ForetellerForeteller

Methodology

Everything on Foreteller is computed from public Solana data. No self-reported numbers, no API from the venue, no manual curation. This page explains exactly how, so anyone can verify any number against the chain.

01

Data source

We index every transaction touching World's on-chain programs, in real time (typically within seconds) plus a rolling 7-day history:

prediCt (market core)prediCtPh6tM

Mints 1 YES + 1 NO per 1 CASH of collateral; merges and redeems. The trustless escrow.

Janus (AMM)JanusXpmEYFP

The always-on counterparty. Prices each side and warehouses opposite-side inventory.

DFlow (router)DF1ow4ts7QBH

Order routing and SOL to CASH conversion.

CASH (settlement)CASHx9KJCASH

Token-2022 stablecoin, 6 decimals. All values on this site are real CASH dollars.

02

What counts as a trade

A fill is attributed to a wallet only if that wallet signed the transaction (for buys and sells) or is a known trader receiving a settlement (redemptions are keeper-executed on World, so winners don't sign their own payout).

This matters because the Janus AMM is the counterparty to every trade: for each user fill there is a mirrored pool fill. Counting both, as a naive indexer would, roughly doubles volume and puts market-maker inventory on the leaderboard. We filter it out; when we measured, pool mirror-flow was 57% of raw recorded volume.

03

How trades are valued

  • CASH-paid trades: valued at the CASH the wallet actually moved.
  • SOL-paid trades: the wallet's own CASH barely moves (DFlow converts internally), so we derive the real odds from the opposite side of the same fill: the pool's payment for the other outcome reveals the price, and value = shares × price. Prices always satisfy YES + NO = 1.
  • Redemptions: valued strictly at the CASH received. Winning shares settle at $1; losing shares burn for $0. We never guess a settlement price.
04

The Foreteller score

The score (0 to 100) measures risk-adjusted realized edge, not win rate. For each trader we take every market they traded that has resolved, compute the return on cost for each, and score the distribution: consistent profit across many settled markets scores far higher than one lucky hit. Small samples are shrunk toward the mean, and traders need 5+ resolved markets to be ranked at all.

PnL and ROI shown are realized, from settled markets only. Open positions never inflate a score.

05

Window, freshness, and honesty

  • Rankings cover a rolling 7-day window; all-time counters accumulate separately.
  • New trades index within seconds; the "indexed Xs ago" chip on the leaderboard is live.
  • If history is resyncing, a banner says so with progress, and totals are labeled as still climbing.
  • Volume shown is real trader volume in CASH dollars. Venue front-ends may display raw base units (6 decimals), which reads a million times larger.
06

Verify it yourself

Every trade row links to its transaction on Solscan. Every market links to its outcome mints. The API (/docs) returns the same data the site renders. If you find a number that doesn't match the chain, tell us: that's a bug, and we fix those loudly.