I spent about 6 months on creating an implementation of @ISDA's SIMM for DeFi. I hope this represents a step forwards in providing an open-source guide to removing the risks around arbitrary leverage in DeFi. As a lender, you can use an AI model to audit collateral LTV limits for safety. This gives you minimum safety bounds, derived from cross-industry quants and experts to rely on. For risk managers, structurers and lending protocols this gives you a benchmark on how to set parameters with a full methodology. Now caveats, I'm really not happy with putting this out there today as I feel like it's not finished and I keep finding errors here and there. But it's at the point, where my errors seem to just be grammatical or bad phrasing, or some random slop that snuck in. Whilst, I remain unable to say this is a final version I think it's well past the 80/20 principle and needs to just get out there. Contributors ... 1) As mentioned, this isn't finished (arg). There are a number of sloppy parts in the back components in particular. 2) We need to extend this to a new crypto risk class - btc, eth, sol, avax etc. - bridges (wBTC, BTC.b etc) - wrappers / staking (wstETH, weETH, wETH etc.) 3) I'll be working on some edits over the next few weeks into FX, Equities, Comm etc as they're a LOT easier than tokenised funds. Link in comments ...

Call for contributors for a paper on DeFi liquidations... Liquidator roles have changed drastically since the addition of tokenised assets as DeFi collateral. Originally coordinators of infrastructure, they are now expected to provide balance sheet to absorb market and credit risk. The problem is that DeFi protocols have been built assuming assets can be sold instantly to a willing buyer. Whilst risk managers have taken this into account somewhat with LTVs, the protocol design has neglected the liquidator's need for a haircut to warehouse these risky assets. Our calls with leading liquidators this week confirmed what we suspected: most are unaware that protocols have hard constraints on the discounts that can be applied to collateral. This means that a slightly unhealthy position in a tokenised fund is unlikely to find any bidders, and will instead sit idle in the market. It is also possible that under certain conditions an asset can never be liquidated, as the maximum discount available is simply not enough for any rational actor. Different lending protocols have different constraints, but we found that in general none were well suited to clearing tokenised assets and funds in their default configuration. Some DeFi lending protocol teams, as well as some risk managers, have been aware of this issue and have been working on fixes. In talking to industry practitioners, one recurring theme was that levered credit positions rarely (if at all) got liquidated. The liquidation business for tokenised assets was therefore a low priority for most would-be liquidators, despite a surge in interest earlier this year. What this means is that for these assets, liquidators are relied upon to provide a service that is objectively negative expected value for them. We expect that some of the largest market operators, Aave for example, can justify this position as a service. But it gives reason to believe that the majority cannot, or would not, in a big market sell-off. The incentives are skewed against them, as being a liquidator in such conditions is effectively being a charity. We consider any third-party agreements (SLAs) with liquidators under such conditions to be largely spurious unless there are significant protocol changes: they are effectively deep out-of-the-money uncleared put options. This concerns us as retail exposure to DeFi increases through exchanges, PSPs and neobanks. It represents a critical fragility vector. Link to the GitHub repo below. It's still a draft, but we've spent about a week tidying it up so that it's ready for outside reviewers and contributions.