Author: @MrMadila
Reviewer: @allan.g @funkmasterflex
Abstract
As part of the product design process, the product pod has built a modest framework in an attempt to holistically model a product’s risk profile.
Rationale
All products going through the design process are evaluated and must make logical sense to end users from a risk-reward perspective. The framework acts as an early barometer to sanity check potential ideas before moving forward in the research and design process; it may also be useful for end user communications and comparative analysis.
Specification
There are many different ways to measure and judge risk. The Product Indicative Risk Framework aims to strike a balance between objective, quantitative data points and more subjective, qualitative observations, culminating in a high-level n/5 risk score for each product (~1 being low risk and ~5 being high risk).
Risks have been broken down further into the following categories and sub-categories:
Intrinsic
Asset - the quality of an asset (ex. the ratio of collateral backing a stablecoin)
Protocol - the audited and reviewed safety and security of a protocol
Strategy - the relative risk of any particular strategy (ex. depositing assets to a collateralised borrow and lending platform like Aave is less risky than LP’ing them)
Extrinsic
Economic - the backtested and expected performance of the product, including; volatility, correlations and beta to relevant benchmarks or reference assets
VaR (Value at Risk) - if an asset or protocol were to completely fail how much loss would the product be exposed to
Contagion - how vulnerable is the product to shocks and counterparty risk for example in the event of acute market stress
Co-operative - are the assets, protocols or strategies dependent on 3rd party entities to perform and function or vulnerable to change or interruption from governance etc?
Asset - Risk modeling in DeFi is most commonly seen across borrow/lending platforms where liquidations of collateral play an essential role in the protocol’s function. Due to their importance, these protocols often have dedicated in-house or outsourced teams completing highly detailed, dynamic and in-depth modeling and reporting.
While Index Coop products and design processes do not require the same level of consideration, many of the modeling outputs by other protocols are public and can serve as useful proxies. For example assets that are listed as collateral across borrowing/lending platforms are considered lower risk than similar assets that are not. We therefore take the LTV ratios (%) across all borrowing/lending protocols and synthesize them into an n/5 score using the formula =(1-Average LTV ratio (%) )*5
So an asset with an average LTV of 80% would = 1, 50% = 2.5 and 20% = 1 and so on.
Exceptions:
- Where an asset is listed on more than 2 protocols the average is taken. If it is only listed on one protocol then the score is halved. (ex. TUSD is only on Aave v2)
- If an asset is only listed on a sidechain/L2 but not mainnet then again the score is halved. (ex. USDT on Aave v3 is only available on Arbitrum)
Protocol - Index Coop uses Defi Safety’s public protocol reports for quantitative measures of a protocol’s safety and security. Smart contracts and their general availability for verification, documentation such as whitepapers, internal testing practices, audits and bug bounty programs, admin controls over key/core aspects of the protocol, quality of oracles as and when they are used.
For a qualitative assessment the following broad assumptions result in the following scoring. Note: scores can never total to absolutely 0.
| Qualitative Scoring | Notes and examples. |
|---|---|
| Tier 1 = +0.5 | Blue chip / Majority sector TVL. (ex. Uniswap) |
| Tier 2 = +1 | Well known / Minority sector TVL (ex. Balancer) |
| Tier 3 = +2 | Pre-launch / Beta / Very low relative TVL (ex. Arrakis) |
| Excellent reputation = +0 | Spotless track record / defi ambassador (ex. Aave) |
| Excellent reputation = +0.5 | Well regarded within general defi circles (ex. Euler) |
| Good reputation = +1 | Well-known, no major PR or defi incidents (ex. Aura) |
| Bad reputation = +2.0 | Signs of unprofessionalism / controversy (proven or otherwise) (ex. CREAM) |
| Terrible reputation = +3.0 | Widely regarded as scam / ponzi etc by the defi community (ex. Terra) |
Strategies - Strategy risks are aimed to be relative to each other and take into account potential losses, volatility, reflexivity/convexity, liquidity, duration, rate sensitivity and FX risks for example.
Economic - These risks consider how the performance of a product as a whole is expected to behave. Again factors such as volatility (both in $ terms, and to a relevant asset / benchmark) are applied but also factors such as VaR (Value at Risk). For example, in a stable coin product equally composed of 3 stable coins, if one were to fail in isolation the max loss would be capped at 33.3%. The same is also applied to protocols. This is an important metric to capture the effects of diversification and subsequent risk mitigation.
Market Stress - Market Stress relates to effects of market conditions during periods of acute stress on any given product. Factors include correlations / beta to risk assets and the risks of contagion should a catastrophic event occur within the ecosystem.
Co-operative Risks - 3rd party dependance and/or governance that can have a material impact on the performance or functioning of products are taken into consideration.
Scoring, ranking and weighting
As previously mentioned, overall scores are n/5. Quantitative metrics are weighted 70/30 over qualitative. Where qualitative measures are not available, applicable or appropriate the quantitative score is carried over and vice versa for missing quantitative scores. Another important distinction when defining risk is a product’s absolute risk vs relative risk. For example: an ETH denominated product such as dsETH is expected to be as highly volatile in $ terms as ETH. However in ETH terms it will naturally track very closely. Therefore dsETH is medium risk in $ terms but med-low risk in ETH terms. (Note: this is in the context of all Index Coop products, leverage products and composites are expected to have med-high to high absolute risk ratings due to being more volatile or containing low market cap tokens etc.)
The intrinsic and extrinsic sub-categories are also weighted in an attempt to best reflect their impact on the performance of the product. These weightings are the same across all products and will only ever be adjusted in exceptional circumstances.
Finally, each numerical score is given a simple label:
- 0-1 = Low
- 1-2 = Med-Low
- 2-3 = Medium
- 3-4 = Med-High
- 4-5 = High
Example (dsETH)
| Risk Weightings | Absolute Scores | Relative Scores | Weightings |
|---|---|---|---|
| Asset | 1.40 | 1.40 | 10.00% |
| Protocol | 2.51 | 2.51 | 10.00% |
| Strategies | 1.25 | 1.25 | 30.00% |
| Economic | 3.81 | 0.61 | 40.00% |
| Market Stress | 0.70 | 0.70 | 5.00% |
| Co-op | 2.50 | 2.50 | 5.00% |
Result (dsETH)
| Absolute | Relative |
|---|---|
| 2.45 | 1.17 |
| Medium | Med-Low |
Summary
The combination of all of the above results in a simple, indicative risk score available to view in this public spreadsheet.
Due to the complexity and subjectivity of measuring, ranking, weighting and scoring risk, the model is deliberately designed not to be rigid or overly scientific. The results are purely for informational purposes only and do not under any circumstances constitute financial advice.
Product scores will be periodically reviewed as well as updated when material changes to protocols/assets become apparent.
Users requiring further information or with questions and/or feedback are welcome to reach out to us in our community discord server.