Investment and Risk Framework for the Cardano Treasury

Governance General Governance Conversation
1

Hi everyone,

I want to propose that we of Cardano create a comprehensive “Investment and Risk Framework” for Cardano’s Treasury - a formal document that defines how the treasury’s capital should be managed, what assets it can hold, how much it should spend and what risk limits apply

The Cardano Treasury currently holds ~1.5bn ADA. At current prices, that is approximately ~$400mn. The entire balance is in ADA - a single volatile asset that has declined over 90% from its all-time high.

The Cardano Treasury currently has:

  • No investment policy
  • No asset allocation strategy
  • No risk limits or risk governance (what to do if price falls below limit)
  • No spending formula
  • No return target
  • No hedging mechanism

The 2026 budget cycle has set a 350mn ADA Net Change Limit (NCL). Which is about to authorize disbursement of over a 1/5th of treasury holdings without a framework for managing the remainder. Cardano does have a governance process for withdrawals from the Treasure that requires DReps approval via on-chain voting and ratification by the Constitutional Committee, but it is silent about how the Treasury should be invested and risk managed.

For context, no institutional investor - pension fund, endowment, sovereign wealth fund - would never hold 100% of its assets in a single volatile asset. A typical university endowment with this balance would hold diversified equities, fixed income, alternatives and cash equivalents with no single position exceeding 20% of the portfolio. Bank treasuries have a formal policy in what assets they can hold and regularly run stress tests on them.

Other blockchain treasuries face the same problem. Cardano is not alone. A few examples include: Polkadot spent ~$87mn in 2024 with no systematic spending policy. Uniswap holds $5bn in UNI tokens with no investment framework. Ethereum foundation only recently adopted a basic guidelines. And no blockchain treasury (to the best of our knowledge) has published a comprehensive policy statement in how the manage their treasury.

The opportunity cost is real. A straightforward 20% allocation into stablecoins and medium risk AMMs would yield a few million of annually, enough to fund grants and in perpetuity.

Consider what happens under stress. If ADA declines by a further 50% - which happened before - the treasury’s purchasing power falls to roughly $200mn. Budget commitments already approved ADA deliver less value (as most companies budget in fiat). There is no reserve buffer in non-ADA assets, no mechanism to prioritize or restructure commitments, and no pre-approved authority to take defensive actions. A framework with a diversified reserve and defined escalation procedures changes this from a crisis into a manageable situation.

What could be included in the Investment Policy Statement (IPS)

  • A governance structure aligned with CIP-1694
  • Investment objectives - 5 to 20% return target, risk tolerance, liquidity requirement linked to the NCL
  • Asset Allocation with tiers: Tier 1) ADA, major stablecoins, tokenized government bonds; Tier 2) BTC, ETH, Cardano DeFi yield from lending and AMMs; Tier 3) Ecosystem strategic investments (e.g. Snek) and venture style allocations
  • Spending policy - adapted from endowment models with smoothing formula to prevent overspending in bull markets
  • Risk management framework - risk taxonomy (what risks are relevant), VaR / Expected Shortfall models adapted to crypto, 5 stress scenarios (crypto winter, stablecoin depeg, governance deadlock regulatory action, protocol failure), risk limits and escalation procedures
  • Implementation guidelines - custody, execution, rebalancing, DeFi eligibility criteria
  • Monitoring and reporting - public dashboard design

The above will go long way for establishing guardrails to avoid the treasury being spent to 0 or drawn on excessively in weak markets.

In addition we should perform on chain analysis to determine how much the prices of ADA has declined as a result of sell pressure from recent selling by projects that received funding from Treasury and Catalyst. This will then be used to inform how much the treasury spend affects the ADA price

A little about myself. I have been with Cardano since the launch of the mainnet, in many different roles: built apps (cardanobeam.app and blockmargin.app); participated in Catalyst; ran block producing nodes (DSIO); written open source libraries (cardano-wallet-connector); built apps for CF (Staking Calculator | cardano.org | Cardano); and more recently serving as a voting member of the Intersect Budget Committee

Previously I spent decades at investment banks risk managing tradings desks and treasury portfolios on how they should be invested and risk managed.

If this discussion gets enough traction then I will submit a proposal in the next budget cycle to work on and develop a comprehensive Investment and Risk Framework for the Cardano Treasury.

I would genuinely value the community’s input on this

P.S. In the age of AI, one can never be sure of what content was AI generated and what was written by a human. So I have separately created an app which proves that content has been written by a human (me in this case). You can check that this text was written by a human by looking at proof of how it was type and additionally signed with a Lace wallet

:writing_hand: Human-written — Just Human

image
image1200×675 25.9 KB

2 Likes
2

Great point, there’s really a lack of a more professional vision in the treasury.

2 Likes
3

Disclosure first: I’m part of the team behind Sentralis, a crypto portfolio risk and scenario analysis tool. Nobody in the Cardano ecosystem asked or paid for this, none of it is advice, and it takes no position on any vote that’s open right now.

This thread asked in March for VaR and expected-shortfall models and stress scenarios for the treasury. As far as I can tell nobody ever posted any, so we ran them: the on-chain pot (1.474B ADA at epoch 642, about $233M at July 13 prices) plus two illustrative diversified variants matching the 5–10% range that’s been floated. A few results, since the votes which are happening this month all draw on this pot:

  • Across two deliberately different Monte Carlo models (25,000 paths each, same seed), the simulated 1-in-20 bad year for the pot’s dollar value lands anywhere between −73% and −94%. Aside from the magnitude (which is to be expected), that 21-point spread is perhaps even more interesting. It’s pure model risk, and it’s the best argument I’ve seen for the risk framework this thread proposed. When the original post was posted in March the same ADA was worth about $400M; it’s $233M now, which is the risk in question having already happened once.

  • We also tried to put a price against the diversification debate: an illustrative 10% BTC+stablecoin mix improves the 1-year 95% expected shortfall by about $14M on the $233M, and softens an FTX-week repeat by about 1.3 points. A 5% mix buys roughly half that. Real money, but a smaller impact than was presented on either side of the discussion

  • The practical cost of the open $100M conversion idea: If the seller stays small enough not to move the price (5–10% of each day’s trading volume), the simulation runs put it at roughly two months in a normal market and close to a year in a crisis, with $27–41M of modeled price impact along the way. One caveat matters a lot here: the day-count depends on what is accept as ADA’s real daily volume. Our data counts a filtered set of venues and shows about $24M a day; CoinGecko counts every listed venue and shows roughly ten times that. Take the bigger number and the whole job takes a tenth as long. The upcoming write-up swill show both.

As with any risk analysis, all of it is model output under some rational assumptions, not fact or prediction. The full write-up with methods and limitations goes up in the Governance category this week. If anybody finds some errors in the data basis, please do tell and I’ll be happy to fix it. And if you or anyone at Intersect wants this re-run against the exact tier structure proposed here, with your assumptions instead of ours, I’m happy to do that as well.