eBay UK Monitoring Bot with Telegram Alerts — Pokémon & One Piece TCG (Python / eBay Browse API)

OVERVIEW I buy and resell sealed Pokémon and One Piece trading card product in the UK. Good listings sell within hours and I can't watch eBay all day. I need a bot that watches for me and alerts me the moment something is listed below market, with a direct link so I can buy it immediately. WHAT THE BOT MUST DO 1. Monitor eBay UK continuously Poll every 10 minutes (interval configurable). Cover Buy It Now and auctions. Keyword list editable by me without touching code. 2. Cover these product types — Pokémon and One Piece, English and Japanese Elite Trainer Boxes (ETBs) — singles and full cases Pokémon Center exclusive ETBs — must be treated as a separate, higher-value product from standard ETBs Booster boxes — singles and full cases Booster bundles — singles and full cases Any other sealed collection box, tin, or premium collection Single cards, raw and graded 3. Multi-item bundle listings — priority feature A large share of the best deals are mixed-lot listings rather than single items — for example a listing containing two ETBs plus two booster bundles, or an ETB plus loose packs. These regularly carry the biggest spreads and are the easiest to miss. The bot must: Parse the full contents of a mixed listing from the title and item specifics — quantity and type of each component Value each component separately against its own market baseline Sum those component values to get the lot's total market worth Compare that total against the asking price to determine whether the lot is underpriced Show the component-by-component breakdown in the alert so I can check the maths Where contents can't be parsed with confidence, alert anyway but flag it as "contents uncertain" rather than skipping it. I'd rather check a few false positives than miss a real one. 4. Work out market value itself — no reference sheet I don't maintain a price list and don't want to. The bot must derive market value from eBay's own live data: For each product, build a rolling baseline from all currently active UK listings of that same product Use median price per unit, not mean, so outlier listings don't skew it Group strictly like-for-like — a given set's ETB compared only against the same set's ETBs, never against a bundle or a box Pokémon Center and other exclusive variants baselined separately from standard versions Require a minimum sample (5+ active listings) before treating a baseline as reliable; below that, flag the alert as "low confidence" Maintain rolling price history per product so baselines track the market as it moves Important: active listings skew high — sellers list optimistically and sell lower. Set the default alert threshold aggressively at 25–30% below active median, configurable. 5. Alert me via Telegram Each alert must include: Product name, and full contents breakdown if it's a mixed lot Direct clickable eBay link Total price plus postage Unit count and price per unit The baseline the bot calculated, and the sample size behind it How far below market, in £ and % Seller username, feedback count and percentage Seller location (UK / EU / US / Asia) BIN or auction, with time remaining if auction Any coupon showing on the listing Confidence flag where applicable 6. Prioritisation Rank alerts by total £ opportunity so the biggest spreads surface first. Exclusive variants and multi-item lots get priority weighting — those carry the largest margins. 7. No spam Never alert twice for the same listing unless the price drops further. Maintain a local database of everything seen. TECHNICAL Python preferred. Must use the official eBay Browse API — not a scraper. Scrapers get IP-banned and breach eBay's terms; if you plan to scrape, please don't bid. Deploy to a cheap always-on host, running 24/7 without my input. All settings editable by me. Sensible error handling with backoff and retry — it must not silently die, and should notify me if it stays down. ON SOLD DATA I'm aware eBay's API doesn't expose completed listings without partner approval. I am not asking you to scrape sold data. The baseline comes from active listings, with the threshold set to compensate for the upward skew. DELIVERABLES Working bot, deployed and running Source code pushed to a private GitHub repository owned by me — I will add you as a collaborator for the duration of the project Setup documentation clear enough that I can add keywords and adjust thresholds myself Short handover video 14 days post-delivery bug support TIMELINE 2 weeks. Milestones: (1) eBay API integration pulling and parsing listings including mixed lots, (2) baseline calculation and Telegram alerts, (3) deployment and documentation. TO APPLY — answer these. Applications that skip them will be ignored. Have you worked with the eBay Browse API before? Link relevant work. How would you approach parsing contents from a mixed-lot title listing multiple different products and quantities? This is the hardest part of the job and I want to see your thinking. What hosting do you recommend, and what does it cost monthly? What would you push back on in this spec?