How we calculate

The model compares confirmed purchases and eligible listings. It considers the collection, model, backdrop and only those number features supported by sufficient market observations. Purchases and listings remain distinct: a seller's asking price is not evidence of a completed sale.

The model's sales history provides the main reference. Telegram's listing book and recent external listings may adjust it; external markets also help when a model has few or no purchases. The gift's own current asking price is excluded from its comparables. A gift listed on several marketplaces is not counted as several independent items.

If there is insufficient evidence for the model and too few eligible listings, we show a preliminary estimate based on the collection. If no positive reference value is available, we do not display a numerical valuation.

Valuations use ordinary purchases with reliable timestamps and positive prices above 0.05 TON. Premarket transactions, giveaways and Lucky Buy are excluded from ordinary purchases. Repeated records for the same gift, source and price within one minute are consolidated.

A purchase's weight halves every 90 days. Recent data therefore has more influence, while older data does not disappear at a single cutoff. Listings undergo separate price, freshness and status checks: buyer bids and unconfirmed latest observations are not treated as seller listings.

The closest comparables are purchases from the same collection, model and backdrop. The chart may show purchases of the model as a whole, including other backdrops.

We first check the model's own listings. Telegram requires at least three eligible listings; external marketplaces normally require five confirmed listings no older than seven days. A smaller external sample is allowed for a rare model without sales only when prices are closely aligned. These listings may provide a preliminary reference, but cannot establish high confidence on their own.

When a model has no purchases, the TBE fallback uses the weighted median of ordinary purchases across the collection and adjusts it for the gift's backdrop. This is a collection-based extrapolation, not a transaction price for that model; the default range is wider, from 0.7 to 1.5 times the central estimate. A rare symbol or number does not create a price premium on its own.

If the collection also has no purchases, TBE cannot produce an estimate. A small but consistent external sample may then provide a separate preliminary reference. If neither collection data nor eligible external listings are available, no numerical valuation is displayed.

The model sets the base price level, with the backdrop applied as a separate adjustment. A number premium is used only where repeated-digit numbers are supported by enough purchases; the model does not invent a premium without that evidence.

A trait's rarity does not increase the valuation on its own. The number adjustment applies only to repeated digits and only with a sufficient purchase sample; with limited data, the adjustment shrinks toward zero. Other number features receive no price premium without transaction evidence.

The central p50 value is a model-based estimate. The p25–p75 range describes the spread of comparable prices, adjusted for backdrop and constrained to limit excessive dispersion. It is not an exact sale price forecast or a promise that the gift will sell within this range.

Confidence describes the quality of the evidence: the number of distinct purchases, their recency and effective sample size, price dispersion and agreement between sources. High confidence requires independent, recent history, sufficient volume and no major source conflict. Confidence is not the probability of a sale.

The market verdict separately compares the asking price with Telegram listings for the same model. It does not replace the p50 valuation: a high Telegram asking price is not validated by an external purchase, and a cheap listing does not guarantee a profit.

Homepage categories apply additional criteria: data quality, comparable listing counts, purchase history and individual item traits. Inclusion in a category is the result of filtering the current snapshot, not an investment recommendation.

The Best Price badge compares an item only with the relevant group and snapshot specified in the interface. It does not mean the lowest price across the entire market.

The combined collection list includes listings from Telegram, Portals, Getgems and Marketapp, plus MRKT observations. Status is accompanied by a check timestamp; a latest observation is not presented as an item available now. Tonnel buyer bids are not shown as seller listings.

A listing link opens a specific item only when a verified address is available. The NFT link in gift details opens TONViewer, a blockchain explorer; it confirms the NFT address, not the existence of a sale listing.

The collection page shows aggregates and grouped models. It does not replace an individual gift valuation; listings from all connected marketplaces are available in a separate list.

Purchase history and median price describe recorded transactions in the selected window. Liquidity shows observed purchase activity and listing volume; it does not promise how quickly a particular gift will sell.

Transaction counts, distinct gift counts and listing counts are displayed separately. Duplicate transaction records and cross-listed items must not inflate the number of independent items.

The interval between two recorded transactions for the same gift is the time between observed transactions, not the exact time the gift spent on sale.

A gift disappearing from Telegram's listing book is not, by itself, counted as a sale. Internal Telegram metrics distinguish confirmed purchase records from listings that simply disappeared.

Analytics describes observed transactions and listings from available sources, not the entire market's total volume. Price changes compare the same model on the same marketplace; collections with different sales compositions are not compared directly.

Cross-marketplace totals use an approximate TON equivalent. For Telegram, it is calculated using available conversion data; it is not the total of on-chain transfers. The snapshot covers complete UTC days and shows when it was prepared.

Gift history consists of available public events: purchases, transfers and listing observations. An event missing from this history means it was not found in available data, not that it definitely never happened.