¶1Okay, so today we are back with our second week of the AI predictions battle on Polyarhead. So as last week we are putting up OpenAI's uh GPT 5.6 versus Claude's Opus 5. Uh on this time we're going to do three different events. So you can see there's going to be like a culture event like a weather event in London and an an employment rate event uh in politics I guess. So you can see from last week you can see Claude is up uh 1 nil here over OpenAI.
¶2Uh it did call a weather uh prediction correct. Uh OpenAI missed both and Claude got the one point here. So it's going to be exciting to see where we end up today. So the events we're going to do is Ariana Grande petal first week album sales. So we're going to let the models try to predict that.
¶3Uh we're going to look at the July unemployment rate and we're also going to look at the highest temperature in London tomorrow, I guess. So these are the markets we going to test the models on. And if you watch the previous video, we have a scaffolding for this. So this means that the models won't see any of the the prices or stuff like that. I guess I can show you like what the models will see when they try to predict this.
¶4So for example for question one here uh the debut week activity total for the album petal by Ariana Grande right uh we have the resolve uh issues here so these kind of the rules it's going to follow and this is what it see so it has to pick uh between here so from 100k and below and up to 600k and below so it doesn't get the prices in the scaffolding and that is kind of on purpose because we don't want uh the poly market prices here to kind of contaminate what we see, right? So, uh we're just going to get started. You can see the 200 to 250 to 300K is kind of the favorite. So, I think we're just going to start on uh codecs uh with GPT 5.6 sol. So, I think I'm going to set both models to high reasoning.
¶5Simple prompt. Uh we're just going to read the instructions to execute the benchmark or the predictions and execute your predictions for this week. That is all I'm going to give it. And we need to run one and one. So we're just going to run the GPT first.
¶6We're going to note down the predictions. Then we're going to clean uh all from the memory. Then we're going to run cloth. And yeah, then we should have both predictions. And I guess we could bet like a tener or something on each.
¶7Something around that. We will see how they converge or correlate. We just have to wait and see. So, I'm just going to let um claw uh no open AAI do its predictions here and we can take a look at the results. Okay, so we have the results in uh I don't know how you well you can see this but basically uh we have the 275 to 300K.
¶8Where is that going to land? If you look at Arion Anna Grande 275 to 300K, that's going to be in the favorites. So that's in the highest percentage. Okay, that's fine. Uh for the second prediction, 4.2% that is going to be the probably the unemployment rate.
¶9Uh that is not the favorite. So that's interesting. So we're going to take uh 4.2 uh yes, I guess on that. And the final prediction is peak London 23° Celsius. Uh that is the favorite.
¶10Okay. So 23° favorite uh Celsius that is also the favorite. So we got one that follows the favorite. This does not and this does. So those are the three predictions from GPT 5.6.
¶11So now I'm just going to instruct um Claude here to just remove all traces from the OpenAI prediction and we can launch this on Claude Opus 5 here. Okay, so we kind of cleaned up everything from the OpenAI run. Now let's just launch our Opus 5. Uh let's check what effort we are on. Let's set the effort to high and let's just do the same prompt and let's yeah let's just do it.
¶12So while we wait for that let's just place some here. So for the highest temperature we're going to go for 23. So let's just place 45 cents a share. Let's just buy this for Open AI or at least just place the order. And for the July unemployment rate, let's do 4.2.
¶13Yes. uh 29 uh cents per share. Let's do 24 shares. Let's place that order. And for the Ariana Grande, we're going to do Yes.
¶14Uh I guess we can just pick like something like 71. That's fine. Uh no, 71. And let's do 14 shares of 71. And let's just place that order.
¶15Uh let's see if we fill after. So that is done for OpenAI and now let's just wait for the research from Claude and we can compare them. Okay, so we had the Opus results back and you can see we pretty much landed on the same. That's not a big surprise, but we did actually land on a different option on the Ariana Grande debut week. So here we're going to go a bit higher and on the July we landed slightly on 4.2.
¶162 uh just ahead of 4.3 and kind of same here on the London uh temperature uh we landed on 23. So I'm not going to place any orders of those. We we already have them but of course I'm going to keep them but we are going to place on the 300 to 325. That's going to be in this right. So let's just do 32 and let's just do something like 30 shares.
¶17place that and let's see if we actually filled here now. Um, let's see. Um, yeah, we filled that. And I guess we filled this, too. So, we should be pretty good to go.
¶18So, that means that we have all the bets placed. And now, let's just wait and see what will happen. So other than that, I have been just letting my um automated system just run. You can see we have a bunch of different options here. Uh this is a new option I have running that kind of went very well tonight or today.
¶19We bet on the S&P 500 to open up and we made a nice 50% win on this. This is a different system I have running in the background. But uh basically like I said uh before this round week one Claude was one nil up. Uh now it's just going to be decisive what happens in this Ariana Grande bet. All other bets went kind of the same way.
¶20So what will decide who wins is where this lands. Uh if it lands outside of this then of course yeah no one wins. But uh let's just wait and see. So, that was going to be the week two predictions uh prediction battle between uh Opus and GPT 5.6. And yeah, have some other stuff in the works.
¶21If you like this kind of content, I'm going to come back uh later this week with a bit of a different model, more like a tutorial machine learning model you can use on Poly Market to try to get a bit of an edge uh or at least learn how to build simple machine learning models to do some prediction. So, if you like this kind of video, give it a like, share. Uh, you don't have to share, but at least you can subscribe. So, hopefully I'll see you again later this week. Enjoy your day and we speak soon.