How we judge a deal

A deal on RealDeals is a product that is in stock at iPon and priced well below what it usually costs, measured against its own recorded prices. Every product gets a score out of 100 from the same formula, and the thresholds on this page are the live ones.

Where the prices come from

iPon publishes its full buyable catalogue as a product feed several times a day. We download it, store every price in forint, and aggregate each product into one row per day: the closing price and whether it was in stock. Out-of-stock days never count as price observations.

No retailer claim enters the judgement. iPon's own crossed-out prices, campaign labels and percentages are ignored. If we have not seen a price ourselves, it does not exist.

What we measure

• Typical price. The median in-stock price over the last 90 days. A median ignores single spikes, so a price that was high for one day cannot manufacture a discount.
• Prior price. The lowest in-stock price in the 30 days before the current price began. This is the reference the EU Omnibus directive requires retailers to show, applied to our own data.
• Real discount. The smaller of the two: how far the current price sits below the typical price and below the prior price. The headline number is always the conservative one.
• Stability. The share of the last 90 observed days within 3% of the typical price. A drop from a stable baseline means more than a drop from a jumpy one.
• Freshness. How many days ago the current price began. Full marks up to 3 days, nothing from 30 days on. A drop that is a month old is a low price, not a deal.
• Range position. Where the current price sits between the 180-day low and high.
• Confidence. How well the price we compare against is backed by observations, and it multiplies the whole score. After a price change it counts the days we saw the product in the 30 days before it (full at 14), so a new product whose launch price held for two weeks can score fully. Otherwise it counts the days we saw it in the last 90 (full at 30). A price we saw on only a few days cannot carry a high score.

The score

Confidence times a weighted blend of five terms. The discount term is full at 40% real discount, the saving term is logarithmic and full at 300 000 Ft, so a large saving on an expensive product counts without letting percentages on cheap accessories dominate.

Exceptional from 85, Strong from 70, Fair from 55. The Deals pages show all three, starting at Fair; narrow them to Strong or Exceptional in the filters.

What a deal must pass

The score alone is not enough. Every deal on the site also clears these gates:

• Real discount of at least 10%.
• Saving of at least 2 000 Ft.
• At least 50% of recent observations near the typical price, or a price at least 3% under every price in the 180 days before it began.
• At least 20 observed days in the last 90.
• The current price began within the last 30 days.
• In stock at iPon right now, and not a data error (a price under a tenth of typical is rejected).
• Not just back from a spike: after a one-day jump to 3 times its earlier level or more (usually a pricing error), the price must be at least 3% under that earlier level.

The badges

• Lowest recorded. The lowest in-stock price we have recorded, and as far under the typical price as a deal must be (the minimum discount and the minimum saving). A small dip, a price that never moved, or one the product has settled at is not a low worth showing.
• Lowest in 6 months / Lowest in 3 months. The lowest price in the last 180 or 90 days, as far under the typical price as a deal must be, with enough observed days in that window to mean something.
• Just dropped. The current price began within the last week and is at least 5% below the prior price.
• Sliding down. No single big drop, but the 90-day trend is down at least 5% per month.
• Too new to judge. Fewer than 20 observed days. We show the prices we have and no verdict.

Price trends

Every category and subcategory gets a price index. Each day we compare every in-stock product with its own price the day before (a price we did not see carries over for up to a week) and take the geometric mean of those changes. Chained day by day, that says whether the group got cheaper or pricier, and products arriving or leaving the catalogue cannot move it. The verdict looks at the last 30 days:

• Getting cheaper. The index fell at least 2% and more products got cheaper than pricier.
• Getting pricier. The index rose at least 2% and more products got pricier than cheaper.
• Holding steady. Anything in between. A group needs at least 30 products followed for a verdict.

Deal lifespan is the median number of days a price held after a drop of at least 10%, over the drops that ended in the last 90 days; it is shown once there are ten of them. The brand table counts, per brand, the share of in-stock products that dropped at least 5% in 90 days and the median size of those drops.

What this cannot do

The engine knows nothing about other shops. A real deal here means cheaper than this product usually is at iPon, not cheaper than everywhere else. It also cannot see a product's launch price before we started tracking it.

Electronics get cheaper as they age. Freshness and the prior price exist so that a slow slide does not read as a series of deals; the Sliding down badge is the honest label for that.

About RealDeals

RealDeals is an independent project. It is not affiliated with iPon and earns nothing from what you buy. Prices are recorded from iPon's public product feed; nothing is scraped from the shop pages.

Questions, corrections, or a category that looks wrong: write to About.

Currency conversion is for display only; every judgement is made in forint. Rates refreshed daily (2026-11-08). Rates by Exchange Rate API.