What categories are the dog water traders on Polymarket predominantly active on? How much flow in this category are squares?
Yesterday, we examined the markets with the highest sharp concentration (short-term crypto and sports); let's take a look at their opposite counterparts.
Again, we segment traders by volume, then volume-weighted immediate, 1m, 10m, 1h, 2h, 1d line movements, and aggregate line movement-based PNL (recall this means that if volume-weighted line movements across time horizons are negative, your PNL is negative and has no bearing on realised PNL).
If the price is worse every time you buy, within the next 10 minutes to 1 day, you’re likely not the best trader.
Results:
Let's compare which categories the best and worst traders by 10m, 1d line movement are doing volume in side by side.
The best short-term traders are trading sports, but also quite a lot of crypto (recall the best 1d traders are BARELY trading crypto). But the worst 10m traders are trading an insane amount of crypto (27.1%).
Recall that the data comes from 2 weeks of November and is filtered for those who have done more than 10,000USD in volume (~2900 wallets). If we see balanced volume percentages on both the top and bottom takers, it means money is cleanly coming out of one pocket and going into the other.
In crypto, we interestingly see something different. This can mean either that the worst high rollers are losing more money than the top guys can rake in, OR that the top guys are so interested in sports that they’re leaving crypto markets to other traders. Both are probably true.
If you have a strong latency edge in crypto markets, there's a lot of dumb flow here.
The above point becomes clearer when you look at top vs bottom 1d traders - the market is balanced ... except for crypto.
If you look at immediate-next trade line movements, the verdict is dire for those trading short term crypto markets. This is the only dataset where there is more crypto volume than sports.
As before - this is the percentage of volume per market done by the worst (10, 20, 50, 100, 200) traders by classification (immediate-next, 1m, 10m, 30m, 1d line VWAP line movement, volume and line movement-based PNL).
Some observations (and scaled / clearer charts so you can make out the numbers):
The worst of the worst short-term traders are in crypto
The worst of the worst long-term traders are in sports
The dogwater medium traders are also in finance.
Notes on the data and methodology as a caveat
Again, this data is based on a two-week window in November. If you would like to see more, please send over a Macbook M3 Max. Specifically the way to interpret this data is - on a per user volume weighted basis, who has the worst line movements within an X timeframe? This is strictly different from "who lost the most money".
For markets that resolve after a trader enters into the trade but before the designated “markout” window (ie. a trader enters a market that resolves in 15m but we are examining 30m line movements) we use the resolution price. The longer the horizons are, the more realised PNL is a consideration, though in practice we found there is not a huge overlap between those doing horribly / or great in longer-term markets and participation in shorter-term markets.
The windows have a 20% grace period to find matching trades. (ie. if 1d window then a 4 hr window).
Originally posted on X.