Manifund foxManifund
Home
Login
About
People
Categories
Newsletter
HomeAboutPeopleCategoriesLoginCreate
mrinallovesbhature avatarmrinallovesbhature avatar
Mrinal Singh

@mrinallovesbhature

Building seiche.info: free early warning for dollar funding stress, every forecast sealed the day it is made, misses published. Lawyer turned builder.

https://seiche.info
$0total balance
$0charity balance
$0cash balance

$0 in pending offers

About Me

I trained as a lawyer, taught myself to code, and now build instruments that hold themselves to a courtroom standard of evidence.

Seiche is the main one: a free public terminal watching the repo and money markets that broke in September 2019 and March 2020. It publishes a stress read twice a day. Every forecast goes into a hash chained ledger the day it is made, so I cannot quietly fix a bad month later. The misses are listed next to the hits, and when my ML model failed calibration, the site published that finding about itself.

Before this I built LiquiLens, an early warning system for Indian small business lenders. Backtested on 36 institutions, it caught nine of ten collapses with a median lead of fifteen months. Seiche is the same method pointed at the headwaters.

I think anyone selling forecasts should be forced to show their whole record. I am building the existence proof.

Projects

Seiche: a tamper evident early warning system for dollar funding stresspending grant agreement signature

Comments

Seiche: a tamper evident early warning system for dollar funding stress
mrinallovesbhature avatar

Mrinal Singh

about 1 hour ago

If you want to check my claims rather than take my word for it: the board is live at seiche.info, updates twice a day. Scroll to the bottom of today's letter, the section called "The record, nothing hidden." That's the backtest with the misses listed by name, not just the wins. All the data is free and keyless (FRED, NY Fed, OFR, Treasury, CFTC, ECB) so you can rebuild any number yourself if you're feeling suspicious. Honestly the most interesting thing to poke at is the ML calibration failure. The model ranks risk fine but its probabilities were junk, and the terminal says so on the page. Ask me anything, including the awkward stuff. I'd rather lose the grant than dodge a good question.