Galaxy Polymarket retail losses under a 50-order daily cutoff

On 1 Oct 2026, Galaxy Research published a study by Will Owens, and that paper is the source of Galaxy Polymarket retail losses. It looks at Polymarket’s international platform. It does not hand you one contract to buy. Of accounts under a 50-orders-per-active-day cutoff, the study counts 2.9 million. In that group, 69.2% finished below break-even, and the group was down $338.9 million. The median loss was about $3. The middle half sat between −$36.64 and +$0.40. Accounts over the cutoff, 125,429 of them, finished +$246.8 million.

TL;DR

What actually happened

The primary writeup is Galaxy’s. CryptoBriefing, the same day, repeats a subset: 69.2%, $338.9 million, 2.9 million accounts, the median, the churn pair, and the $246.8 million. Where this page cites a figure CryptoBriefing is not credited with repeating, the figure stays attributed to Galaxy.

Read the study here: Galaxy Research. The same-day repeat is CryptoBriefing.

The cutoff is the spine. Accounts under 50 orders per active day are the 2.9 million. That is the group in which 69.2% finished below break-even and the group that was down $338.9 million. “Below break-even” is the study’s phrase. This page does not rename the other accounts as winners, because finishing at break-even, if any did, is not spelled out here.

The median loss is about $3. About is part of the sentence. The middle half sat between −$36.64 and +$0.40. A tiny median next to a large group loss is the shape Owens is showing: many small results, and a group total that is still largely negative. This page does not invent a mean to “explain” the gap between about $3 and $338.9 million.

Accounts over the cutoff are a different population. There are 125,429 of them, and they finished +$246.8 million. The study does not, in these facts, say that crossing 50 orders causes profit. It says which side of the cutoff finished where. Do not turn a cutoff into a strategy.

How the cutoff splits the story

Churn is measured after the outcome, inside 30 days. After a loss, 15.2% had not traded again within 30 days. After a win, 6.1% had not traded again within 30 days. CryptoBriefing repeats that pair. The facts do not say why someone stopped, only that the not-traded-again share was higher after a loss than after a win in this study.

Specialists get a definition, and the definition matters more than the label. A specialist has more than 60% of markets in one topic, and at least five markets. Under that rule, specialists were 44.1% of traders. The profitable rate is not one number for all of them. Sports specialists’ profitable rate was 25.1%. Tech and science specialists’ profitable rate was 41.2%. Those two rates are Galaxy’s. This page does not claim CryptoBriefing repeated them.

Median position size is also Galaxy’s in this telling: $13.96 for profitable accounts versus $10 for unprofitable accounts. The difference is printed as those two medians, not as a rule that betting a few dollars more creates an edge. Both medians are small next to the group totals. Small is the point of reporting them, not a suggestion to copy either size.

Galaxy Polymarket retail losses is the right name for the under-cutoff result, and a sloppy name if you forget the over-cutoff result. The 2.9 million accounts and the 125,429 accounts are both in the study. Quoting only 69.2% and skipping +$246.8 million edits Owens. Quoting only the smaller group’s gain and skipping $338.9 million edits him the other way.

The platform scope is “international,” as the study states. This page does not extend the percentages to a different venue, a different year, or a single election contract. There is no one market ID in these facts, and no Yes price to clip onto the study.

What a second outlet repeated

CryptoBriefing’s same-day piece is a repeat, not a second sample. It repeats 69.2%, the $338.9 million group loss, the 2.9 million accounts, the median, the churn pair of 15.2% and 6.1%, and the +$246.8 million on the far side of the cutoff. When both pages carry those lines, you still have one study.

The middle-half band, from −$36.64 to +$0.40, is cited here from Galaxy. The specialist share, 44.1%, the sports rate, 25.1%, the tech and science rate, 41.2%, and the two position medians are cited here from Galaxy. If a recap wants those, it should say Owens, not “reports agreed.”

Author and date stay attached. Will Owens. Galaxy Research. 1 Oct 2026. CryptoBriefing the same day. A stat stripped of that byline becomes internet folklore fast, especially a scary group total like $338.9 million next to a median loss of about $3.

No contract block sits under this story. There is no single Polymarket question, no Yes price, and no event URL to treat as the trade. The research URL is the source. Anyone who turns the study into a market ticket has changed the genre.

What to do as a reader (not a trade)

Read the cutoff before the headline percent. “69.2% finished below break-even” applies to accounts under 50 orders per active day, a set of 2.9 million, on Polymarket’s international platform, in this study. It is not a law of prediction markets, and it is not a description of the 125,429 accounts over the cutoff.

Do not trade it. The paper is not a signal to take the other side of retail, to raise your order count past 50, or to specialize in tech because 41.2% is higher than 25.1%. Those are historical rates inside definitions Owens set. They are not a playbook, and this page will not write one.

If you quote one number, quote the pair that stops the misuse. Group loss $338.9 million, median loss about $3. Under cutoff, and over cutoff. After a loss 15.2%, after a win 6.1%. Sports specialists 25.1%, tech and science specialists 41.2%. One side of any of those pairs is a different story.

Galaxy Polymarket retail losses should not be shortened to “retail lost.” The median is about $3. The middle half includes a top end of +$0.40. The over-cutoff accounts finished positive in aggregate. All of that can be true while 69.2% of the larger group finished below break-even. Hold the whole shape.

This site’s other notes are not part of the sample. The DogeOS public testnet is unrelated. The Gemini 4 Argon launch is unrelated. Do not drag them into a trader study to make the chart feel current.

Keep Galaxy Polymarket retail losses attached to Owens, the date, and the cutoff. A fair notebook line is boring on purpose. Date 1 Oct 2026. Author Will Owens, Galaxy Research. Venue: Polymarket’s international platform. Cutoff: 50 orders per active day. Under it: 2.9 million accounts, 69.2% below break-even, down $338.9 million, median loss about $3, middle half −$36.64 to +$0.40. Over it: 125,429 accounts, +$246.8 million. Churn: 15.2% versus 6.1% inside 30 days. Specialists: 44.1% of traders, with the two topic rates above. Positions: $13.96 versus $10. CryptoBriefing repeats the subset named above. No buy, no fade, no “so bet specialists.”

The useful discomfort in the paper is the scale gap, and you do not need a new statistic to feel it. A median around a few dollars and a group total in the hundreds of millions describe different layers of the same count. Report both layers, name the cutoff, and leave other people’s orders alone.