A football RNG gacha bolted onto the real 2026 World Cup: roll the dice, pick one footballer out of three, build an XI, play a match, build a stadium. This is a teardown of our own game — not a competitor's. Based on a recorded live session from 13.07.2026 (53 minutes, 108 screenshots) and cross-checked against the sources of that exact build.
The main point: the first ten minutes are exemplary — the roll gives you a choice rather than a random card, the match pays, the stadium grows. After that the game stops asking anything of the player: the squad resets after every match, you can roll endlessly and for free, and OVR 87 beats absolutely everyone. By minute 38 of the recording there was no reason left to carry on — and that is the same cliff analytics sees as D2 ≈ 5%.
What this game is, what already works in it, and where it loses the player.
The game already earns and already feels like football: a roll is a decision, not a lottery ticket. But everything that retains fits inside the first ~15 minutes. After that the loop stops pushing back: the squad wipes itself, the dice roll free and without end, the opponents run out. The recording catches the moment where there is no point playing on — and it lines up with the second-day collapse in analytics.
"Roll the dice — build a dream team out of real stars — beat the world champion"
The fantasy works, and it is rare for the RNG genre: a roll gives you a choice of one out of three for a specific position, not a random card into a bag. That is already half a squad builder.
The game dismantles itself: after a match it demands you rebuild the squad from scratch, and the free dice roll without limit.
Attachment to the club never has time to form, and progression meets no resistance. By minute 31 OVR 87 beats the whole opponent ladder; by minute 46 you can walk away for an hour and nothing changes.
The core loop, the first minutes, and where the session breaks.
The loop is assembled better than in a typical RNG competitor: the dice hand you not a card but a choice of three for a position, and every choice moves one legible number — OVR. The loop breaks where it should close: instead of returning the player to his squad, the game invites him to take it apart and build a new one.
Cash goes into the stadium; luck is fuel for the next rolls. The last node is not a metaphor: it is the game's literal instruction after a win.
The first rolls are not random: the game scripts them to teach synergy and to stop you failing. The technique is good, but it also exposes how little the loop steers anything after that.
The first roll in the recording produced Grimaldo, Cubarsí and Gavi — three Spaniards. That is no coincidence: on the first two rolls all three candidates are forcibly swapped for the weakest cards of a single nation, so the player spots the link himself and builds his first bit of chemistry. The game then spells it out: "Put players from the same country together".
| What the onboarding scripts | How it looks in the recording | Why |
|---|---|---|
| Rolls 1–2: three cards of one nation, deliberately weak | Grimaldo / Cubarsí / Gavi — all Spain 03:16 | show synergy before the player starts optimising |
| The nation hint | "Put players from the same country together" 04:33 | explain the green links in the squad |
| The reveal cutscene only on the first rolls | after that the animation gets shorter 05:36 | not to get in the way of an experienced player's pace |
| The first matches are rigged | a 4:0 win straight away 09:39 | give a taste of winning before any risk appears |
What the game actually computes — checked against the screen, not guessed from it.
Card ratings run 72–99, and each rating has its own denominator: the chance of pulling a card = luck / denominator (capped at 1). The tier ("Pro", "Elite", "Legend", GOAT) is a label on the card: it takes no part in the draw. The roster size disagrees across sources — the on-screen index shows 423 players, the team quotes "400+ base cards and ~93 moment skins", and the config gives 472 entries by tier. The difference is most likely about whether moments count as cards; one number is worth pinning down.
This is not a reconstruction — it matches the screen to the letter. Every "1 in N" the game shows on a card is exactly equal to the denominator of its rating:
| Card in the recording | Rating | Shown in game | Denominator in code | |
|---|---|---|---|---|
| Sow, CM | 79 | 1 in 55 | 55 | matches |
| Meunier, CB | 80 | 1 in 70 | 70 | matches |
| Ajer, DEF | 81 | 1 in 100 | 100 | matches |
| Ueda, ST | 84 | 1 in 175 | 175 | matches |
| Fernández, CM (moment) | 90 | 1 in 1,800 | 1800 | matches |
On all five cards legible from the screen, the displayed chance is exactly the denominator of the rating. So the odds display is honest — a rare plus for the genre. But an unpleasant corollary follows: since only the rating sets the chance, the rarity tier tells the player nothing beyond the number on the card. Eight tiers work as a colouring scheme, not as an axis of progression.
The tiers are labelled on the cards like this: Любитель (Amateur), Перспектива (Prospect), Повышение (Rising), Профессионал (Pro), Элита (Elite), Мировой класс (World Class), Легенда (Legend), GOAT. The third in that list is a translation error: Rising in a football sense means "a rising star", whereas «Повышение» in Russian means a pay rise. The player sees a word on the card that describes neither the rarity nor the player.
| Rating | 72 | 75 | 80 | 84 | 88 | 90 | 92 | 95 | 99 |
|---|---|---|---|---|---|---|---|---|---|
| 1 in N | 3 | 12 | 70 | 175 | 800 | 1,800 | 8,000 | 30,000 | 300,000 |
| Chance at luck 7 (start) | 100% | 58% | 10% | 4% | 0.9% | 0.4% | 0.09% | 0.02% | 0.002% |
| Chance at luck 37 | 100% | 100% | 53% | 21% | 4.6% | 2.1% | 0.5% | 0.1% | 0.01% |
| Chance at luck 67 | 100% | 100% | 96% | 38% | 8.4% | 3.7% | 0.8% | 0.2% | 0.02% |
This shows what luck is for — and why its value does not read in the recording. Between luck 7 and 37 the chance of an 84 grows from 4% to 21%, but the player sees neither this table nor a before/after: he sees only the "luck" number and the fact that good cards drop anyway. A roll is never empty — if no group clears the threshold, the game hands over the most common available card.
In the recording it jumps out that some cards carry a "+N" and others do not, and the logic does not read. Here is the logic — and it does not depend on luck at all:
| Mutation | Rating bonus | Chance | Effect |
|---|---|---|---|
| Shiny | +2 | 1 in 20 (5%) | foil |
| Pyro | +3 | 1 in 50 (2%) | fire |
| Aurora | +5 | 1 in 125 (0.8%) | aurora, enlarged VFX |
They are checked from rarest to most common, and the first one that fires wins; the total chance of any mutation ≈ 7.7%. The mutation rolls together with the offer — so the player sees exactly the card he will get. Nothing in the interface explains what it is, or why your neighbour has one and you do not. recording 34:38
The formula is simple and honest: chance = 50% + (your OVR − opponent OVR) × 3%, clamped to 1…92%. Client and server read the same number, so the displayed percentage is real. The problem is not the formula, but where it leads.
| # | Opponent | OVR | Unlock | Luck per goal | Chance at OVR 87 | Chance at OVR 100 (ceiling) |
|---|---|---|---|---|---|---|
| 1 | Academy XI | 70 | — | +2 | 92% | 92% |
| 2 | Sunday League | 73 | $500 | +3 | 92% | 92% |
| 3 | Group Stage XI | 76 | $1,500 | +4 | 83% | 92% |
| 4 | County Cup XI | 79 | $4,000 | +6 | 74% | 92% |
| 5 | Continental XI | 82 | $10,000 | +8 | 65% | 92% |
| 6 | National Side | 85 | $20,000 | +10 | 56% | 92% |
| 7 | Elite World XI | 88 | $35,000 | +13 | 47% | 86% |
| 8 | Champions XI | 90 | $50,000 | +16 | 41% | 80% |
| 9 | World-Class XI | 95 | $80,000 | +20 | 26% | 65% |
| 10 | Legend XI | 100 | $125,000 | +25 | 11% | 50% |
| 11 | All-Star XI | 110 | $250,000 | +30 | 1% | 20% |
| 12 | Galactico XI | 120 | $500,000 | +36 | 1% | 1% |
| 13 | Icon XI | 130 | $1,000,000 | +42 | 1% | 1% |
| 14 | GOAT Squad | 140 | $5,000,000 | +55 | 1% | 1% |
| 15 | Ultimate XI | 150 | $10,000,000 | +70 | 1% | 1% |
Cards in the game run 72–99. So Team OVR tops out at roughly 100. Yet opponents #12–15 have OVR 120–150.
Even a perfect squad of nothing but 99s takes Galactico, Icon, GOAT and Ultimate at a 1% chance — that is, loses 99 times out of 100. And they have to be unlocked for $500,000 … $10,000,000. The last four opponents are scenery with a 16.5 million price tag.
The chance ceiling is 92%. At OVR 87 the first six opponents are already indistinguishable.
The Academy (OVR 70) and Sunday League (73) both give 92%, although a whole rung separates them. The player stops choosing: there is no difference, and the luck per goal is higher on whichever one is unlocked. This is exactly what the recording renders as "OVR 87 is more than enough to beat everyone".
Unlocking all 15 opponents costs $17,076,000.
This is the only substantial cash sink in the game. And in the recording the entire stadium cost ~142,000 — meaning the opponent ladder is 120 times more expensive than the stadium, and it dead-ends in opponents who cannot be beaten.
In sum: with a realistic squad (OVR 87) the meaningful choice is opponents 3 through 8: below that it hits the 92% ceiling, above it falls to 26% and lower. Even a perfect squad of nothing but 99s gets honestly as far as #10 (Legend XI — exactly 50/50) and stalls there: #11 is taken at 20%, and #12–15 at 1%. The last four opponents are unbeatable by design, and together they cost $16,500,000.
Where cash comes from, where it goes, and what we sell.
The design intent is right: a match draws a crowd, the crowd pays for tickets, tickets build the stadium, the stadium holds a bigger crowd. A compounding loop tied to actually playing. The problem is that in the recording the money arrived bypassing it.
| Parameter | Value at start | How it is calculated |
|---|---|---|
| Capacity | 100 seats | grows from stadium pads |
| Ticket price | $5 | grows from pads |
| Fanbase | 100 | +4 per match, +4 per win, +3 per sell-out, +1 for each skin worn |
| Demand per match | 45–110 | fanbase × random 0.45–1.10, capped at capacity |
| Revenue per match | $225–$500 | tickets sold × ticket price |
| The whole stadium | ~$142,000 | 23 upgrades, from $200 to $16,800, 4 levels (gates at 5 and 16 purchases) |
Galaxy Roll is the best thing in the monetisation: a cheap impulse right inside the fun. 96% of all first purchases are it. But 93% of those who walk through buy nothing else, and only 2% ever take a game pass.
| SKU | Robux | What it gives | How it performs in the data | Verdict |
|---|---|---|---|---|
| Galaxy Roll ×1 | 6 | a roll with a Legend floor, luck ×20 | 96% of first purchases; median between purchases 0.4 min; record — 197 rolls one at a time | KEEP |
| Moment Pack | 10 | a random moment skin | the first one is given free | KEEP |
| Cash Stack $8K | 13 | 8,000 cash | sells progress itself, not acceleration | KILL |
| Icon Pack | 20 | an ICON moment skin | ~5,000 Robux, a third of buyers came back | KEEP |
| Featured Moment | 39 | the current rotating skin (rotates ~2h, pool of 8) | a flop: 23 buyers. Expensive next to the pack at 20, and invisible | FIX → 19–25 |
| Galaxy Roll ×10 | 55 | 10 rolls | shipped on 13 July — a rung the digest had not yet seen | SHIP |
| 2× Cash Gain | 109 | ×2 cash from every roll | becomes inert the moment cash printing is switched off | KILL |
| 2× Luck Gain | 119 | ×2 luck from matches | 148 sales in 3 days against 39 in a week at 149 — the price has been found | KEEP |
| Galaxy Roll ×25 | 129 | 25 rolls | shipped on 13 July | SHIP |
| Auto Roll | pass | auto-roll | the same thing is buyable for in-game cash in the upgrade tree — the pass sells something free | FIX |
| Auto Equip Best | pass | auto-fielding the best | gives nothing: auto-equip fires for everyone unconditionally, with no ownership check | KILL |
The digest wrote that between the roll at 6 and the pass at 119 there was not a single rung. By the evening of the recording there are five: 10, 13, 20, 55, 129.
The 10-for-55 and 25-for-129 bundles — exactly what the digest proposed shipping — were already in the shop. So pain #2 on our list needs re-measuring, not solving again: the question now is not "is there a rung" but "does it convert".
"2× Cash" becomes a dud after the economy fix; "Auto Equip Best" is a dud already.
This is not a hypothesis: auto-equip is called on every card added, with no check at all that the pass was bought. Both SKUs are a direct refund and reputation risk, and both are fixable before the World Cup final.
The top 1% brings in 16.7% of revenue against a genre norm of 40–60%.
The most expensive item in the game costs 129 Robux. A big spender has nothing to buy — not because he does not want to, but because there is nothing. Tellingly: the things that would fill the vacuum (VIP, expensive packs, Icon/GoatHunter) already existed and were cut.
The one thing competitors do not have — and the thing we do not have.
The game is bolted onto the real World Cup: a fixture grid with countdowns, a ticker of real results, moments tied to real goals, a bonus to cards of whichever nation won yesterday. No RNG competitor has anything like it. But today it is wallpaper: you can look at it, you can barely act on it.
The live layer demands nothing of the player and rewards him with nothing for coming back.
The nation bonus arrives by itself; you do not need to know about it. There is no "field a Spaniard before Spain play", no "log in after the final", no set closing out. The event happens next to the player, not to him.
The 2026 World Cup final is once-and-for-all content.
A commemorative trophy for everyone online at the final; a limited skin; the WC26 set closing at the final whistle. Miss it and it is gone for good — there will not be a second World Cup. And the plan for it, by the digest's own assessment, is 26 steps with no owners and no dates.
A ~15-minute session is above the Roblox benchmark. And daily time in game of 14–19 minutes is at the bottom of the band. Divide one by the other: 1.27 visits per day. The player enjoys the visit and does not come back the same day.
D7 doubling on the day the stadium and Galaxy Roll shipped is the most honest signal in the whole dataset: shipping an update is itself our retention mechanic. The tails of old cohorts bent upward that day — the dormant came back. It works, but it is not a mechanic, it is manual labour.
| Hook | What it does | State |
|---|---|---|
| "Free luck" | like + favourite + group = +30 luck | one-shot for the lifetime of an account |
| Weekly updates | the only thing that actually lifts D7 | works |
| The World Cup ticker and fixtures | a reason to know what is going on | does not require a return visit |
| Daily reward (7 days: $1K → GALACTICO pack) | the table exists in code | not wired to anything |
| First match | +5 luck, once | ok |
| First 2 matches rigged | guaranteed wins | ok |
| Pushes / drop countdowns / happy hour | — | none at all |
The peaks land on new territory: the first roll, the first match, the first moment card. The troughs are not about difficulty but about its absence: the squad wipes, the dice roll free, the opponents run out. Below is engagement minute by minute across that same 53-minute recording.
How to read this. The green zone on the left is the ramp-up: new territory every ninety seconds. The red zone on the right is the stretch where the curve dropped below the boredom threshold and did not come back. Important: this is one session — and the average session in the game lasts 15 minutes, so the mass-market player leaves at the very peak, before the collapse. He meets the collapse on his second visit. That is precisely what prints D2 ≈ 5%.
Everything is new and everything is free.
Rolls, the first squad, the first win, the first moment card. The onboarding rigs it and does so invisibly. Here the game works without a single complaint.
Passive boredom: the squad resets, the match is watched.
The first "why bother" arrives at exactly the moment the game asks you to break up the team it has just helped you build. Then 45 seconds of simulation with no decisions in it.
Active pointlessness: no resistance is left.
OVR 87 beats every available opponent. The dice roll free and forever. You can walk away for an hour. You cannot lose. The game has stopped asking for anything — and this is not a trough you pull a player out of with a reward, it is the end of the content.
First what the team has already discussed and quantified. Then what is visible in the recording and was absent from those discussions.
These are not fresh observations — they are what the analytics session and the 13 July call already named, quantified and assigned. Here they are gathered with the numbers, the people who raised them and their current state, so it is visible what moved on its own while we were tearing the game down.
| Pain | Numbers | Who raised it | What was decided | State right now |
|---|---|---|---|---|
| The D2 cliff "problem #1" |
D1 ~10% → D2 ~5%. Every cohort halves | Yelnur opened the meeting with it; confirmed by the digest | the economy "with intent, not by intuition": a day-1 cliffhanger → the first 7 days → day 28. Happy hour, dailies (Arda) | in progress, no mechanic yet |
| The leak: rolls print cash | ~170 Robux print the entire $142,000 stadium | the digest + Yelnur independently | "the most urgent fix in the whole plan". Owner Danyil, by Friday | closed the same day — the 22:27 save is already leak-free |
| The 6↔119 cross-sell gap | 96% of first purchases = Galaxy Roll · 93% bought nothing else · 2% took a pass · median time to a pass 12 min | the digest | 10-for-55 and 25-for-129 bundles + a 2× Luck upsell after the third paid roll | the bundles are already in the game — re-measure, do not solve again |
| The whale vacuum | top 1% = 16.7% of revenue against a norm of 40–60% | the digest + Arda + Yelnur | scale (bundles) + scarcity (limited skins 399–999) + completion (era sets). "The answer is never power" | not started |
| Thumbnails < 2% CTR | CTR 1.5% → 1.1%, target at least 2% | Danyil | 5 thumbnails → 3 permanent + 2 experimental. Owner Mike + Yelnur + Martin | Mike's iterations have already failed a round |
| The second daily visit | session ~15 min (above benchmark) + daily time 14–19 min (bottom of the band) = 1.27 visits | the digest | drop countdowns, happy hour, a day-2 gift, a push after the first Legend | not started |
| The top-of-progression cliff | 23 upgrades, 4 levels, ~142,000 cash. "There is almost nothing in the top level". Plus 150–250 Robux = a full Legend XI in an hour | the digest + Yelnur | Martin is expanding the 3D; levels 5–8; keep the Legend floor and track it | partly |
The call set a goal of raising the average session to 30 minutes. The digest wrote the same day that 15 minutes is already above the Roblox benchmark.
If the digest is right, that is a completely different lever: the problem is not "make the session longer" but "manufacture the second visit". Two teams are pulling in different directions, and it has not been resolved.
The digest: "dead forever", the config is empty. The call: "remove them, but maybe keep bags worth 1–3 pads", and the biggest bag buys the entire stadium.
Two incompatible statements about what is even on sale. The recording settles it: a Cash Stack $8.00K at ◎13 is in the shop — so the packs are alive, and the call was right.
The "2%+" goal was set against CTR (1.1–1.5%). But a playthrough rate of 1.87% appears alongside it in the discussion — that is a different metric.
You cannot line 1.87% up next to 1.1–1.5%. The risk: seeing "nearly 2%" where it is actually 1.1%, and closing the task.
Below is what only becomes visible when you sit down and play. This is not "it just seemed that way": for every item a mechanic was found, not a symptom. Grouped by how hard it hits trust.
The badge means not "you have not seen this" but "you have not taken this".
A card is written into the index only when it is chosen. And chosen cards are excluded from future offers. So the two sets almost complement each other: everything you are offered is something you did not take, i.e. "new". The badge lights up on every card, always, and carries no information. Everything hanging off it breaks with it: "duplicates are shown more modestly" and the new-card sound almost never fire correctly.
The "Worse than your …" label exists in code and is suppressed at render time.
The display condition explicitly excludes the "worse" state — such a card arrives with no label at all. Together with the previous two: three labels in a row, and all three can only deliver good news. The player stops reading them fairly quickly — which is what happens in the recording.
What dropped: "Fernández 90, Legend, 1 in 1,800" — what landed in the squad: Fernández 87, World Class.
A moment is a skin on the hero's base card. On receipt the game swaps the moment for the hero's base card and takes that card's rarity. The display promised a Legend at 1-in-1800; the squad got a World Class. Nothing in the ceremony warns about this — hence the hunt for "that 90" who does not exist. The +3 bonus itself is real and does reach Team OVR; exactly one place ignores it — that same "Better than your" flash.
«Вежливый» is not a glitch on one card but proof of method: the player list was run through a machine translator that cannot tell a surname from a common noun. The scrambled portraits probably come from the same place. What needs checking is the entire roster (four hundred-odd cards), not just these two — and first of all the surnames that coincide with ordinary words: Courtois, Silva, Fernandez, Williams, Sow.
| What | Where it is visible | Why it matters |
|---|---|---|
| «УЛУЧШЕНИЯ» (“UPGRADES”) and «ПРОВАЛИТЬ» (“ROLL”) overlap, rendering as УЛУЧШЕНИЯПРОВАЛИТЬ | 15:02 · 19:47 | a layout collision on the game's main button, at a standard 1920×1080 |
| The celebration VFX sits on a cheap card: golden pillars and confetti around a "1 in 3" card, while the rarest card in the offer stands there dull | 08:11 | it breaks gacha's main language — "rare looks rare" |
| The luck total does not add up: "+6 for goals" + "+1 bonus" = 7, displayed as +7.4 | 39:18 | 0.4 out of nowhere; either a hidden multiplier with no visualisation, or a bug |
| Left and right backs are mirror-swapped relative to the pitch | 18:28 | for a football game, conspicuous sloppiness |
| The Invite Friends button does not work | 11:58 | the only viral channel in the game |
| The luck editor explains nothing: a heading, a "0…7" number and a bar. No text, no confirmation, no preview of the effect | 12:16 | luck is the central currency, and its main screen says nothing about it |
| In the upgrade tree, locked is indistinguishable from unaffordable: a node costing 1,000 at a balance of 1,610 is shown dark, and nowhere does it say what needs unlocking | 13:02 | the player reads it as "expensive", not as "a prerequisite is needed" |
This is the cheapest and most underrated item in the whole teardown. There is no localisation table in the game — Russian is assembled by Roblox auto-translation straight from the English strings. The result is in places not merely clumsy but meaning-changing. Neither the digest nor the call raised it even once.
The entire shop ("SKINS PACKS", "A random MOMENT skin to wear on your card", "instant cash — spend it anywhere", "premium rolls — every card LEGEND or better"). The upgrade tree tooltips ("+8% luck on every match. Rarer cards, better pulls."). Buttons: CLAIMED, BUY, CLOSE, PLAY, TAP TO OPEN. Half the match reward table: «Забито голов» (“Goals scored”) and «Бонус» (“Bonus”) are in Russian, Tickets Sold and New Fans are not. Every stadium structure. The entire news ticker and part of the commentary feed. The value inside the luck indicator: the heading «УДАЧА» (“LUCK”), and inside it 7 / 7 luck · 100%.
A separate document, a separate project. Here: what it closes and what it does not.
The target shift is simple: from a gacha machine to a squad builder. The squad lives for weeks, rolls feed it rather than re-rolling it; money comes from matches; wins set the pace, not luck. Below is an honest map: what is proposed, what is already being prototyped, and where the holes in the package are.
| Pain | What is proposed | The proposal's key number | State |
|---|---|---|---|
| The D2 cliff | a persistent squad + a start at OVR ~40 with a slow curve + a weekly season + a push after the first strong pull and before the season closes | the goal is "~2 hours a day across a week" instead of "everything in 2.5 hours" | the luck knob is done, the mechanic is not |
| The cash leak | switch off printing on rolls, money only from matches, the stadium priced in wins | a single flag flip | done by the developers on 13 July |
| The 6↔119 cross-sell | a ladder of packs between 6 and 119 + position packs "a bit pricier than a random one" + the next tier for Robux | not a single Robux price in any document | structurally closed, numerically not |
| The whale vacuum | bring back what was cut + a whale ladder + status content (limited stadiums, trophies, WC26 once-and-for-all) + an endless pool of 100–110+ packs | "the thing that would fill the vacuum already existed and was cut" | not started |
| Thumbnails <2% CTR | — | — | GAP: zero mentions |
| The second daily visit | season and drop countdowns + pushes | the cadence is weekly, not evening-by-evening | GAP: the lever is blocked by our own principle |
| The top-end cliff | 100–110+ packs for Robux only + rebirth + a season + icon PVP + a collection by ranges | "6 tiers are the basis; beyond that the pool expands almost without limit" | not started |
On CTR, cover art, traffic and discovery, the five redesign documents contain zero lines.
The whole package is about the core loop and monetisation. Yet a CTR of 1.1–1.5% against a 2% target is a pain that constrains entry, not retention: however much you fix the loop, nobody walks into it. The task formally sits with Mike, but it is not in the package and its iterations have already failed a round.
The proposal states it outright: "Nothing is earned while idle".
It is an elegant and correct principle for an economy — and it also blocks the main lever of the second visit. The genre classic (it has accrued over 3 hours — come and collect) becomes forbidden. What is left is weekly timers and pushes, i.e. a treatment for "tomorrow" and "this week", whereas the metric is about "this evening". It is a deliberate trade-off, but nowhere is it named as one.
The client was told "6 tiers = 6 stadium levels". In the prototype the tiers are decoupled from the levels, of which there are still 4.
Done deliberately, to avoid modelling new 3D. But the proposal reads otherwise, and it is worth saying so out loud before someone notices. Same place: the proposal describes adding "low" cards at 40–71, while the plan cancels that in favour of a remap.
Below are our changes, not the game. They are not in production. They are listed so it is visible what has already been tested against the code and what is still just text.
| Phase | What | State |
|---|---|---|
| 0 | Sources under git + write-back into the place file | done |
| 1a | The master luck knob: a scale up to 40, a convex two-parameter curve, three comment-vs-code discrepancies removed | done |
| 2a | Tier ↔ effective rating; the opponent ladder remapped 70–150 → 40–100 so the displayed chance becomes true | done |
| 2b | The roll shows the card at its tier (visuals are placeholder) | done |
| 2c | The persistent squad — the biggest mechanical shift and the package's central promise | not done |
| 3 | Three coefficients, rebirth, a weekly season, the "icon" | not started |
| 4 | PVE depth: a "tier × class" matrix, a positional penalty | scale rebalance only |
| 5 | Dailies, the collection, a battle pass, weaving in the live World Cup, PVP | not started |
A cross-check against the cards on neighbouring RNG games. The main finding is an unexpected one: two of the five tricks are already implemented in our game, the player simply does not know about them. That is cheaper than building.
| Trick (whose) | What it gives | Our state | Verdict |
|---|---|---|---|
| Stacking card mutations Spin a Soccer Card |
a layer of variation on top of the card: a reason to re-roll what you have already assembled, and something to sell whales through chance rather than through power | already there: Shiny +2 (1 in 20), Pyro +3 (1 in 50), Aurora +5 (1 in 125) with VFX. But explained nowhere — in the recording the "+N" reads as randomness | explain, do not build |
| Cohesion / chemistry FC |
links by nation/club/league give +rating to the whole team: depth in squad building and a reason to go for nation packs | already there: +2 for each neighbour of the same nation, stacking, plus +2 to cards of a nation that won a real World Cup match in the last 12h. Visible as "base 73 +4 synergy" | extend to club/league |
| Tier packs with a rating floor FC Mobile |
predictable "no lower than N" packs instead of "1 in N": steady OVR growth and ready-made price rungs | no. There is only the Legend floor on Galaxy Roll | take it |
| A free roll on a timer Lucky Spins |
a gift every N minutes — a direct lever on the second daily visit | no. And our redesign package forbids it, via the "nothing is earned while idle" principle | allow an exception |
| A "pay for luck" whale ladder | fill the whale vacuum through rate/chance/status, not through power | no. The spend ceiling is 129 Robux. Meanwhile a ready-made ladder lies dormant in the code: a 20-level "mutation machine" | take it |
A conclusion worth stating out loud: mutations and chemistry are not development tasks but explanation tasks. The mechanics sit in the code and are counted into Team OVR, but the interface says nothing about them, so for the player they do not exist. The cheapest gain in depth in the whole document.
What the product is worth and what to do first.
This is not a prototype — it is a working game with a unique trump card and real revenue. Everything below is about the fact that it stops exactly where it ought to begin.
the roll as a decision + OVR as a single number — stronger than the genre competitors
the onboarding rigs it invisibly and teaches synergy without a word. The best part of the game
an honest "match → fans → tickets" model, but in the recording the money bypassed it
an excellent conversion door, a whale vacuum, and two passes selling nothing
the only working mechanism is shipping updates by hand
a dense HUD, three lying labels, machine Russian
the live World Cup — something nobody else has. Used at a quarter of its potential so far
The game wins the first fifteen minutes and loses everything after that. The cause is single and structural: the loop does not push back. The squad wipes itself, the dice roll free and forever, the opponents run out at the seventh of fifteen, and the last four cannot be beaten in principle. There is nothing left for the player to want — and that is exactly what analytics sees as D2 ≈ 5% and what the recording catches at minute 31.
What to do, in order of cheapness:
And one observation outside the list: two of our six pains have already moved on their own while we were tearing the game down. The leak was closed the same day. The rungs between 6 and 119 Robux — the bundles at 55 and 129 — are already in the shop. They need re-measuring, not solving again.