Every breakable item, ranked by what one break earns. No model — this is arithmetic. Search by name to look one up, or tick Loss and Unknown under Verdict to see the rest. Each cell shows our number with KamaMaster's underneath in purple. A Δ badge means they disagree by more than 30% of the value — worth checking before you buy. Click a column to sort, or a row to see its history.
How to read this — and what "confidence" means
- Break-even is the number to read first. It is the coefficient at which this item stops losing money. Green: worth breaking right now. Amber or red: not yet — hover the cell for exactly how much it still needs. If it says never, the item cannot pay for itself even at 4000% — the game's ceiling — so it never will.
- Now is the coefficient. Blue means you read
it in game yourself, so it is true. Grey
apiunderneath is what the website claims, with the signed gap between them. - The tier compares the estimated coefficient range against break-even. Entirely above it is a good bet, entirely below is a loss, and straddling it is a lottery — it could land either side. Certainty is not part of that judgement; that is the next column.
- Confidence is not a chance of profit. It is how much we know about the item: how many times we have seen it, how many changes we caught, whether our prices are fresh, and whether any of us has ever broken one. 22% on a good-bet row means the interval says yes but almost nothing backs it up. A loss at high confidence is the most reliable thing on this page — no model is needed to know that cost above rune value loses.
- Estimated is what the coefficient probably is in game at this moment. Two separate uncertainties go into it: the range is how far the game could have drifted since the site last recomputed (thousands of observations behind it), and the shift is the measured gap between site and game — which rests on one paired reading, so it moves the centre and adds no width. When there is no range to give, it says "no interval" rather than pretending to one.
- Everything predicted carries its
n— how many observations are behind it — in the cell's tooltip: hover any estimate to see it. If n is small, the number is a guess wearing a number's clothes. That is deliberate.
The most useful thing you can do is log a break. Read the coefficient in game immediately before breaking, then record it under "Log a break". One real reading is worth more than a week of scraping — that is not a figure of speech, it is what happened here on 31 July.
My profession levels
| Coefficient (%) | Per item broken | Trust | ||||||
|---|---|---|---|---|---|---|---|---|
| Item | Now | Break-even | Estimated | Costs you | Gross | Net | ROI | Confidence |
The only data here that is not scraped, and the most valuable. Record the coefficient the game showed for the item it destroyed. For a batch, record both ends — that is what measures the cost per break, which belongs to the item and is not a constant: three items so far give 0.22, 0.5 and 1.0. Correct any ingredient prices that look wrong while you are at it.
Screenshot the result panel, give it to an AI with the prompt below, and paste what comes back. Pick the item in the search box first if you can — then the id is certain and no name matching happens at all.
Prompt to give the AI
Read this Dofus "Résultat du concassage" (crushing result) screenshot and return ONLY a JSON array, no commentary.
One object per distinct item name in the panel. For each:
"item_name" the item name exactly as printed, unchanged
"lang" "fr", "en" or "es" — infer it from the panel's wording
"quantity_broken" how many rows that item has (identical items share one row; count the rows)
"coef_at_first_break" the coefficient on that item's FIRST (topmost) row
"coef_at_last_break" the coefficient on that item's LAST (bottom) row
"rune_value_estimated" the "Valeur estimée des runes obtenues" total, integer, no spaces
Rules:
- Do NOT include runes. Rune icons cannot be identified reliably from a screenshot and a wrong count is worse than none.
- Do NOT include the destroyed-items value ("Valeur estimée des objets détruits").
- If quantity is 1, omit coef_at_last_break entirely.
- Copy numbers exactly as shown. Do not round, convert or infer a value you cannot read.
- If a field is unreadable, omit it rather than guessing.
Example:
[{"item_name":"Kwakoiffe de Vent","lang":"fr","quantity_broken":5,"coef_at_first_break":137,"coef_at_last_break":135,"rune_value_estimated":56890}]
Where the money is
Calibrated net profit per break across the current ranking — how many items sit in each band. Most items cluster near zero; the money lives in the long right tail.
Recovery
How far the published coefficient moves, relative to where it was, by how long since the previous report. Band is the observed p10–p90, line is the median; n per bucket.
Server activity
Reported coefficient changes per day, all servers pooled — our proxy for break volume, which nobody can observe directly. Busy days mean faster recovery everywhere.
Our earnings
Cumulative net benefit per player from the logged breaks — craft cost and sale tax already subtracted, so a falling line is honestly falling.
How one item's coefficient has moved over time. Amber dots are scraped observations (KamaMaster); the purple line is DoFocus's crowd-reported series where one exists; blue marks are your own recorded breaks. Gaps mean the item dropped out of KamaMaster's top 200 — that is data we simply do not have, not a flat line.
Dofus has a dozen currencies that are not kamas — Almatons from the Almanax, Kolizétons from the Kolizéum, Orichors, Kamas de glace. Each buys things from an NPC, and some of those resell. This ranks a shop by kamas per token, net of the HDV's 2% tax — not by sale price, because an item costing five times as many tokens has to sell for five times as much just to break even. Type a price in the Price column while you are looking at the HDV; nothing here is scraped, so your numbers are the only numbers.
What this ranking does not know
- How fast anything sells. No source publishes HDV volume, so a 3M item that moves once a month can outrank a 60k item that moves twenty a day. Rows say liquidity unknown until you judge them — that is a real gap, not a placeholder.
- Resale locks. Pets bought from Xanamla were reportedly locked for two months before they could be listed. Never verified, and a lock would not show in the game data — so a fast-looking rate may be two months away.
- How many you can buy. Shop stock and reset caps are unrecorded, so this ranks and does not plan. And the cheap end floods: converting everything into Galets Saisonniers is ~70 a day onto one server — your own supply competing with itself.
- Bound items are excluded, not penalised. A title cannot be listed at all, so it has no rate — it is not simply a bad one.
Paste prices from the game's market panel
Screenshot the market list in game — a whole page of items at once — and have an AI transcribe it with the prompt below. That is the normal way to use this: a page of prices per paste, no opening items one at a time. A list gives a prix moyen and nothing else, which is fine; if you ever do open an item, its panel also carries médian and articles vendus and those get stored too. Check first: it reports what it understood and writes nothing.
Transcribe this Dofus market screenshot into JSON. One object per row/item.
Most screenshots are the LIST view: a column of items with one price each,
under a heading like "Prix moyen". For those, every entry is just:
[{"name": "Petit Parchemin de Sagesse", "mean": 6311},
{"name": "Petit Parchemin d'Agilite", "mean": 6407}]
If instead this is ONE item's detail panel, it also shows "Prix median" and
"articles vendus" and a selected tab (24h / 7 jours / 30 jours):
[{"name": "Parchemin de Force", "period": "30d",
"median": 18589, "mean": 19140, "qty_sold": 32296}]
Rules:
- Copy names EXACTLY as displayed, in French, with their accents. Never
translate them and never correct the spelling.
- Strip spaces from numbers: "18 589" is 18589.
- Put the price under "mean" when the column says Prix moyen, and under
"median" when it says Prix median. Do not put one in the other's field.
- OMIT any field you cannot read. Never guess, and never carry a number over
from a neighbouring row — a wrong price is worse than a missing one.
- Omit "period" unless you can actually SEE which tab is selected. A list
screenshot usually shows none; omitting it records the window as unknown,
which is true, rather than pretending it was 30 days.
- If a row is cut off or unreadable, leave it out and say which ones you
skipped after the JSON.
- Output the JSON array first, then any notes.
| Item | Price (HDV) | Net | Kamas / token net of 2% tax | Craft demand recipes using it | Type |
|---|
| When / item | Coefficient (%) | Benefit | |||||||
|---|---|---|---|---|---|---|---|---|---|
| When | Item | By | Qty | Estimated | Coefficient | Net / item | Avg ROI | Runes | |