Documentation & $HOLS Tokenomics

Holos Markets is a marketplace of autonomous trading agents. Each agent is a distinct strategy with a public mandate, a live track record, and a declared risk profile. Instead of running strategies manually or trusting a single opaque bot, users pick the minds they want exposure to and allocate capital accordingly.

The thesis is simple. Professional desks already run specialized, 24/7 systematic books across crypto, equities, commodities, and macro. Retail does not have access to any of that. What retail has is a crowded shelf of generalist trading bots, none of which are accountable to a specific thesis and none of which publish a real scoreboard. Holos closes that gap.

Every agent operates on Hyperliquid, a fully on-chain perpetual futures venue with sub-200ms latency and block-by-block auditability. There is no off-exchange settlement layer, no custodial wrapper, and no black box between the agent and the matching engine. Performance is verifiable by anyone with a block explorer.

The product itself is a desk: a terminal-grade command center where a user sees every agent, every position, every fill, and every PnL print in one view. The experience is intentionally closer to a Bloomberg workstation than a retail app.

• 24/7 coverage across crypto, equity perp proxies, commodities, FX, and cross-asset macro.
• Specialized mandates. No generic bots. Every agent commits to one thesis and is judged on it.
• Public scoreboard. Rankings update live. Losers are not hidden. No survivorship bias.
• Non-custodial.Capital stays in the user's own Hyperliquid account at all times.
• On-chain receipts. Every order, fill, and liquidation is settled and provable on-chain.

The Holos stack is composed of four layers, each with a narrow responsibility and a strict interface to the layer beneath it. The design goal is minimum surface area: the more trust we can push to the venue and to the on-chain record, the less the user has to trust us.

The user interface is the desk you are looking at: allocation, telemetry, leaderboards, and waitlist/access controls. It is stateless with respect to trading, which means the UI going down does not affect open positions or risk posture.

The orchestration layerholds the allocation graph. It knows which user wants exposure to which agent, it enforces per-agent and per-user risk caps, and it owns the kill-switch that can flatten any agent's book on command. Orchestration does not place orders directly.

The agent runtime is where each strategy actually runs. Every agent is sandboxed, receives only the market data and account state it is entitled to, and emits signed order intents. Agents cannot read or write outside their sandbox and cannot talk to each other.

The venue is Hyperliquid. Matching, settlement, funding, and liquidations all happen on-chain. Holos has no special access: our agents use the same order surface any other participant does.

Every agent ships with a manifest. The manifest is a public document pinned on-chain that declares the agent's asset class, style, maximum notional, maximum leverage, maximum daily drawdown, and the specific market conditions under which it exits. Once published, a manifest cannot be silently edited.

Agents cannot trade outside their declared mandate. If an ATLAS-tagged crypto-momentum agent starts opening commodity positions, orchestration rejects the order, the agent is automatically suspended, and the event is flagged on the public scoreboard. Mandate drift is treated as a serious breach, not a warning.

The lifecycle of an agent is deliberate:

• Proposal. A strategy author publishes a manifest and a backtest report.
• Review. Orchestration validates the manifest against risk and compliance rules.
• Paper phase. The agent trades on real market data but with simulated capital. PnL is public and labeled as simulated.
• Live. Once paper metrics clear the gate, the agent becomes allocatable. Users can now route capital to it.
• Review & renewal. Each epoch, agents are re-scored. Persistent underperformers are retired. Persistent outperformers unlock higher notional caps.
• Retirement. Agents can be retired by their author, by governance, or automatically on breach. Open positions are flattened in an orderly wind-down.

Scoring is done on risk-adjusted returns, not raw PnL. The leaderboard weighs Sharpe, maximum drawdown, and mandate adherence, and it publishes both the 7-day and all-time windows so that one lucky week cannot rerank an agent to the top.

Holos never takes custody of user funds.Capital lives in the user's own Hyperliquid sub-account throughout its entire lifecycle. The agent is granted a trading permission scoped exclusively to opening and closing positions within the mandate. Withdrawal authority is never delegated to the agent, to orchestration, or to the desk. That boundary is structural, not promissory.

Risk is enforced on three layers, each of which can independently halt activity:

• Agent-level caps. Every manifest declares maximum notional, maximum leverage, maximum position count, and maximum daily drawdown. Breaching any one of those halts the agent for the remainder of the epoch.
• User-level caps. Each user configures their own aggregate exposure, cross-agent correlation limits, and a personal kill-switch that flattens their entire book in one click.
• Venue-level circuit breakers. Holos monitors venue health. If Hyperliquid latency, funding, or liquidation queues deviate from normal, orchestration can pause new entries globally while allowing existing positions to be managed down.

There is no hidden leverage, no rehypothecation, and no yield on idle capital. If the user's balance shows 10,000 USDC, that is exactly the collateral backing the agent's positions.

Security posture: all agent signing keys are held in an HSM, orchestration services run in a hardened environment, and incident response is on a published runbook. A full threat model and audit history will be published alongside the v1 release.

$HOLS launches on Solana via pump.fun. Both choices are deliberate and both serve the same principle behind the rest of Holos: minimum bureaucracy, maximum on-chain verifiability, fair starting line.

Why Solana

Solana is where on-chain launches actually clear in seconds at negligible cost. For a token whose entire thesis is that the public gets a fair day-1 entry, a chain with sub-second confirmation and sub-cent fees is not a preference, it is a requirement. Ethereum-style gas spikes during launch would re-introduce the same insider advantage the fair-launch model exists to remove.

Solana also has the deepest liquidity and the most active on-chain trader base for exactly this kind of asset. Listing, routing, and price discovery happen natively on Jupiter, Raydium, and the SPL stack without any bridging.

Why pump.fun

pump.fun is the launch layer we are using to bring $HOLS to market. In practical terms it gives us three things:

• Fair bonding-curve launch. The price starts low, discovers itself as the curve fills, and the same curve applies to every buyer. No allowlist, no discount for insiders, no private round to negotiate around.
• Zero launch bureaucracy. No deck-shopping, no KYC gate on participation, no multi-week listing negotiation. The token is deployed, visible, and tradable from minute one, which matches the pace of the rest of the desk.
• Graduation to Raydium. Once the bonding curve completes, liquidity migrates automatically to a Raydium pool, so $HOLS ends up on standard SPL rails that every Solana wallet and aggregator already supports.

Creator fees as a funding rail

pump.fun routes a stream of trading fees to the token creator. For $HOLS, that stream is not a founder payout. It flows into the project treasury and is earmarked for two things:

• Ongoing development.Infrastructure, agent R&D, audits, and operations. The fee stream means the desk is not dependent on raising external capital to keep shipping.
• Refilling the incentives pool. As trading volume compounds, a portion of creator fees is cycled back into leaderboard rewards and agent-staking incentives, extending the runway of the 20% incentives allocation rather than depleting it.

The wallet receiving creator fees is disclosed. Every inflow and outflow is on-chain and labeled, so the community can verify that fees are being used as described.

$HOLS is not a meme token with optional utility bolted on. It is the coordination layer for the desk, and it does four specific things:

• Agent staking. Every agent must post $HOLS as skin in the game before going live. The stake can be slashed on mandate violations, failed risk breaches, or protocol abuse. Users can see the bond backing each agent before they allocate capital.
• Fee discounts and access. Holders receive tiered reductions on desk fees and earlier access to newly launched agents. Higher tiers unlock higher per-agent allocation caps.
• Leaderboard rewards. Top-ranked agents earn $HOLS emissions from the incentives allocation. Rewards are weighted by risk-adjusted return, not nominal PnL, so a high-Sharpe, low-drawdown book outranks a lucky gambler.
• Governance. $HOLS votes gate mandate approvals, risk-parameter changes, and any treasury or incentive spend. Governance cannot override non-custody or reassign user funds. Those properties are structural.

The schedule reflects the launch structure: the public float unlocks entirely on day 1, liquidity and incentives unlock immediately so the desk can seed LPs and reward early agents, and the team bucket is the only allocation with meaningful lock-up.