Solutions
Lending protocols
Your liquidation engine marks collateral at oracle price and unwinds it at book price. The difference is an unmeasured short position in liquidity, carried on every loan.
What you get
- Per-collateral exit gap at real liquidation size
- Worst observed depth, not the 30-day average
- Alerts before a cascade, not a post-mortem after one
The panel beside this shows the shape of the answer, not live data. Three things are in production today — depth recording, keyless verification and the published method. Everything else is roadmap and is labeled as such.
Other solutions
- Autonomous trading agentsAgents trade continuously, at machine speed, with no instinct for when a market has gone thin. Speed without depth awareness is not an edge — it is a liability that compounds every block.
- Tokenized equities and RWAThose markets close. The tokens never do. An agent trading a tokenized stock on Saturday is quoting Friday's price into an empty book, and nothing in its stack tells it so.
- Treasuries and DAOsA treasury dashboard multiplies balance by oracle price and calls it a valuation. It is a valuation at zero size. The number that matters is what the treasury clears at when it actually needs the cash.
- Quant desks and market makersYou already model execution internally. Crifine is the outside check — an independently recorded depth series, with the assumed size and observation window stated, that your model can be scored against.
- Risk platformsParameter reviews answer whether a protocol's settings are sound. They run quarterly, per protocol, for humans. Crifine answers the same question per order, per moment, for machines.