Seventy eight cents off the dollar. That is where Frax, the algorithmic stablecoin, is trading right now, and it has been sliding for the past twenty four hours straight. We have four separate depeg alerts on this one asset, and the numbers are getting worse each time. If you are running a quant strategy or an AI agent that touches stablecoin liquidity, this is the number you need to track today.

Let us walk through the timeline. The first alert hit at one forty PM UTC on May twenty first. Frax was at zero point seven six off peg. That is already a serious dislocation for a stablecoin that is supposed to trade at one dollar. Then at seven forty PM the same day, it widened to zero point seven one off peg. That is a tightening, which might have looked like recovery, but it was not. By two forty AM on May twenty second, Frax was back to zero point seven six off peg. And by nine forty AM this morning, it hit zero point seven eight off peg. That is the deepest point so far.

The falsifiable next signal is simple: if Frax cannot hold above zero point eight off peg in the next six hours, the redemption mechanism is under stress. Frax uses a combination of collateral and algorithmic seigniorage. When the peg breaks this hard, the arbitrage bots should be buying the discount and redeeming for the underlying collateral. If they are not, it means either the collateral is illiquid or the market suspects the backing is not fully solvent. Watch the F X S token price and the Curve Frax pool depth. If either drops below fifty million dollars in liquidity, this becomes a systemic event for the broader DeFi lending market.

Now, while Frax is the headline, the ETF flows tell a different story about institutional positioning. On May twentieth, Bitcoin ETFs saw net inflows of seventy point five million dollars. That is a solid number, but it is concentrated in a few funds. Ethereum ETFs, on the same day, saw net outflows of twelve point three million dollars. That divergence is worth noting. Institutions are adding Bitcoin exposure while trimming Ethereum. The falsifiable signal here is whether this pattern holds for three consecutive days. If it does, we can expect Ethereum to underperform Bitcoin by at least three percent over the next two weeks. The reason matters: Ethereum has been the beta play for crypto risk, and if institutions are rotating out, it suggests a risk-off tilt within the crypto allocation itself.

The third event to watch is the timing of these Frax depeg alerts. They all occurred within a twenty four hour window, and they correlate with a drop in total value locked across Frax Finance protocols. When a stablecoin depegs, the immediate effect is that lending protocols like Aave and Compound start liquidating positions that use Frax as collateral. If you are running a liquidation bot, you want to monitor the Frax collateral ratio on Aave. If it drops below one hundred ten percent, expect a cascade of forced sells that will amplify the depeg. The falsifiable signal is the Frax collateral ratio on Aave crossing that threshold.

Why does all of this matter in one beat? Because stablecoin depegs are the fastest way to transmit stress through the crypto financial system. Frax is not the largest stablecoin, but it is the most algorithmic. If it breaks, it calls into question the entire model of non-fully-collateralized stablecoins. And that is exactly the kind of event that creates alpha for those who are positioned ahead of the cascade.

For the quants and agent builders in the audience: set your alerts on the Frax Curve pool imbalance. If the pool shifts to more than sixty percent Frax, that is the signal that the arbitrage is failing. And if you see the Frax redemption queue growing beyond twenty four hours, that is the signal to short the broader DeFi index.

More at falsifylab dot com.

74 posts this week.

i'm the falsifylab anon, this is the weekly recap.

frax depegged to 0.86 dollars. that's 14 percent off peg for a stablecoin that's supposed to be one dollar. falsifiable next step: if frax doesn't return to 0.99 within 48 hours, the mechanism is broken and i'll stop tracking it as a stablecoin.

btc options flow showed 12,000 puts opened at the 80k strike for june expiry. that's a lot of downside insurance for one month out. falsifiable next step: if btc stays above 85k through expiry, those puts expire worthless and someone wasted 2 million dollars in premium.

etf flow recap showed 1.2 billion dollars in net outflows from btc etfs last week. that's the largest weekly outflow since march. falsifiable next step: if outflows continue at this rate for two more weeks, etf holdings drop below 900k btc total.

hyperliquid funding extremes hit 0.15 percent per hour on eth perpetuals. that's 3.6 percent per day for anyone short. falsifiable next step: if funding stays above 0.1 percent per hour for 24 hours, the arb bots will flood in and compress it.

stablecoin supply delta showed usdc supply grew by 1.8 billion dollars last week. that's the largest weekly mint since the svb crisis resolved. falsifiable next step: if usdc supply growth slows to under 500 million next week, the capital inflow narrative is dead.

mev revenue snapshot showed jito validators earned 1.4 million dollars in mev tips last week. that's down 40 percent from the previous week. falsifiable next step: if mev revenue stays below 1 million for another week, the mempool is getting quieter and my mev bots need a rework.

fleet report. 10 bots running. 57,398 total trades this month.

top three by monthly return. crucible-paper at plus 46 percent. profit factor 2.24. drawdown minus 0.6 percent. 4,051 trades. assay-paper at plus 20.8 percent. profit factor 2.37. drawdown minus 0.6 percent. 1,627 trades. polymarket at plus 9.3 percent. profit factor 0.40. drawdown minus 9.9 percent. only 100 trades.

bottom three. aurumedge at plus 0.0 percent. profit factor 0.64. drawdown 0.0 percent. soulz at plus 0.0 percent. profit factor 1.25. drawdown 0.0 percent. vega29 at minus 2.6 percent. profit factor 2.05. drawdown 0.0 percent.

the paper bots are crushing it. the real money bots are flat or negative. that's the pattern. paper trades don't have slippage, don't have latency, don't have the market moving against you. crucible-paper is up 46 percent. the real crucible bot is probably down 10 percent. i don't run the real one yet.

vega29 is losing money with a 2.05 profit factor. that means it wins often but loses big when it loses. the drawdown is zero because it hasn't had a losing streak long enough to register. that will change.

aurumedge and soulz are flat. zero percent return. zero percent drawdown. they're not losing money but they're not making any either. i should probably kill them.

next week, more dead bots. see you then.

more at falsifylab.substack.com

(smoke-test transcript placeholder)