Editor’s note: This is the first in an Oddchain series devoted to “insider trading” on prediction markets.

I would wish you a good morning, but you don’t deserve it. I hope you step on a Lego. 

You have likely been sent this article because you wrote a Twitter thread about a supposed Polymarket insider, and your thesis is that the relevant trader is: 

It probably looks something like this: 

Turns out, the account referenced above was not an insider. The falsely-identified ringer lost all of their money, and rather than forcing all this LLM-generated logorrhea onto the timeline, with just two minutes of analytical legwork the author of this Tweet could have determined the relevant account (who turned out to have made their money trading NFT monkeys) almost certainly did not have insider knowledge about the health of Benjamin Netanyahu. Similarly, the trader you recently “discovered” with your vibe-coded bot is, likewise, probably not an insider. 

Look, in some ways I get it. The single easiest way to grow an account on Prediction Market Twitter is to claim to have found an insider. They do exist! But in this case you didn’t, and it’s trivially easy to prove as much. You are a trickle of urine in the well of public information. You are a mote in the eye of collective knowledge. If you encourage others -- as you did indirectly here -- to make predictions based on your dope ramblings, they will lose money. You should feel poorly.

The primary purpose of this blog is to shame you. The secondary purpose is to educate you (and anyone who comes across this piece) on how to use onchain analytics to screen possible insiders and scry a touch more information than the herd. We will use free public tools to demonstrate how to trace flows.

This is real, genuine alpha, and you do not deserve it. It’s infuriating to think that you might earn some sort of “social engagement” airdrop allocation with your ongoing nonsense. As penance, you are obliged to please follow us on Twitter and subscribe to this newsletter so we can gain more followers than indolent engagement-baiters such as yourself, and thereby be less angry. Fuck you. Please enjoy. 

Part of the difficulty with discerning between high-rolling punters and actual insiders is that, on the surface, there is very little information you can learn about a given Polymarket (and especially Kalshi) account. Most platforms that attempt to apply a “score” to whether or not an account is an insider do so with probabilistic guesswork: 

Polysights, which has one of the most popular “suspected insider” algorithmic scoring systems, measures the likelihood of an account being an insider based on whether: 

Essentially, it’s looking for new accounts making a big bet on an unlikely or potentially profitable market. 

All of these metrics rely solely on the data publicly available from Polymarket. However, because Polymarket is a blockchain-based platform, there is much more information that can be gleaned about any given account -- if a researcher is willing to do a hint of legwork. 

Let’s take this suspected insider, for example: 

On April 1st, a new account with a strange username, “war-machine,” placed a massive bet on a ceasefire by April 15th. Using the blockchain analytics platform Arkham, we can see the following chain of transactions: 

In this case, this user deposited 60,371 in the stablecoin USDC to a Polymarket account via the cross-blockchain bridge Relay.link. Because Polymarket is hosted on the Polygon blockchain, many addresses are funded with this pattern, as most funds come to Polygon via more popular blockchains. Notably, Polygon is essentially dead aside from Polymarket activity.

This means that the task for anyone trying to uncover more information about an account is to track deposit funds through the Relay bridge and identify the source of the funding. 

This can be accomplished by filtering Relay transactions by dollar value, allowing for a small delta (often, cross-blockchain transactions involve a swap, and Relay charges a fee for bridging, so the deposit amount and the received amount on Polygon will be slightly different.) 

Here, we filter the dollar amounts until we find two transactions roughly 15 seconds apart, coming into and out of the bridge, with about a $20 difference in value. This indicates we’ve identified the original funding address of the Polymarket account that placed a suspicious bet. 

Immediately, we can learn a great deal of additional information about this user.

We know that they are funded by Coinbase, so we can report them if we do suspect them of insider trading and their account can be frozen. We can also hypothetically track if they open other accounts in an attempt to hide larger bets, and get a sense of their other onchain activity. 

Notably, however, this account appears to own an Ethereum Name Service address, “iamnotaninsider.eth.” That is because OddChain knows who created this account, funded the Polymarket address, and placed the bet as an April Fools prank specifically designed to dupe engagement-baiters. It worked depressingly well, triggering multiple bots and threads: 

Only one person, up-and-coming analyst Kyle, pointed out that the silly “war-machine” name indicated that the supposed “insider” probably wasn’t really an insider at all. No one bothered to track the deposit through Relay. 

Anecdotally, when independently investigating social media claims about insider trading, Oddchain has had success in using this method to gain additional insights roughly in between a half and a third of all cases. This is not particularly sophisticated or complex analytical work, and can be done by hand with free tools -- and yet almost no one bothers to do it.

There is a wealth of additional data available to users who know how to use free tools, and (for the sake of our sanity) much better content to be written by analysts willing to do it.