How tickr works

tickr is a trading-first chat platform built on top of Deriv's public WebSocket API. Every market you can see here is a real Deriv instrument — volatility indices, Boom / Crash synthetics, forex, crypto, metals, stock indices — and every number on screen is updated tick-by-tick from the exchange.

Around that live feed, we've built three distinct ways to engage: a conversational AI analyst that can draw charts and place trades for you, and two gamified prediction products that turn the same tick stream into something you can actually play with. Everything non-trading is demo-only; nothing touches real money unless you explicitly place a real contract in chat.

The chat is a Next.js streaming interface wired into Claude. Ask it a question in plain English — “chart EUR/USD with SMA” — and the model decides which tool to call. Tool results render as inline widgets (candlestick charts, trade tickets, signal dashboards), not plain text.

A 2-D grid is overlaid on a live price chart. Columns are time (5 s each), rows are price bands (sized by the instrument's volatility). Tap a cell to bet that the price will visit that band during that time window. If the live price touches the cell at any tick, you win stake × multiplier.

The pricing model

Each cell's multiplier is derived from the touch probability of a driftless log-GBM — the odds that the price passes through a narrow price band during [t_start, t_end]. We use the reflection-principle first-passage formula:

Where B is the nearest band edge, S₀ the current spot, σ the annualized volatility, and T the window length. The displayed multiplier is 1/P, capped by a tier (so sub-percent bets can't pay unbounded), and the platform takes a 5 % margin:

Toggle Parlay mode in the GridRush sidebar, tap 2–6 cells, and submit them as a single bet. Every leg must win for the parlay to pay out. The trick is that the platform charges its 5 % margin once on the combined multiplier — not per leg — so parlays pay strictly more than the equivalent stack of independent singles:

A 3-leg parlay of 2× / 2× / 3× pays 13.68× vs 10.97× from independents (~25 % more). If any leg refunds (no ticks in that window), the whole parlay refunds and your stake comes back.

Deriv's Boom and Crash indices generate uni-directional spikes at a known average tick interval — Boom 500 spikes once every ~500 ticks, Boom 1000 once every ~1000, and so on. Each tick is effectively a weighted coin flip, which maps cleanly onto a Poisson process:

Two bet families ride on top:

The detector uses a rolling-median outlier test on log-returns so it flags real spikes in real time without being poisoned by the spike itself. Everything resolves instantly when the spike hits; otherwise the round ends when the window closes.

Everything on tickr — every chart, every quote, every game outcome — is derived from Deriv's public WebSocket (wss://api.derivws.com/trading/v1/options/ws/public). We subscribe to ticks for every supported symbol on connect, cache the latest price in a shared MarketDataContext, and let every widget read from that.

For the games, demo-only means no PAT token is needed — tick data flows through the public endpoint using a registeredapp_id. Only the chat's “place a real trade” flow would require user auth, which is opt-in per action.

• Next.js 16 app router, React 19, TypeScript
• Tailwind CSS v4 + shadcn/ui + lucide-react
• Deriv WebSocket API v3 singleton client with heartbeat + exponential-backoff reconnect
• HTML5 Canvas for grid & gauge rendering (imperative draw loop, no per-cell React)
• Supabase for chat history + future gamification tables
• Anthropic SDK streaming with tool use for inline chart / trade widgets