Frequently Asked Questions
- If SPICE's appraisal engine is turned off, SPICE will see a degradation in performance that is proportional to downtime. As a heuristic, downtime shorter than 10 days has little to no impact on appraisal error rates, since collection appraisal models get retrained roughly every 10 days. After 10 days, the degradation is linear.
- If price feeds go down, SPICE stops placing bids. There are no loss of funds.
- If bid placement systems go down, SPICE stops placing bids. There are no loss of funds.
- If rebalancing instructions go down, SPICE stops rebalancing aggregator vaults. Underlying funds continue to earn yield in their respective deposited area.
- Users are still able to fully withdraw funds as any liquid funds in the marketplaces SPICE routes to or the bidding vaults SPICE uses are available for withdrawal at any time.
- SpiceDAO hosts two main modules used for bid placement off-chain: a module that calculates a bid for a given asset or collection and a given duration, and a module that places bids.
- After a bid (and LTV) is calculated, the bid is passed from the module that calculates bids to the module that places bids through a simple function call. Thus, the LTV is enforced through code hosted off-chain.
- The module that places bids is essentially an SDK for the underlying peer to peer marketplace.
- SPICE protocol's off-chain modules currently represent a centralization risk for the protocol. As NFT finance marketplaces, oracles, computational layers and underlying crypto technologies evolve, the SpiceDAO team will look to decentralize accordingly.
- At the current stage of the NFT finance industry, lending marketplace order-books are largely hosted off-chain. Pricing feeds and ML appraisal by industry participants are also computed off-chain, requiring portions of the SPICE protocol to maintain off-chain components.
- SPICE's lending book currently composes a <5% market penetration rate among profitable lenders. Assume no NFT finance growth, SpiceDAO anticipates a 5-10x in TVL (~5,000 ETH) before returns begin compressing significantly.
SPICE calculates bid parameters based on both historical floor price for collection and individual bidding bidding and historical appraisal price for individual bidding. The term is specific to a given collection or asset and a duration, and is calculated following a series of steps, listed below:
- First, SPICE uses volatility estimation algorithms for a given collection or asset. To do so, SPICE uses historical pricing data as the input to a machine learning model that determines the optimal historical dataset to minimize future volatility prediction error.
- Next, SPICE runs a proprietary Black & Scholes Options Pricing Model with adjustments for total collection leverage, low volume and high volatility. Appropriate adjustment are applied to account for liquidation timelines, floor-ask spreads, market-derived Implied Volatility to calculate an optimal risk-adjusted bid.
- LTV limitations are applied as a function of (1) maximum historical drawdown of prices over the duration of the loan and (2) conditional value at risk from Black and Scholes Options Pricing Model.
- A maximum duration limit of 30 days is applied to all bidding terms.
- Bidding limits are currently set at a 20% weight to maximum historical drawdown and an 80% weight to a cVAR of 4.0%.
It’s largely economically infeasible to wash farm a collection SPICE appraises into a high appraisal value, since it would require paying royalty fees or marketplace fees across thousands of trades for an uncertain and unmeasurable benefit as SPICE's appraisal algorithms are not made public. That said:
- SPICE only uses appraisal when lending to individual assets, not collection bids. Thus, the potential risk is siloed into specific assets within a collection, rather than the whole collection, which is much more dangerous.
- Collection additions are periodically revisited by the SpiceDAO. Collections with unusual and spoofed volume are not added to SPICE Vaults.
- Low liquidity or high levered collections are heavily penalized with SPICE's conditional value at risk approach. For example, collections that quickly move 20% to the upside results in a significantly lower acceptable LTV than prior to the move upwards.
- Market manipulation remains the highest risk for NFT market participants. SpiceDAO's lending strategy is largely designed around identifying market manipulation and algorithmically adjusting lending terms accordingly.
- SPICE does not custody NFTs, so SPICE vaults do not receive airdrops when an NFT is in escrow.
- cVaR and maximum historical drawdowns limitation prevent SPICE from lending at LTVs highly susceptible to losses incurred from airdrops. Bid-ask spreads set by NFT market makers and Implied Volatility modeling aim to capture potential event driven price movements.