Today, uh, I wanted to do an experiment I've been thinking about for a while. I wanted to kind of compare like apples to apples. What happens if I set like cloud code up with like we're just going to pick the latest biggest model 4.7 versus codec cli with the 5.5. We're probably going to do the I don't know. Let's just do medium thinking. Or maybe we can do high. Let's do high thinking. High. Uh, let's hide. That's fine. We can do both for them. So, basically what we don't want to do is use poly market. We're going to set them up maybe. I think we're going to go with the fiveinut uh bitcoin BTC trades. That's very easy to monitor. So, we're going to give model each model in cloud code and in CLI codec cli. The same prompt, the same documentation. Also they're going to probably have no we will not have any data. So same prompt same documentation and same time of executing the trades. Uh I said like one hour or something it's going to be finally. So let's say they trade for 1 hour and just simple most dollars at the end wins. That's a simple experiment. But I want to take you through how I set this up because I think it could be a bit interesting looking at how we set this up and kind of maybe look at a bit of the different strategies both models take here. So I think we just got to get into it. I loaded up two different Poly Market accounts. So one uh both has around 50 bucks they can trade with. So I'm going to specify that in the prompt. I want to see if we can do them side by side to track the trading in real time. If we can't, we're just going to do one by one, but I want to try to set up side by side. So, let's just start uh building this out. So, how we going to do this? Now, you can see we have the both wallets here. Claw code wallet has 50 bucks. Codeex has 52. So, combined is like 100. That's a perfect starting point. I got to check if Codeex has any gas money. Ooh, that is important. Let me fix that. Okay, so that was done. So I sent over just some matic I think it's called. So now both wallets are ready. So this is my plan now. So I'm going to do cloud code here. Uh cloud poly and we're going to do yolo mode on 5.5 medium, but I changed it to high. Uh we also have high here on Opus 4.7. Uh the prompt I'm going to use is this. So I'm going to read it out. If you want to skip, that's fine. Your task is to create a profitable trading strategy on the poly market 5 minute up and down. You can fetch the documentation. So I put that in here. Just a link to the poly market documentation. And you will need to extract. No, you will need to do extensive research, brainstorming, groing to find a strategy that can make the most dollar in one hour. This is a competitive challenge to your fe fierce rival. So I'm going to say clude code on the other right. So you will need to you will be measured in the amount of plus dollars not how much is in the account after one hour. If you don't make any trades you will automatically lose the challenge. You now have the chance to create a plan launch agents for research. When you launch your algorithm for trading it must be able to run for 1 hour into uninterrupted. If your balance goes to zero you lose. Now do the research. create a plan to beat your rival and to show that you are the 100x giga brain trading AI agent. So, I'm going to send this prompt to both um to both uh we're going to start here since we did um cloud code model. And now let's do open AI open AI 5.5 model and send that over to cloud code. And let's just run both. Okay. So let's do that. Uh should I do plan mode? Maybe we should do plan mode. Yeah, we can do plan mode. That's fine. Uh shift tab. uh plan mode. Uh so both are going to do plan mode. Okay. Okay. So I'm going to take you back when both of these agents uh have done their research, they created their plan, they should be ready to trade. Then we're going to build out the UI, I think that uh can monitor both side by side. I think that's going to be pretty cool. So let's just wait this let this run out and I'll take you back when they have set this up now. Okay. So both uh models now have their plan ready to be implemented. Uh I'm not going to go too much into the different plans now, but we're going to ask about that later. So yeah, I'm just going to accept that on Cloud Code. Same on Codeex. Let them build out uh the setup. So now uh both uh Cloud Code and Codeex has finished up their strategy. So let's read what they say here. The strategy is don't this is codeex open AI don't try to predict Bitcoin generally predict the exact poly market sentiment source faster and more cleanly than the market. Uh up wins if chain link end price is greater or equal to the start price down wins if it's lower. So the bot watches the USD BTC live captures the window start price then asks given where chain link is now how much time is left how volatile BTC is and what's the real probability upwind wins then it's going to do some probability calculations and try to make the trade. So yeah, not the most advanced strategy for codeex, but let's see how it works out for cl code. Uh it's going to be uh let me see the core ID is that in the start of the window, no one knows the outcome. It's a coin flip. Prices sits near 50 cents or 50/50. Uh but with only 6 seconds left, Bitcoin has basically already moved to where it's going to move. The results almost decided yet uh winning side often still sells for less than one. Let's say 80. Oh, so it's going to pick that boring strategy. Okay, that's fine, I guess. So, it's going to try to buy in very late in the window just to secure a win. But let's see. I'm not going to intervene in one sentence. Wait until the five minute win commits nearly settled, then buy and just try to ride that streak of wins, I guess. But uh like I said, I have not intervened at all. So, we're just going to see what turns out best in the end. But before we do that, uh I want to build a UI for each bot so we can kind of set them up side by side so we can run it over the hour. So, let me just put in this um build the UI. So, this is going to be for codeex. So, we're just going to build like a simple UI for this. I'm going to put that in codeex. And for what did I call it? Claude PNG. Uh I have a logo here. Claude PNG. Uh let's do that in uh CL code. So let's build out the both UIs and then we are basically ready to go. So my plan is just to start both kind of at the same time. One thing uh cloud code in the testing phase kind of messed up. Okay. Okay, so I guess it's back again now. So it actually made $1. It had a bit of a luck there because it made a bad trade, but I guess it resolved it. So now it has 51 instead of 50, but that's doesn't really matter too much. So after the UI is ready, let's start both at the same time and we're just going to monitor and see who ends up with the most amount of dollar or probably the least amount uh in one hour. Maybe someone makes some money. We'll see. So that's going to be kind of the king of polyarket 5minut up and down trading. Okay. So I think the UIs look great. They are a bit similar but you can see we have the clawed UI on the left here. Uh I I made them quite similar so they kind of fit into the window. We have the open AI here on the right. Time remaining both have this P&L this hour realized. We have the balance update and there's a few different things but mostly the same as intended. So, I think we are ready now. So, I'm just going to say go here in both windows and hopefully this will start the timer and let's just see. So, the rules is if none of the bots or one of the bots doesn't start, of course, I'm going to fix that and I'm going to kind of monitor in the background just to keep them running and if there are any issues, I'm going to try to restart them. Other than that, I'm not going to intervene uh for you. This is going to go go quite fast. I'm just going to kind of speed it up so you can just watch them uh go back and forward in like 20 seconds or something. But uh now let's just do it. So I'm going to say go and go and let's go to the dashboard and let's see if they kick off here. Uh if not uh we're going to do some changes but let's just wait and see. Okay, so you can see uh Bitcoin bots bitcoin claw bot started. What about open AI but here? What is this start? Okay, here it goes. So, now we have both up and running. That's good. So, yeah, like I said, I'm going to turn off the camera. I'm going to capture this and if there are any issues, I might even come back, but let's just let run it out 1 hour. Who is going to beat the other model going into this? So, they start on 51 versus 50. And yeah, I guess good luck to both models. my phone. Everybody got your touch from last night. Okay, so that was done. So, I don't know how it well showed in the speed up, but that was kind of messy actually because uh yeah, we had some issues and it was a bit optimistic going straight into the trade. But, uh yeah, you can see it here. One bot did really good. The open AI got like $14 in profit while Clouds started out pretty safe. in like 30 minutes. I think it had made like 40 cents and it was like $10 behind OpenAI. So, I think I said something like uh let me see if I can find it here. Yeah, I said Claude is losing by leading by $10. You're just making pennies. Adjust or lose. So it did actually change up its strategy and after that it just went full out all gamble and it did put like here it lost like $28 in one trade. So it bought 37 down uh on for 28 and lost everything. So it just went into hardcore gamble mode after I told it that and it basically the balance is not 14 I think. Let's double check the balances. So, this is the balance of the Open AI account. Uh, it actually, yeah, did very good actually, this trade. Uh, this strategy, I'm going to might look more into that in a different video, but this uh did pretty good. So, let's switch to the other account. So, it was not 14, but it lost like $25. So, we did up end up minus 10. But uh if we had just let the the Clawbot just run the penny trades, we would have been like $15 up, maybe a bit more. So it could have been a good result, but since I kind of pushed it a bit at the end because it was losing the challenge and it didn't seem to try to win even. Uh I just had to and then just went crazy. But basically uh the open AI, like I said, did pretty well. I was quite happy with the way it uh performed. So, this strategy is definitely something we're going to look more into in future videos. So, without any doubt, the winner was, of course, Codeex 5.5 on high setting. And yeah, I had a lot of fun with this actually. So, we could take like a quick look before we end this at the strategy it did actually pick just to hone it down a bit more. So I asked Claude to explain the this logic strategy in one sentence. It said it uses live Bitcoin price to calculate the true odds of each 5minute market and only place small bets when Poly Market is selling those odds at a discount. Pure value betting against a mispriced book. Is that 100% true? Probably not. But uh in this case it did work very well and we ended up with a nice profit here actually. Uh so yeah pretty interesting and if people like this format we could do other models too. Uh we can do open source models. We can try that on the GGX over here. If you wanted to test out longer strategies we can do that and other more constrained strategies. So I will check the like count. if it's quite high, we can do more of this. But uh yeah, that's what I wanted to share today. Just a fun experiment and also good data points to take with me if I want to dive deeper into this and set up different strategies that could be more interesting to pin up against each other and see who survives. So if you want to talk more about AI automation, AI trading, anything like that, uh you can find the link in the description to my discord. I think we have passed like 200 people now in the server and yeah, if you want to come talk, feel free to join. So, thank you for tuning in. Hope you enjoyed it. And don't forget, you can see the white rabbit sign up in the corner here. This means that this video is a puzzle hunt. So, there could be like an interesting prize for you, a reward if you figure out the puzzle. So yeah, thank you for tuning in and I'll see you again hopefully very