own. Cards marked X and A give bonus points, so to maximize points, all cards should be level A (which gives the highest points).

Here’s how a card can earn points — these attributes are cumulative (except, of course, for the card type, because logically, a card can’t be both X and A at the same time):

A card can have between 0 and 8 arrows. In other words, for each of the 8 slots, there may or may not be an arrow, which corresponds to 256 possible permutations. We can easily verify this by representing each slot as a boolean—0 or 1. If all slots are filled, we get a sequence of eight 1: 11111111, which is the binary representation of 255 (since binary 11111111 equals decimal 255). So actually, there are 256 possible combinations from 0 (no arrows) to 255 (all arrows).

We can even describe, by number of arrows, the number of possible combinations with this formula:

This gives us the list of possible combinations for each number of arrows, showing that we don’t have to rely on weak patterns to reach the highest level. In fact, there are 163 combinations including at least 4 arrows:

While to reach the maximum, you need all the cards and they must all be level A, you only have to worry about about ~39% (100 out of 256) of the different arrow patterns. Nice.

If you’ve been paying attention, you might have noticed something. We said each card could earn a maximum of 17 points, and there are 100 cards, so you’d expect a total of 1700 points. But the level table stops at 1698. So, what happens if you actually reach 1700? You might expect a big surprise, like a rare item or a congratulatory NPC. Well, the answer comes from ik141 (as reported in the sombre document; unfortunately, I couldn’t find the original message), who took the time to discover the maximum achievable level in Tetra Master:

So what happens at 1700, you ask? Well, under your collector pts. where Master was written it now says, superimposed over everything, 'Would you like to discard?'. So, after a hundred hours of playing this game I am rewarded with a F***ING GLITCH!!!"

I’m delighted to end this long dive into the rules of Tetra Master on such a cheerful note. You’re not dreaming — the guide where this testimony of frustration was recorded is roughly contemporaneous with the game’s release. ik141’s feat is truly impressive because:

• he didn’t have the guide I used to understand the rules;
• the state of emulation back then wasn’t nearly as advanced as it is today. So, we can reasonably assume (though it might just be speculation) that this card prodigy tried to complete the game, including the Tetra Master quest... the old-fashioned way.

A chapter closes. I discovered FFIX over 20 years ago, and every time I tried, Tetra Master always left a bitter taste in my throat. After two attempts (one in 2012 via emulation and another in 2020 on Nintendo Switch) to understand the rules of the game within the game, I gave up and turned to the internet for help. It was a very entertaining investigation that made me question the true quality of the game. Diving into the rules of Tetra Master altered the conclusion I thought I had when I started writing this article. Having been a huge fan of the ninth installment of this saga that marked me so deeply, my frustration and bitterness towards Tetra Master grew. When I first began researching, I was furious about what I read and convinced my conclusion would be scathing. However, probably because I developed a sort of Stockholm syndrome, since learning the rules, I’ve developed a morbid fondness for Tetra Master. I’m very happy to have finally learned how to play it.

To wrap up this article (which took me over 15 days to write, though I took plenty of breaks), I’ll try to give a balanced opinion on the game.

It’s quite fun to develop an interest in a game by laying out its rules because they were originally incomprehensible. So, it’s possible I won’t stop here. In fact, writing this article has given me the desire — if time allows — to create a free implementation of the game (though it’s likely this project will never see the light of day, due to lack of time and motivation).

More than twenty years later, a chapter closes. It’s hard to see myself as an adult without having understood the rules of Tetra Master. Now it’s done! I can finally focus on less interesting and less formative projects! Jokes aside, I really enjoyed this work of archaeology, writing (and a bit of implementation). It was a lot of fun, and I’ll try to be more consistent with my writing, maybe continuing to describe game mechanics! Thanks again for reading, and enjoy the heatwave!