0xtalks / alt.fun decoded
whitepaper breakdown · may 2026

Memecoins that pump on their own.

The reserve isn't SOL. It's a 2x/3x/5x perp on BTC, HYPE, or a stock. Your bag has a heartbeat now — it ticks with the underlying, even when nobody trades. Right regime → the curve graduates almost by itself. Wrong regime → time bleeds you out. la riserva non è SOL. è un perp leverato Hyperliquid. il tuo bag ha un battito anche quando nessuno scambia.

opening market cap
$4,000
constant by design
graduation threshold
$12K USD
or 750M tokens sold
leverage tiers
2x · 3x · 5x
long or short
LP at graduation
100% locked
forever, no recovery
scroll to decode
tl;dr

In 30 seconds.

Three things. Don't scroll past if you'll skip the rest. tre cose. se salti il resto, almeno leggi queste.

01 — what

Your reserve isn't SOL. It's a 5x perp.

On pump.fun the reserve is spot SOL. Here it's a BounceTech Leveraged Token — an ERC-20 that wraps a leveraged perp on Hyperliquid. The token's price moves even when nobody trades.

02 — why new

Time is no longer free

The reserve moves with the underlying. Trend in your favor → graduation can fire without a single new buyer (curve almost frozen). Choppy market → vol decay drains the reserve: graduation drifts away and everyone bleeds.

03 — the catch

4 systems can rug you before alt.fun does

Hyperliquid (oracle, CLOB) · HIP-3 (builder-chosen oracle) · BounceTech (NAV, redemption capacity) · alt.fun (smart contract, MEV). Safe? No. More transparent than standard pump.fun? Yes.

the paradigm shift

What's different from pump.fun.

Same math (constant product, bonding curve). Different physics. stessa matematica, fisica diversa.

pump.fun · spot reserve

Time is neutral

  • Reserve = SOL (spot). Worth what SOL is worth.
  • Token price moves only via trade flow on the curve.
  • Waiting costs nothing. Diamond hands is a valid strategy.
  • Historical graduation rate: ~1%. >98% of tokens die.
p_token = (y / x) · p_SOL
alt.fun · leveraged reserve

Time costs you or pays you

  • Reserve = BounceTech LT: a leveraged perp position.
  • Token price moves with trade flow + underlying × leverage.
  • Waiting in chop drains the reserve (vol decay).
  • Graduation can fire with zero buyers if the underlying rips.
p_token = (y / x) · p_LT
key insight p_token = y · p_LT / x has two independent drivers. On pump.fun you only get one (y/x, trade flow). On alt.fun you get two: y/x (trading) + p_LT (underlying × leverage, ticking on every oracle update from the Hyperliquid perp). That single line is the entire whitepaper.
the stack

Under your memecoin sit 4 layers.

Click a layer to expand. The stack is vertical and additive: every layer adds functionality, and also a dependency you inherit. clicca un layer per espanderlo. ogni layer aggiunge una dipendenza.

🌍 Real-world asset
underlying
BTC, HYPE, equities, indices, gold. Quoted 24/7 (even stocks, after HIP-3).
↓ oracle (Pyth / RedStone)
HIP-3 perpetual
hyperliquid
A perp market deployed by a "builder" (trade.xyz, Felix, etc.) on the Hyperliquid CLOB.
↓ USDC margin in vault
🎚 BounceTech LT
erc-20
Wraps the perp into an ERC-20 with fixed 2x/3x/5x leverage. Atomic mint/redeem.
↓ used as bonding curve reserve
🪙 alt.fun curve
launchpad
x · y = k constant product. The "y" side is LT, not SOL/USDC.
↓ graduation
💧 post-graduation AMM
post-grad
Liquidity migrates to a standard constant-product AMM. LP locked forever.
pricing, in two strokes

One formula. Two engines.

A token's USD price is the product of a "curve ratio" and the LT's NAV. Sounds obvious. It is the entire mechanism. due motori spingono il prezzo: trade flow e underlying drift.

token price in USD p_token = y · p_LT / x
engine 1 · trade flow

y/x

Moves only when someone buys or sells on the curve. The classic "pump.fun behavior".

buy → x ↓  y ↑  → p ↑
sell → x ↑  y ↓  → p ↓
engine 2 · underlying drift

p_LT

The LT's NAV. Ticks on every oracle update of the Hyperliquid perp, amplified by leverage.

underlying ↑ → p_LT ↑ (× leverage)
underlying ↓ → p_LT ↓ (× leverage)
not obvious Opening MC is always $4,000 regardless of which LT you pick. Proof: substitute into the formula — p_token,0 = (V/p_LT)·p_LT / S = V/S, so MC₀ = V = $4,000. Initial price-per-token is identical across all launches. It's a deliberate design choice — tokens are comparable from the second they exist.
sandbox · 01 / 03

Try it: two drivers, one price.

Move the sliders. Watch trade flow and underlying drift combine. Leverage only amplifies engine 2 — not engine 1. muovi gli slider. vedi come i due motori si combinano.

Two-driver price simulator live

p_LT,0 = $1.00 · S = 10⁹ · V = $4,000 · constant-leverage rebalancing
3x
LONG
+10%
25% (250M)
USD price per token (live)
$0.0000080
+100.00% from launch
trade engine alone
+33.3%
drift engine on top
+30.0%
how to read Engines compose multiplicatively: (1 + trade) · (1 + drift) − 1 = total. They don't add.
distance to graduation
$8,400 to go
USD trigger · $12,000 target
the bonding curve

Constant product, with one twist.

The "twist" is the virtual reserve: the contract accounts for more liquidity than it actually holds, giving the curve a sensible opening price without needing real USDC seed. le virtual reserve permettono alla curve di partire a un prezzo sensato senza liquidità reale.

invariant x · y = k
launch setup x₀ = S = 10⁹ tokens
y₀ = V / p_LT = $4,000 / NAVLT
k = x₀ · y₀
virtual vs real The pair contract actually holds 0.75·S = 750M tokens (curve reserve) and 0 LT. The remaining 250M tokens (25%) sit outside the curve — they seed the AMM at graduation.
buy

Pay ΔLT, receive Δtoken

(x − Δx)(y + Δy) = k
→ Δx = x · Δy / (y + Δy)

First you mint LT from BounceTech with USDC, then spend it on the curve. The UI bundles it in a single tx.

sell

Spend Δtoken, receive ΔLT

(x + Δx)(y − Δy) = k
→ Δy = y · Δx / (x + Δx)

ΔLT is atomically redeemed for USDC at NAV. If the LT's idle USDC buffer is depleted, the sell tx reverts.

sandbox · 02 / 03

Walk the curve.

Slide tokens-sold and current LT NAV. See price, reserve, and how the two graduation triggers race each other. muovi gli slider: vedi prezzo, riserva, e come i due trigger competono.

Bonding curve walker live

S = 10⁹ · V = $4,000 · T = $12,000 (default)
200M
20.0% of supply · 26.7% of curve reserve
1.00×
flat · LT at launch value
current state
on curve · 26.7% sold
graduation fires when EITHER trigger hits
USD trigger
$1,330 / $12K
11% to $12K
supply trigger
200M / 750M
27% to 750M
graduation

Two triggers. Only one needs to fire.

First one to hit wins. Graduation is permissionless — anyone can close it, the protocol can't block it. due strade alla graduation, in OR. chiunque può triggerare.

01
USD trigger
captures: underlying rips
y_real · p_LT $12K

The LT held in the curve crosses the USD threshold. Even with a near-frozen curve, if the underlying pumps, p_LT rises and the product breaks the threshold.

example Curve at 50% sold + HYPE pumps 30% on a 3x long LT → reserve crosses $12K with zero new buyers.
02
supply trigger
captures: flat market, heavy flow
x_real = 0

All 750M curve tokens have been sold. The curve empties regardless of p_LT. Classic pump.fun-style "buy frenzy".

example Underlying flat, but viral narrative + KOL pump → all 750M get bought, graduation via supply.
migration math At graduation, the AMM pool opens exactly at the last curve price. The formula t_LP = ℓ · r₀ / r₁ forces the same reserve ratio onto the new pool.
where the math gets pretty

Why exactly 25% of supply?

The protocol holds 25% of supply outside the curve. It's not arbitrary. It's the maximum of a parabola — proven, not hoped. il 25% non è scelto, è il massimo di una parabola dimostrata.

tokens needed to seed the AMM, as a function of s t_LP(s) = s · (S s) / S

Maximize by differentiation: d/ds = (S − 2s)/S = 0 → s* = S/2. Plug back in: t_LP(S/2) = S/4 = 25%.

For any path to graduation — few buyers + big underlying pump, or zero pump + many buyers — the protocol has always enough tokens to seed the AMM. Anything not needed gets burned.

Notable: p_LT doesn't appear in the final formula. It cancels out in the derivation — sufficiency is pure unit math.

x: tokens sold (s) · y: tokens needed to seed AMM (t_LP)

5 properties

What makes alt.fun safer than pump.fun.

Five design choices that move trust away from the team. cinque scelte di design che spostano la fiducia via dagli operatori.

prop · 01

Anti-snipe 3-block window

Trades gated for the first 3 blocks via EIP-1153 transient storage. Bots can't buy in the same block as the deploy → narrows (doesn't kill) the sniper-vs-human asymmetry.

prop · 02

CREATE2 suffix mining

Every alt-coin deploys to an address that ends in a fixed suffix. The contract rejects non-matching deploys. Frontends and indexers filter fakes at the structural level — no whitelist lookup needed.

prop · 03

LP permanently locked

LP tokens minted at graduation get sent to a custodian with no withdraw(). Liquidity locked forever. No governance recovery path. Anti-rug, structurally.

prop · 04

Permissionless graduation

Anyone can close graduation once the trigger fires. The protocol runs a backstop keeper, but it's not a privileged actor. The protocol can't censor a graduation.

prop · 05

Zero-arb migration

The AMM's first tick = last curve price. No bot snipes the graduation arbitrage gap. Sounds like a detail: in practice it redistributes MEV to late-curve buyers.

5 risks

5 ways you get cooked that didn't exist on pump.fun.

Whitepaper-listed. Here with the numbers that hurt. cinque rischi che con la riserva spot non avevi.

risk · 01

In chop, you lose ~2.5%/day doing nothing.

A constant-leverage instrument that doesn't liquidate must rebalance. Rebalancing in chop converts realized vol into NAV bleed. Approx: decay ≈ −½·L·(L−1)·σ²·dt.

5x at σ=5% daily: ~−2.5%/day in pure chop
risk · 02

Your reserve can drift to zero with zero sells.

The LT never liquidates, but in a sustained crash NAV converges asymptotically to zero. The curve's reserve loses USD value — even with zero sells on the curve.

gets worse with leverage. 5x is the most fragile.
risk · 03

Graduation can un-happen.

Distance to USD trigger can grow back. At 80% of threshold, a −20% move in p_LT drops you to 64%. No ratchet. Sitting in a marginal position is negative-EV.

graduation is not a one-way ratchet
risk · 04

Sells revert when everyone runs.

LT redemption is atomic, capped by the LT's idle USDC buffer. In a mass-sell event the buffer depletes → sell txs revert. Sellers face asymmetric execution risk.

an async queue would absorb pressure but add latency to every trade
risk · 05

4 systems can rug you. None of them is alt.fun.

Hyperliquid downtime → oracle frozen. HIP-3 builder compromised → bizarre mark price. BounceTech vault exploit → LT goes to 0. 4 layers = 4 single points of failure.

the design doesn't insulate against any of these
sandbox · 03 / 03

How much does waiting cost?

Drift (trend) compounds in your favor. Vol decay bleeds you in chop. See how much NAV remains in 30 days. drift vs decay. quanto il tempo mangia (o nutre) la riserva.

Vol decay calculator NAV trajectory

NAV_t ≈ NAV_0 · exp((L·μ − ½·L·(L−1)·σ²)·t)
5x
4.0%
0.0%
30 d
final NAV
82.5%
−17.5% over 30 days
break-even drift
+0.40%/d
daily trend that cancels the decay
read this The break-even drift is the minimum daily trend the underlying must move in your direction to offset decay. Below that, you bleed even with zero sells — time costs ½·L·(L−1)·σ² per day.
game theory

How to play it, by role.

Same whitepaper, three players, divergent optimal moves. That divergence is where the edge lives. stesso paper, tre attori, mosse ottimali diverse.

buyer · degen

Buy momentum + leverage.

Entry signal: a strong and recent directional move on the underlying, not just meme narrative.
Time preference: short. Waiting in chop = guaranteed decay.
Skip high leverage if the underlying is choppy. Pick 2x in flat regimes, 5x only in strong trends.
Watch redemption: before big sells, check the LT's idle USDC buffer. Mass-sell on illiquid perps → txs revert.
creator · launcher

Launch with the wind, not against.

Direction × regime: 5x long on an underlying that's dumping = technical death of the token, regardless of meme quality.
Underlying choice > meme choice. Pick a HIP-3 perp with real flow and active narrative (e.g., HYPE, NVDA, indices).
Timing: launch right after a breakout or catalyst, not before.
Leverage: 3x is the sweet spot. 2x decays little but scales slowly; 5x scales hard but decays just as hard.
observer · researcher

Find the mispriced setup.

Short setup: an alt-coin just launched with direction opposite to the underlying's trend — short half-life.
Long pre-graduation: token at 70%+ of threshold with underlying trending your way → graduation imminent, AMM opens with LP locked forever.
Alternative: if you only want the leverage, mint LT directly from BounceTech and skip the momentum game. The curve adds social, not guaranteed alpha.
Track health: HIP-3 oracle uptime, BounceTech buffer levels, Hyperliquid open interest on the underlying.

The point, distilled.

alt.fun isn't "pump.fun on Hyperliquid". It's a different mechanism: the reserve is an active financial instrument. Buying an alt-coin means buying two things at once wrapped in a single ERC-20, a price that ticks even with zero trades, and the chance that time works for you or against you depending on the macro regime of the underlying the creator picked.

What makes alt.fun an existential bet isn't the curve mechanics — it's whether memecoin traders will accept that time is no longer free. On pump.fun the median holder survived by waiting. Here they don't. For real degens: more edge for the chart-readers, less edge for the pure Twitter-trackers.

if you remember one thing Pump.fun rewarded patience. alt.fun rewards reading charts. Pick your timeframe carefully — the curve doesn't wait for your vibe.
fine print · what's verified, what's not
  • Whitepaper PDF May 2026, alt.fun official. x·y=k, p_token = y·p_LT/x, t_LP(s) = s(S−s)/S verified by re-derivation.
  • Discrepancy: whitepaper states T = $12,000 (default); docs.alt.fun (snippet) mentions $9,000. Verify in-app.
  • Discrepancy: BounceTech marketing says "no rebalancing up to 10x"; alt.fun whitepaper acknowledges rebalancing-driven decay. Exact schedule (opportunistic mint/redeem vs daily) not published.
  • Underlyings live in BounceTech private beta as of research date: BTC, ETH, SOL, HYPE, PAXG. The "stocks/indices via HIP-3" claim is infrastructurally enabled (trade.xyz already lists TSLA/SP500) but not yet wrapped by BounceTech.
  • Vol decay formula ≈ −½·L·(L−1)·σ²·dt is the classical leveraged-ETF approximation, not BounceTech-specific. Real numbers depend on exact rebalance schedule.
  • alt.fun team: anon-default, no public funding round indexed at the time of this study.
  • Single-shot vs continuous: the price simulator (sandbox 01) uses the constant-leverage geometric formula (1+r)^L — accurate for instantaneous moves with rebalanced position. The vol-decay calculator (sandbox 03) uses the continuous-time log-drift approximation.