The less a series is certified, the more expensive the certified copies are – and the difference is not marginal.On statements from September 5, 2026, Dave Sim's series has 6% certified ads, with a median of $169.00 compared to $12.00 for raw copies, a ratio of thirteen. Two recent Marvel characters noted on the same day display 22% and 26% certification, for ratios of ten and four and a half. The frequency of certification and its level vary inversely.
The common intuition is that a poorly certified series is so because it is of little interest. The figures tell something else: it is in the least certified corpus that certified copies reach the highest amounts.
This article measures this reversal on three corpora taken on the same day, proposes an interpretation, and above all indicates how not to make a mistake when transposing an observation from one corpus to another.
Trois corpus, trois régimes
The three readings were taken on September 5, 2026 on the American market, in the same category, a few minutes apart. All values aresommes demandéesby sellers; no sales are observed, and a photograph taken on a single day does not measure any evolution.
The first corpus, that of the independent series that concerns us, includes 285 raw announcements and 17 certified — 6% of the total. Gross Median: $12.00. Median of certified: $169.00. The ratio between the two reaches thirteen.
The second corpus, a Marvel character that appeared in the 2010s, has 133 raw announcements and 37 certified — 22%. Medians of $9.89 and $95.00, a ratio of almost ten. The third, another recent character, has 121 raw and 42 certified — 26%. Medians of $14.99 and $67.50, a ratio of four and a half.
Raw medians look similar
$12.00, $9.89, $14.99. The ordinary booklet is negotiated at the same level in the three corpora: it is elsewhere that everything is at stake.
The certified share varies from simple to quadruple
6% versus 22% and 26%. Out of 304 advertisements, only seventeen carry certification in the first corpus.
The certified/raw ratio is reversed
Thirteen where certification is rare, four and a half where it is common. The two curves go in opposite directions.
The rarest is the most expensive
$169.00 certified median in the least certified corpus, compared to $67.50 in the most certified. The rarity of the gesture goes with its level.
Before going any further, a reservation about the workforce. Seventeen certified announcements are not many: the median of $169.00 is a position in a small group, sensitive to the departure or arrival of a single line. The other two corpora, with thirty-seven and forty-two certified announcements, rest on more solid foundations. What the survey firmly establishes is the order of magnitude and direction of variation, not the exact value of each median.
The most economical reading of this reversal can be summed up in one sentence: when certifying is rare, we only certify what deserves it. The cost of the gesture being fixed, it only makes sense on a copy whose value justifies it - therefore on the high pieces of the corpus.
When certifying becomes common, the gesture also applies to ordinary copies, and the median of the certified segment mechanically descends towards that of the corpus. It is not the certification that loses its value: it is the population that receives it that changes.
What this reading allows, and what it prohibits
It helps to understand why a high certified median is not a good argument for certification. If the observed level is explained by a selection at the entrance - only the best pieces pass - then it describes these pieces, not the effect of the gesture on any specimen.
Above all, it prohibits transposing a report from one corpus to another. Multiplying the value of a copy by thirteen because it is the ratio observed in the least certified corpus would be a complete error of reasoning: this ratio does not measure a supplement, it measures a difference in composition between two baskets.
Finally, it does not establish any causality. I note a joint variation between the frequency of certification and its relative level, on only three corpora. Three points do not make a trend, and other factors differentiate these corpora - the age of the series, the number of copies in circulation, the existence of an investment market. None of these factors can be measured from an ad survey.
It is worth making the mechanics concrete, because it is verified by simple reasoning. Let us imagine two corpora identical in every way, except that the first certifies one copy out of twenty and the second one out of four. In the first, only sellers convinced that they have an exceptional piece incur the cost of the gesture: the certified segment therefore only contains the top of the corpus. In the second, certification has become a habit and applies to ordinary copies: the certified segment then resembles the entire corpus, shifted upwards by the service rendered, but nothing more.
The certified median of the first corpus will necessarily be higher than that of the second, without any difference in intrinsic value coming into play. This is not a market effect, it is a selection effect - the same mechanism which causes a very selective exam to display better average marks than an exam open to all, without the candidates of the first being better in themselves.
This reading has a practical virtue: it is refutable. If it is correct, we must observe, in the least certified corpus, that the certified copies are sought-after numbers rather than any sample of the series. I did the check rather than exempting myself from it, and it is clear.
The seventeen certified announcements focus on the very first ranks and on particular examples. The number 1 appears four times, between $2,595 and $4,650. Number 2 twice, at $225 and $700 — the second signed by the author. Number 4, billed as carrying a first appearance, also twice, at $224.99 and $399.99. A signed copy of rank 71 completes the top of the segment. The high ranks which make up the bulk of the series only appear at the margins, at the lowest amounts of the segment.
In other words: we do not certify this series, we certify its three or four founding numbers. The selection at the entry is not a convenient hypothesis, it is read directly in the composition of the segment.
I therefore present it for what it is: the most economical hypothesis compatible with three observations, not a conclusion. The distinction matters, because it is exactly the kind of reasoning that is then transmitted as an established rule even though it is based on three points and an intuition.
Six reflexes for reading a certified segment
Count before reading
Seventeen certified listings do not produce a reliable median. The number of employees is systematically quoted next to the amount.
Look at the share, not just the level
6% or 26% certification describes two different markets. The part completely changes the interpretation of the level.
Do not transpose a report
The factor thirteen of one corpus does not apply to another. It measures a composition, not the effect of a gesture.
Compare the same number to itself
The only comparison that isolates the effect of certification concerns an identical booklet, certified and raw, at the same time.
Remember the baskets
The certified ones of a poorly certified corpus are its high pieces. Comparing them to brutes is like comparing a peak to an average.
Date the observation
A segment of nineteen ads is quickly recomposed. A median without its date becomes unverifiable in a few weeks.
What these statements do not establish
They are not saying that certification would be more valuable on one series than on another. They say that the certified populations of three corpora do not have the same composition, which is an observation about the corpora, not the process.
They do not allow any projection on a given copy. Having an ordinary fascicle from the low-certified series certified will not bring it closer to the observed $169.00: this amount describes pieces that were already exceptional before being certified.
A correction is also necessary on my own statement. My first read yielded 19 certified listings and a median of $159.00: two of those lines were actually for another series published by the same publisher, captured by an overly broad query. The series only segment has 17 listings and its median is $169.00. The order of magnitude does not change, the rigor of the figure does — and this is exactly the kind of contamination that a request by editor produces without reporting it.
Finally, the similarity of the three raw medians — $12.00, $9.89, $14.99 — does not mean that these three markets are equal. It only means that their median fascicle is negotiated at the same level. All that really distinguishes these corpora is found above the third quartile, where the logics diverge.
Translate this into a file
The rule:note the certified portion of the corpus next to any estimate from a certified segment.A figure like “certified median $159” cannot be read again in six months if we ignore that it related to nineteen advertisements representing six percent of the market.
For a certified copy that you have, record four things: the fact that it is certified, the grade, the organization and the date. The last three are what will make the rating comparable later; without them, the mention is a checked box that provides no information about anything.
If you are considering certification, the only honest calculation is comparing the same fascicle to itself: this precise certified number at a plausible grade, against this precise raw number. Any reasoning based on overall medians will overestimate the gain, because it compares baskets of different contents.
One last point, valid beyond this case. When a segment has fewer than twenty observations, the most useful thing to record is not its median but its number. Writing “certified segment: 17 ads” is accurate and durable information. Writing “certified value: $159” makes a small group of opinions look like a market price.
The simplest reading is a selection at entry: when certifying is rare, we only do it on the documents that justify it, the cost of the gesture being fixed. The certified median then describes these tall rooms, not the effect of the certification. It is a reading consistent with the three corpora noted, not a demonstration: three points do not make a proof.
No, and that is the mistake this report invites people to make. Thirteen is the ratio between two medians relating to different sets: the certified ones are the upper parts of the corpus, the rough ones are the whole. The only valid calculation compares an identical fascicle to itself, certified and raw, at the same time.
Not at all, and the statement shows the opposite of intuition: the least certified corpus — 6% — displays the highest certified median, at $169.00. The certified share describes a market practice, not the quality or value of a series. It is read next to the level, never in its place.
In the center, yes: the three raw medians stand between $9.89 and $14.99. Beyond the third quartile, no: the logics diverge, and a relationship observed in a corpus is not transposed. The useful comparison concerns internal relationships within each corpus, never on amounts from one corpus applied to another.
Its workforce, above all. Seventeen certified announcements do not produce a stable median: a line entering or leaving moves it significantly. Record « segment certifié : 17 annonces au 5 septembre 2026 » is correct and reads back; recording an amount alone gives the appearance of a set price to a small group of opinions.
The number next to the amount, always
My Comics Collection keeps your estimates with their date, their number and their origin, so that a figure remains judgeable months later.