DSS 2026 — Follow the Collateral: A Mental Model for Auditing Polymarket V2 Paul Clemson Source video: https://www.youtube.com/watch?v=xbOIklS2r6g Automatic captions — not human-reviewed. Complete spoken-audio coverage is unverified. Source language tag: en (English). Source export SHA256: 6681743e30e9d103fc7a8448094235c717c12837d6a73b1532be187e6d2aecb7 [0:00] Thank you. [0:01] Hello, everyone. [0:03] Uh, who in the room is a security [0:05] researcher? [0:07] That's a good number of hands. How many [0:10] of you have looked at Polymarket V2? [0:14] No hands. No, one hand. [0:15] Okay. So cool. I think as someone at [0:20] Quantstamp who one of a few companies [0:23] that have been looking at Polymarket a [0:26] lot over the past year, [0:28] I think it's really cool and I think [0:30] it's a really under explored area of [0:34] crypto, D DeFi kind of. [0:37] And I think more people should take a [0:38] look. So what I want to do in the next [0:41] seven minutes apparently is give you [0:44] guys a bit of a a run through of a [0:47] mental model that you can go away with [0:50] and hunt that huge Polymarket bug [0:52] bounty. So let's go. [0:57] Oh, it worked. Okay, that's the first [0:59] slide. Okay, so the base of Polymarket [1:03] is a simple yes and no market. [1:06] [clears throat] So for example, will my [1:08] talk actually finish in five minutes? [1:11] They've given me seven, which makes me [1:14] think there is insider trading at play. [1:17] So I think they might have manipulated [1:19] this market. But yeah, essentially [1:22] every Polymarket [1:24] condition should always add up to the [1:27] sum of the collateral. So if you put $1 [1:30] in, you get 100 yes and 100 no. [1:33] And this always has to hold. This is the [1:36] key thing I want you guys to take away. [1:39] Oh. [1:41] Did Donald Trump moment, sorry. [1:42] [laughter] [1:43] What I want you to take away is this has [1:46] to always hold. [1:47] Whatever they add, every new module, [1:49] every bridge, every upgrade, if there's [1:52] any way that you can get more than $1 [1:55] for a $1 worth of both yes and no, [1:58] then there's trouble. [2:01] Okay, so yeah, there is three very core, [2:03] just kind of speed on this, three very [2:05] cool actions you can do. Split is $1 in, [2:09] 100 yes, 100 no. [2:11] Merge is the opposite of that. I've got [2:13] 100 yes and 100 no. I can get $1 back. [2:16] And then redeem is obviously after the [2:19] Oracle [2:21] confirms Pulse Talk did indeed last [2:23] longer than 5 minutes. [2:25] If you're holding no, you can redeem for [2:27] $1. If you're holding yes, then you [2:30] lost, so [2:32] GG. [2:35] And then [2:36] to go a little bit further, there's also [2:38] neg risk markets. [snorts] Neg risk [2:40] markets are essentially [2:44] a group of markets where only one of [2:46] them can be true. [2:47] So for example, what color will my [2:49] t-shirt be? [2:52] Unless I'm wearing two t-shirts or I do [2:55] like a Lady Gaga costume change, there's [2:58] going to be one answer to this question. [3:00] And today the answer is black. [3:02] About 15 minutes ago it was white, but [3:04] we won't get into that right now. [3:07] But yeah, so this is another layer of [3:09] complexity [3:11] where [3:12] the invariant changes a little bit and [3:14] instead of yes plus no equals $1, the [3:18] sum of all yes, blue, black, red, I know [3:23] colors, well done me, [3:25] all must equal $1. [3:27] And then as of Polymarket V2, there's [3:30] also an other position and this is [3:33] essentially if [3:35] I didn't turn up to the talk, [3:38] kind of evacuate emergency, [3:41] the event didn't happen, so cancelled [3:43] essentially. [3:47] And then [3:49] the Oracle, another big source, the [3:51] Oracle always has to report [3:54] a legitimate result that sums to a [3:56] dollar. So if they change Uma or Uma [3:59] updates how they handle the results [4:03] always has to stay with this core [4:06] invariant of like [4:08] one. [4:14] And then as a V2 [4:16] you can there are markets that are not [4:19] going to resolve anytime in the near [4:21] future that exist on the Polymarket V1. [4:25] For example, the 2028 [4:27] American presidential election, this [4:29] market already exists. If they want to [4:32] take this market over to Polymarket V2 [4:35] to get all the benefits of the cool new [4:37] stuff they've added [4:39] they need to map the existing market [4:41] over to V2 [4:43] or have a split of the liquidity which [4:46] is not cool. So yeah, and when they [4:50] suddenly have presidential market on V1, [4:54] presidential market on V2, you can [4:56] imagine here [4:58] even more ways that [5:01] yes plus no [5:03] if somehow yes [5:06] person A [5:08] uh could be true on the legacy V1 and [5:12] person B is true on the new Polymarket [5:16] V2 [5:17] suddenly yes plus no or yes plus yes [5:21] is more than $1 and you can quite [5:24] quickly [5:26] drain lots of money and we don't want [5:28] that to happen. [5:30] So yeah, just as a [5:32] kind of sum it up, if you guys want to [5:34] take a look, these are a few of the key [5:36] questions I think you should keep in [5:38] your mind when you're planning things. [5:41] When you are minting tokens or whether [5:44] you're redeeming tokens, is there any [5:46] way this can happen twice? For example, [5:48] in the legacy [5:49] to uh V2 migration I spoke about, is [5:53] there any way I can claim on both? That [5:55] would be terrible. [5:57] And another one would be [5:59] can I have [6:02] the market of what color t-shirt Paul is [6:05] wearing, black. [6:06] Can the condition idea of that somehow [6:09] collide with [6:12] who wins [6:14] the football tomorrow, whatever game? [6:16] If that's the case, then [6:18] again, [6:20] how do you resolve the same question [6:22] twice to the same condition [6:24] problems? [6:25] And then finally, [6:27] similar to the migration question, [6:30] is there any way that yeah, the Oracle [6:34] on V1 and the Oracle on V2 can come to a [6:37] different conclusion? [6:39] Cuz if that happens, we got problems. So [6:41] yeah, if you want to take a look at [6:43] Polymarket, I would definitely [6:44] recommend. It's very cool, and they're [6:46] doing some fun stuff. And I've finished [6:49] under 7 minutes, [6:51] but not under 5. So [6:54] I hope you guys lost a lot of money. [6:56] Okay, thank you.