How Kimi K3 Crushed Polymarket with a 6,871% Return

[00:02] by the hype around Kim K3. So yesterday I thought I just give it a shot. So I added some credits to open code and I put down Kim K3 as our provider. So the

[00:15] test I wanted to do yesterday was to point Kim K3 using open code on some data on Poly Market around a batch I was going to watch. That was the Spain. No,

[00:27] not the Spain. That's today. That was the France England game in the World Cup the France England game in the World Cup yesterday. So what I wanted Kimik K3 to do was to analyze all the data, the market data we had on um this match. Uh

[00:41] this was before the match, right? And we wanted to look for some expected value. What was the best prices we could find using the data we had? And I wanted to using the data we had? And I wanted to try Kim K3 on that task. And as you can

[00:55] see here, we did very well, right? You can see uh yeah, I might zoom in this, but we have one trade that uh earned 6,800% 6,800% gains and the other one was like 613%

[01:09] gains and the other one was like 613% gains. So, that went really well. And uh I actually sold out half time on one of those. And if you look at the other one, Bitcoin trading going on there. Hasn't been doing really well. And yeah, here

[01:25] is the other one. This is the exact uh score. So if one of the teams had over four goals, I'm going to go into this. Uh this went in. So we got $324 on that one. So that was a really positive uh uh first impression of Kim

[01:42] K3. So today I thought I'd go through what we did and we're going to do the same thing for the World Cup final that goes today. So we're going to see if we can make some find some interesting expected value bets on poly market by

[01:57] analyzing that data too. So before we run today's analyze uh an analysis uh I want to show you last night uh you can see this uh the returns we have here. So see this uh the returns we have here. So I just want to go through uh what kind

[02:11] of how Kimmy kind of came to the conclusion that they wanted to place these bets. So you can see uh the first point here is we don't predict we find inconsistencies. So this is kind of we use math and code

[02:25] So this is kind of we use math and code to look at the data to find uh some kind to look at the data to find uh some kind of bets or lines that are probably price you can see poly market is a market prices are set who ever has the best

[02:39] information fastest mo most betters trade vibes. Our edge was never knowing England would win 4-nil or lead 4-0il to halftime. Nobody knew that. The edge was finding prices that were mathematically inconsistent with each other and buying

[02:55] the cheap side. So that was the strategy uh Kimik K3 came to the conclusion of following. Uh I also want to mention we spent actually $8 on this. So we have to you can read it here if you want to see kind of the the start it had here. So

[03:12] basically it came up with a model it wanted to follow that was reverse engineer the lambdas from market prices and by doing so it made kind of a discovery it wanted to follow. So you can see here we kind of did this

[03:25] calculation on the numbers we extracted from poly data and the model said that we found a gap in kind of based on our calculations and the market shares. Right? So basically we uh gave kind of the fair value or our calculations said

[03:41] that the price was supposed to be like 19 cents per share but the market offer 14 and that was really good way to just enter this trade based on our calculations or I guess Kimy's calculations. So this was the first

[03:56] trade we did because we found this gap in uh based on our math and the market price. That was pretty interesting. And you can also see uh Kimmy kind of took it a fee into consideration here. And this was kind of the gap we found that

[04:10] was 4 cents per share from our calculated fair price or correct price. calculated fair price or correct price. So basically Kim just looked to find a mispricing you can say that in the market uh compared to its uh fair value

[04:26] calculations and this is the trade we did and basically we did the same with some other things we found uh England minus 3.5 was the one we filled

[04:39] minus 3.5 was the one we filled uh yeah we tried to do minus 4.5 but uh we couldn't fill that I guess we were some part we were up like 24,000% but we couldn't really fill that so it didn't really matter. Uh this was a loss

[04:53] because this was France right and France here was a loss. Uh the cost was $65 total and the estimated expected value was $29. was $29. So that was basically how Kim K three

[05:08] came to this conclusion. And now I kind of want to go in. Yeah, I had some other lessons and stuff here if you wanted to check this. But now I want you wanted to check this. But now I want to focus this strategy on today's game

[05:22] and that is going to be the Spain versus Argentina game today. So I think I just want to keep it the way we did and I'm going to point Kim K3 to this game now

[05:34] and we're going to see if we can find similar mispricings and make some bets on this. Okay, so I did uh instruct Kimmy to construct a prompt to use to check for the same expected value positions we can use for

[05:47] future games. So if you want to kind of copy this, you can just uh I'm going to scroll to this slowly. So you can kind of read through this. We have some methods here and we have the rules. Okay. Uh do not place orders until I

[06:01] prove the table. So basically want me to also draft a version two with a in plane mode. So we can also run this live. So, we have like a monitor on this, but uh now we're going to keep it quite simple. Uh I'm just going to point Kim K2 to

[06:15] today's World Cup final event. Should be great liquidity and stuff like that. So, we're going to see if we can find any similar bets or uh like yeah, very similar or the same to see what we can place on in today's game. So, I'm just

[06:31] instructions. There is a new event today, the World Cup final, Spain versus Argentina. Should be great liquidity. So let's take a look at for similar plus expected value positions for tonight's game. Use the same data math codebased

[06:46] game. Use the same data math codebased approach as yesterday and work hard. So I'm just going to let Kim K3 now go through all the data. So uh I did not show how I set up this but basically how we connected our open code Kimik tree to

[07:03] we connected our open code Kimik tree to the API on Poly Market is really simple. I guess I can just give you like a quick instruction. So on poly market here we have something. I guess I can just open here you have this documentation. So

[07:16] here you have this documentation. So here you can use uh API for free to gather data to run your um your trades, your betting everything. And yeah, you can just scroll through this. And they also have a web socket if you want to do

[07:30] more low latency and stuff like that. So, this is basically what we use to do these type of setups programmatically. So, I'm just going to let Kimmy K3 run

[07:42] here now and I'm going to take you back when we have something if we have something for tonight's game. Okay. So, you can see we have some results back here. You can see we spent about 150 more. So, the final recommended

[07:55] portfolio for the Spain Argentina World Cup final is the following. So we have six or maybe it's just five because we're going to skip that. So we have any other score. This is a four plus goal score line I think for a World Cup

[08:09] final. That's pretty crazy. But it did find some uh net expected value here. find some uh net expected value here. Argentina the under or minus5 Argentina the under or minus5 25 35 very speculative. Spain 3-0. I

[08:23] think this is pretty interesting. uh that is of course with the lowest expected value but uh this is what we found and I checked all markets using Kim K3 for this. So now I'm just going to instruct

[08:37] um execute those uh trades for me since we have programmatically connected uh poly market to um

[08:50] to uh Kim K3 here and open code. These trades can now be executed programmatically. And as I mentioned we could also set up like a live monitor here. You can see this execution plan by the five legs now line monitor fix. So

[09:07] we can run this on like live with some guard rails on halftime protocol. We can kind of check the halftime and yeah we can say it doesn't really matter but basically like I said we could set up like a web socket to run this live

[09:20] during the match. Maybe we can set it up to find some find some of these expected value mispricings during the match too. But uh let's just see now. Uh okay, so

[09:32] we have we're in plan mode. So I'm just going to change that. Okay, so I just switched to build mode. Uh I didn't see that. So execute those trades for me. And now uh Kim Kater should just do this automatically. So let's just open up our

[09:45] poly market here. And you can see Kimmy is writing the code to execute this. And we should see those pop up here in kind of our open positions first because we need to fill those orders. Uh, some of them might

[09:59] have the liquidity to be filled straight away, but let's just see what happens. Some might get into positions, some might get into open. Let's see. Uh, we haven't seen anything yet. Yeah, you can see we have the spread Argentina minus

[10:14] one minus2 exact score 3-0. We have the minus3 spread. So, looks like everything minus3 spread. So, looks like everything uh went on. Uh, yeah. And this too. Good. So, I want to do one thing. I know this

[10:29] yesterday maybe doesn't really matter too much um in this game, but uh we could have been a bit more ambitious here, but uh that's what we're going to do. Uh I want to mention one more thing to Kimmy.

[10:43] Uh yeah, that looks good. Everything is filled. So, now we are ready for tonight's game. But what's interesting now, these prices is going to change during the day. Now you can see already this have lost some value.

[10:57] So we need to monitor that. But I want to say one more thing. Yesterday I noticed that when the lineups came out there was some uh volatility. So, I want to set Kimmy on one last task that is looking at some

[11:15] one last task that is looking at some volatility around the lineup coming out volatility around the lineup coming out and if there's anything we can look into to make like a quick dollar or something on that um on that uh signal, I guess.

[11:30] So, I'm just going to do yesterday we saw volatility around the lineups coming out. investigate and looked for any possible expected value positions. Very short term of course on the lineup news. Just want to see what happens. I haven't

[11:44] tried this before. Also, uh I waited a bit for this and now you can see the prices have changed a bit here. You can see we are $1 up on the exact score and yeah, some small changes around this. So the prices on polarity they always

[12:00] fluctuate. So yeah, we could have made going to keep it for resolution, I think. Or maybe not resolution, but I'm just going to keep it. So uh yeah, I'm

[12:12] going to take you back when we have the if we can do anything around the lineup news uh later today. Okay, so we found some interesting things here. You can see uh it has some scenarios that is going to pre-map. So if Lionel Messi is

[12:29] out, this is going to be like a smash Spain yes move. And if Yamal or Rodri is out, this is also going to be interesting for the Argentina uh play.

[12:41] And yeah, but most likely it's going to be full strength lineups. So it's not going to be anything, but I asked what is kind of the the the technical setup you have here. So I just wanted to double check that.

[12:55] So the first setup is one bot that is just basically it's going to monitor inplay dislocation hunter. So this is going to look for uh it's going to reload all the books and basically look for any orders that uh shows a net or

[13:09] positive edge again. And the second one is going to be the lineup watcher. So this is going to record uh prices and the rumor drift curve whatever that is. and it's going to pull on ESPN for lineup publications. Okay, but that is

[13:27] going to move so fast if like messes out. I don't think we kind of have the latency setup to actually capitalize on that with this setup we have now. But it's just interesting to catch the data on that. So yeah, that is basically what

[13:42] we found out. It was pretty interesting, but uh I don't think we're going to act on this setup, but uh pretty interesting to see how it works during the day. Might do a follow-up video on it. Uh how how this is doing. So, I'm just going to

[13:57] remind you this is the setup we're going to run today. And yeah, that was basically uh the video I have done on Kimmy Trade 34. I got to say, I was super impressed. This is such a good model for being like

[14:14] This is such a good model for being like uh open weights. Uh it's a huge model. locally, but it's open weights and that means that it's possible to run it locally if you have the hardware. And this means that it's more competitive on

[14:29] pricing and stuff like when it comes to open router and it could get cheaper in uh the future. So really cool to see a model like this come out and yeah it's just up there now. I don't think there's like six months difference anymore

[14:44] between the uh closed source providers, the biggest providers and the open source at this moment in time. So that's pretty cool. Pretty impressed by Kimmy. Let's see how it goes tonight. Might do a follow-up video if uh that is needed.

[14:58] If we do well maybe we'll see. So yeah uh hope you enjoy this. Hope this gave Kimmy and stuff like that. If you like this video, give it a like, subscribe, and I'll come back next week for more stuff like this.