The decomposition
layer for equities.
A stock is a bundle of separate economic drivers, but the market only lets you trade the bundle. Tranch splits a company's reported fundamentals into individually tradable claims for revenue, EPS, margin and segment revenue, each settling against the figure the company itself publishes.
- Live markets
- 9
- Collateral locked
- $12.77M
- Tickers
- 5
- Metrics tracked
- 9
NVIDIA Corporation
Q3 FY2027
price
Buying the stock buys every driver at once, whether you have a view on them or not.
Stocks are bundles. Tranch lets you trade the pieces.
Each market isolates one figure from one earnings event. Depositing collateral mints a matched pair of claims on that figure: one that gains if it beats the reference value, one that gains if it misses.
Continuous payoff, not a coin flip
Settlement scales with the magnitude of the surprise rather than a yes/no threshold. A 1% beat and a 12% beat pay very differently, and each metric gets its own sensitivity. Margin markets run 12x leverage where revenue runs 2.5x, because a 1% margin move is a far bigger event.
Fully collateralised by construction
The two sides of a market always divide exactly one unit of collateral, so total claims can never exceed deposits. Payouts are clamped at both ends, which means no reported figure, however extreme, can make the market insolvent.
Objectively settleable
Every live market settles against a number the company states explicitly in its own quarterly disclosure. No judgement, no modelling, no valuation opinion. Settlement is a matter of reading a line item.
Collateral in, two claims out.
Split and merge are first-class protocol actions, not an implementation detail. They are how exposure is created and destroyed, and they are what guarantees a complete set is always redeemable for exactly what was deposited.
A complete set is worth exactly one unit of collateral under every possible outcome, so it can always be merged back before settlement. After settlement the two claims divide that unit according to the reported figure. One side gains exactly what the other gives up, which is why total claims can never exceed deposited collateral.
NVDA Q3 FY2027, decomposed
Four independent markets on one earnings event. Each has its own reference value, its own settlement definition and its own payoff sensitivity.
Q3 FY2027 · Revenue
Q3 FY2027 · EPS
Q3 FY2027 · Gross Margin
Q3 FY2027 · Data Center Revenue
“NVIDIA revenue will crush expectations, but margins will disappoint.”
Buying or shorting the stock nets these two views against each other. On Tranch they are two separate instruments, so the position is simply both legs at once.
Pays more the further above the reference value revenue is reported, up to a capped maximum.
Pays more the further below the reference value gross margin is reported, up to a capped maximum.
Two spot positions. No options chain, no calendar spread, no view on implied volatility.
Build this positionEvery reported driver becomes its own market.
Tranch covers quantitative earnings metrics because they settle deterministically from public disclosure. The same architecture extends to any ticker and any reported figure, and eventually to a genuine on-chain sum-of-the-parts market.
- RevenueLive
- EPSLive
- Gross MarginLive
- Data CenterLive
- GuidancePlanned
- AutomotiveLive
- EnergyLive
- RobotaxiPlanned
- OptimusPlanned
- AWSLive
- RetailPlanned
- AdvertisingPlanned
- CloudLive
- SearchPlanned
- YouTubePlanned
- WaymoPlanned
- LaunchLive
- Space SystemsPlanned
- ContractsPlanned
A weighted earnings score, decomposable into its parts
The contract interfaces already anticipate a composite market whose score is a weighted blend of individual metric scores (revenue, EPS, margin, guidance) that can itself be decomposed into single-metric tranches. Nothing in the protocol today depends on it, and no relationship between unrelated metrics is faked in the meantime.
Split the stock.
Trade the fundamentals.
Pick a company metric, choose your side, and hold a claim that settles against the number the company reports.