Case study · Telegram

Ghost Arb Bot.
It watched
the books.

Two Telegram bots, built for arbitrage betting — plus a sub-bot we put straight into a group of paying members. They scraped live prices, found the gaps, and the alert landed where people were already waiting. They ran on their own — and they made money.

Seven-day profit and loss of plus £1,054.25
Betting account showing £1,630.92 profit this month
Sports betting profit and loss breakdown, £537.36 month to date
Copper paths branching between a central tile, an envelope and two signal nodes.
THE ART OF CONNECTIONComplex behind the scenes.
Simple where it matters.
Original conceptual illustration. The case-study screenshots above remain the real project evidence.
The short version

A pair of bots
that paid for themselves.

Arbitrage betting is simple on paper and miserable by hand. Two books disagree, you back both sides, and the difference is yours. The miserable part is spotting it in time. That's what Ghost Arb did — then the slips it caught, live.

3

The watching pair, plus a sub-bot inside a paying members' group

Live

Ran unattended. Prices in, alert out, no one sat watching a screen

API

Bookmaker and exchange prices pulled in as they moved

TG

The alert landed on a phone with the market, the books and the stakes

The errors

The slips it
caught live.

Mispriced markets the bot found while they were still open. Arrow keys, swipe, or the buttons. After that: how it actually worked.

Open bets with cash-out already in profit
Live betting slips found mispriced, cash-out in profit
Tennis selection at 8 to 1 with cash-out in profit
How it worked

Watch. Spot.
Size. Fire.

Four jobs, in that order, every time. If any step was late, the window was gone — so the whole thing was built to be quicker than a person. After that: the working parts, as they shipped.

01 / Watch

Keep a live book

The bot sat on bookmaker and exchange APIs and kept a running picture of prices. Not a daily scrape — a watch, so a move was seen as it happened.

02 / Spot

Find the disagreement

When two prices on the same event didn't agree, that was the arb: a short window where both sides could be backed and still leave a margin.

03 / Size

Work out the stakes

It calculated how much to put on each side so the return was the same whichever way the event went. No back-of-a-napkin maths at 11pm.

04 / Fire

Ping Telegram

A message landed with the market, the books, the prices and the stakes. The same ping also went out through a sub-bot, into a Telegram group of paying members — so it reached a room, not just one phone.

The sub-bot was built for that group and wired in directly. Members were already paying to be there; Ghost Arb just made sure the useful alert arrived in the same chat, without anyone copying it across by hand.

What it actually did

The working
parts.

Built in Python, talked to Telegram over the Bot API, and left to run. This is the list as it shipped — not a wishlist. Need something like it? That's the last stop on this page.

Live odds over API

Pulled prices from bookmaker and exchange feeds as they changed, rather than waiting on a page refresh.

Cross-book matching

Lined the same event up across books so a mismatch was obvious, instead of two tabs and a guess.

Stake maths

Worked out both sides so the payout didn't depend on the result — that was the whole point.

A floor on the edge

It didn't ping every half-percent wobble. If the numbers weren't worth acting on, the phone stayed quiet.

Telegram alerts

Market, books, prices, stakes — one message, on the phone, while the window was still open.

Unattended running

Started it, left it. The pair kept watching overnight and through the afternoon without anyone at a desk.

A pair, plus a sub-bot

One watched the books. One spoke Telegram. A third sat in a paying members' group and posted the same alert there — built for that room, not bolted on later.

Paid group, live

The sub-bot was integrated straight into a Telegram group with paying members, so the people who were in on it saw the arb in the chat they already used.

Python · Telegram Bot API

The same stack we still use for bots: scrapers and alerts in Python, delivery over Telegram so it reaches a pocket — or a group — not an inbox.

Need something like this?

We can build
the next one.

That's the path: what it was, the slips it caught, how it worked, the parts. Alerts, scrapers, booking bots, anything that should land on a phone instead of an inbox. Custom builds start at £349 — a few sentences is enough to start.