Estimation of Metrics & Parameters for Term Vault and Summer.fi
Anticipating the deployment of a new higher risk Ark on the Summer.fi protocol that will allow the fleet to provide liquidity to a Term Finance vault, we conducted a statistical analysis to determine the optimal parameters for the Term vault. Using the expected deposit volumes and collateral composition, we will feed this data into our risk model on the Summer.fi side to compute parameters for the specific Ark. Based on discussions in the forum (rfc post, sip post, and sip_post), the final setup will include a new Ark added to the Higher Risk fleets, while the Term vault will use the Lower Risk fleet on Summer.fi as its idle market for the respective asset.
Data Acquisition
Using Term’s API via subgraph and RPC calls to on-chain data, we collected data for 13 strategy vaults currently live on the Term protocol. This includes collateral composition (including utilized collateral), current vault TVL, APY, and other relevant metadata such as concentration limits, required reserve ratios, and current liquid assets.
Once the hypothetical values for the Term vault are defined, we will integrate them into our risk model to determine the initial parameters for the corresponding Ark on the Summer.fi side.
Methodology
We analyzed all strategy vaults that use the same assets intended for deployment as Summer.fi branded vaults: USDC and WETH. Since Summer.fi aims to optimize yield while keeping assets safe, there is a tradeoff between attracting borrowers and enforcing a high enough collateral ratio to protect the underlying funds. For the upcoming Term vaults, the proposed collateral assets are: for the USDC Term vault, cbBTC, wstETH, weETH, sUSDe, LBTC, and PT-sUSDe 31JULY2025; and for the ETH Term vault, cbBTC, wstETH, weETH, rsETH, and ezETH.
We analyzed existing Term strategy vaults using each of these collaterals where the deposit asset matches ours. For instance, Figures 1 and 2 show the frequency of strategy vaults using cbBTC and sUSDe as collateral, along with their respective minimum collateral ratios, for USDC deposits. To remain competitive while prioritizing safety through higher collateral ratios, we will define our vault parameters based on the following formula:
where
and 
are the frequency and the total backed debt from all vaults at the respective minimum collateral ratio i. This way we trade between having potentially good demand while putting more weight on the highest values.
Fig. 1. Number of vaults using the same minimum collateral ratio (left), and the same histogram with debt-weighted values. This data is taken for cbBTC in USDC Term vaults.
Fig. 2. Number of vaults using the same minimum collateral ratio (left), and the same histogram with debt-weighted values. This data is taken for sUSDe in USDC Term vaults.
For cases where there is no debt-collateral assigned to a given collateral asset in the proposed Summer.fi Term vault, we select the highest observed minimum collateral ratio from other vaults. In the case of the ETH Term vault, none of the proposed collaterals are currently used in existing vaults, so we assign the highest and only available values from other strategies. It is worth noting that most ETH-related vaults are heavily concentrated in ETH+.
To compute the expected TVL, idle liquidity, and required reserve fraction, we apply a similar logic. We analyze the required reserve ratios and their frequency across all vaults with the same deposit asset. This allows us to define realistic input assumptions for our model.
where 
and 
are the number of vaults and the assets in idle for each specific.
. Finally, for the collateral ratios, TVL, and expected APY, we make a simple weighted average of each value among all vaults (also per collateral for the case of the collateral ratios). By doing so, we arrive with the following values for the vaults:
| Parameter | USDC Term Vault ($) | ETH Term Vault (ETH) |
|---|---|---|
| Exp. TVL | 680,000 | 660 |
| Exp. APY | 3.7% | 2.8% |
| Exp. Liquidity Reserve | 20% | 50% |
| Exp. Min Required Reserve | 20% | 10% (we suggest 20%) |
And for the collaterals, for the USDC vault:
| Collateral | Exp. Distribution | Min. Maint. Ratio |
|---|---|---|
| cbBTC | 13.4% | 1.25 |
| wstETH | 0.0% | 1.16 |
| weETH | 0.0% | 1.25 |
| sUSDe | 28.1% | 1.1 |
| LBTC | 0.0% | 1.5 |
| PT-sUSDe 31JULY2025 | 38.5% | 1.08 |
And for the ETH vault, since all the existing vaults have most assigned distribution to ETH+, we have divided the remaining 50% evenly among the ETH-related collaterals:
| Collateral | Exp. Distribution | Min. Maint. Ratio |
|---|---|---|
| cbBTC | 0.0% | 1.16 |
| wstETH | 12.5% | 1.05 |
| weETH | 12.5% | 1.05 |
| rsETH | 12.5% | 1.16 |
| ezETH | 12.5% | 1.16 |
Both vaults will have a Concentration Cap of 60%, and a 45 days of Weighted Average Life Cap (WAL) Cap as suggested by Term Finance.
SummerFi Higher Risk Fleet Parameters
Taking into account the market data from the previous sections, we have used our risk model to compute initial parameters for the new Ark on the Higher Risk Fleets on Mainnet for both assets, USDC and WETH.
Higher Risk Fleet Mainnet - ETH
The current fleet caps are the following:
| chain | fleetAsset | fleetCap | FleetMinimumBuffer |
|---|---|---|---|
| Mainnet | ETH | 15,000 | 0.37 |
With the new ark:
| Ark | Symbol/Vault | maxCap | Max. %TVL | maxInflow | maxOutflow |
|---|---|---|---|---|---|
| Term | WETH | 1,280 | 53.0% | 1,280 | 1,280 |
Higher Risk Fleet Mainnet - USDC
The current fleet caps are the following:
| chain | fleetAsset | fleetCap | FleetMinimumBuffer |
|---|---|---|---|
| Mainnet | USDC | 20,000,000 | 1,000 |
With the new ark:
| Ark | Symbol/Vault | maxCap | Max. %TVL | maxInflow | maxOutflow |
|---|---|---|---|---|---|
| Term | USDC | 1,400,000 | 54.0% | 1,400,000 | 1,400,000 |
The parameters above represent the final caps after three weeks of progressively increasing them. The reason for these three weeks is due to new implementation at the smart contract level from the summer.fi side, and will serve as a slow roll up of the full caps after verification that the ark is working as intended. Thus, the maxCap and Max %TVL will be planned to be executed as (it may differ from actual execution if any issue arises):
The current cooldown period for all ARKs is set to 10 minutes. These parameters ensure the fleets operate within controlled risk levels while supporting expected demand. However, we should keep in mind that there is a lockdown period of max 45 days for supplied assets, and in the case of full utilization, only 20% of the Ark TVL will be available for immediate withdrawal at all times. Furthermore, once the Term vault is live, we will start collecting actual data and eventually revise the ark parameters.