Fixed yield on Pendle, with the parts nobody prints
Vantias reads every live Pendle market on eight chains, keeps its own dated history of them, and says what a fixed-yield buyer would actually receive at a size they choose -- after the cost of getting in, after the cost of getting out, and with the risk that is being taken for it named rather than scored out of ten.
It also publishes what it has NOT checked. 2 of its 31 published figures have never been graded, and the service names which ones. That sentence does not appear on any competing screen, because nobody volunteers it.
How a number reaches the screen
Six steps, in the order they run. Two of them end in a claim rather than a reading, and the service says which two.
- 1What comes in
Three sources, and the browser is none of them: Pendle's index for the market list and its published rates, each chain's own contracts for the wrapper and who may change it, and Pendle's router for a quote at a real size. Every reading is stored with the moment it was taken.
- 2The fixed side
A PT is a bond. The rate is the discount over the days remaining, recomputed here rather than copied -- and a market whose two published figures disagree is rejected rather than shown with a number nobody can defend.
- 3The floating side -- the guess the ranking rests on
The only guess. Today's reported rate on a leash to the 30-day median, chosen on a backtest of the stored history. Where an asset reports a rate the market will not pay for, every figure built on it is withheld and says so.
- 4What it costs to get in, and out
Every yield is net at a size you choose. Entry is a router quote at that size; the exit is a second quote at the size you would actually hold, and a market the router refuses costs the position points rather than being scored as though leaving were free.
- 5The risk band
Six layers, each scored on its own evidence. The band is the worst of them, never the average, and the card names which layer set it. There is deliberately no single score out of a hundred.
- 6What is kept, and what is rejected
A refused market is not deleted -- it stays in the list carrying the rule that refused it, so it can be sorted beside the rest and argued with.
What it says that others do not
Three habits, each of which costs the service something and is kept anyway.
- It grades its own estimates and publishes the marks. The floating-rate estimator has been scored over fifty thousand days against what followed. The record is allowed to say the estimator in use lost -- and on the average day it did.
- It names the figures nobody has checked. Every published number is classified: somebody else's reading copied, arithmetic with no free parameter, or a guess. Only the guesses can be wrong on their own, and each one carries its own state -- scored, pending, or never.
- It refuses questions it cannot answer. The risk bands have never been checked against outcomes and the service says so on its own page. A backtest that would have been a tautology is published as a refusal with the counts that prove it.
Is it used with real money
The fairest question a stranger can ask. The answer is in bands rather than figures -- the scale is what a judgement should turn on, and a public page printing a book to the dollar would be publishing a wallet's size to anyone who asked.
Computed from the same book the service's own portfolio page reads, and updated with it. Nothing here is typed in.