¶1Okay, so I think I actually found something here. I don't know if I even I should make this video and just use this myself, but uh I guess we can just go through it. So you can see here at 1647, right? Uh we get this exposed and here you can already see the price starts to move, but that is 42 seconds afterwards. So if we were pretty quick put up a position here, we could catch this price r.
¶2So today uh I wanted to share kind of my latest experiment uh on this is time is going to be on polymer market but this works on cali too doesn't really matter and I'm trying to use I think this capes is uh codeex again on GPT6 with Astra and what I wanted to test uh in a more like automated way is to see if we can go from like X posts that yeah runs they come up all the time. We want to search for a specific expose to compare them to an event on let's say poly market or calc and we want to kind of try to compare and turn those exposts into like a numbers you can call it like that so we can kind of compare it to the price of the event right so just to go through the the setup I have on the VPS um we want to run this on a VP VPS because we want it to be 24/7, right? So, I have uh a codeex uh installed on the VPS uh with uh my subscription of course and we have my surf agent. This is basically an an agent that can operate on Chrome to so we can kind of log in on X account. So, we don't have to use the API.
¶3And when we have the surf agent and we run chrome on the VPS, we can navigate on X to find posts based on basically latest search. We can save all this text links and times, right? So what we want to do next when we have all those post on a specific event uh in this video is going to be on like uh I guess I can show you here. So it's going to be this event Saudi oil pipeline restarts by. So this is the 30th of December event.
¶4I just thought that could be a lot of information that comes out on X. So this could be like a nice event. And as you will see here soon, we had a big spike here. I'm going to tell you later in the video what our model picked up. And if this gave us an edge uh on this spike, you can see on 5 cent here that happened.
¶5I think it was last night or something when I was sleeping. So here we had a spike. So we're going to check if our model picked this up and if we had possibilities to enter before this spike happens, right? That is basically the goal. So you can see here we are using AI to turn those expost.
¶6That is of course uh semantics. We try to turn those into numbers. So we check for relevance. Are they relevant to this event, right? We give that a score.
¶7We check if they are yes or no. like are the sentiments is it going to open before 30th of December is it not that is a signal novelty uh because we don't want duplicates so we want to see have we had this before yes or no right and that's what are kind of the tree we are doing in this early testing so we're not going to do any machine learning models or stuff like that we basically just going to plot this evidence signal I call it it's not like a prediction [snorts] and then we're going to compare here so you can kind of see we just look at the evidence on X and we compare that to the price on for example poly market who moved first. So let's say we found that X posts uh the price tends to lag behind the expost. That's what we want to see right then we can try to take advantage of that. But if we start to see that the pricing is ahead of the expost then there's basically nothing there.
¶8So this is the experiment and I think we have some early positive uh indications that we will look at soon. So yeah, hope this was kind of easy to understand. Um it's basically just very fresh. I thought of this yesterday and I kind of lined it up for today. So yeah, let's just see what we have found out so far.
¶9And yeah, there's nothing much else to say about it. Let's just have a look at some of the numbers, check some events, and some posts we have collected so far. But before we do that, let me just show you how this codeex and surf agent kind of works together in this debugger chrome here. So if we open up this, you can see now I opened up the port here with the debugger chrome and you can see I'm logged into an X account here. So in yeah, I'm just going to make this codeex a bit small here.
¶10But basically now we can use codeex to control this. So I can just give it like a yeah, a task here. So I can just do something like check the three latest X posts on the Saudi pipeline also the top post and just do that. And now you can see we instruct surf agent to use Chrome to control um yeah this debugger window. And now it should be able to just use surf agent to navigate X as long as we are logged in right.
¶11So just give it some time here and we you should see surf agent now starting to search. You can see it's searching for Saudi pipeline and yeah this is not me of course and now it's looking at the three post. It's sorted by latest now you can see and hopefully it's going to also do a sort by top and yeah that should be it. This is what we are using for research. So this system you see here now this exact setup is running in the VPS uh in the cloud right and this is just going to run on like a interval so now we are fetching this so let's see if we switch to top and here you see okay so now we are actually reading the three posts we found you can see here we are reading these post and we can uh switch to top perfect and now we see the top one this was two hours ago and Pakistan is starting available options regarding the Saudi pipeline stuff stuff right you get the point but this is general right this this is not custom in this case you can see it's custom for and we get the summary here right in this case it's custom for this event Saudi pipeline but this could be anything right we can track anything on like a 24/7 basis and store all the data and try to turning into like a matt model so let me show you kind of the setup set up I have for this now and where where we are.
¶12This is just a testing phase. I'm not putting any positions on this. This is just early testing for me and hopefully we can find something interesting. So let me just show you kind of the dashboard I built for this and yeah at least one interesting finding. So if we just start here you can see this is the Saudi oil pipeline restarts at 30th December.
¶13So now you can see something has happened here right because if we go to the event now you can see something has happened app I haven't seen this so now we have had a big spike this could be interesting now you're kind of getting this live look at this spike so here something must have happened so I think we're going to use this now to actually just do some research live here because I never seen this big of a spike we went from 49 9 till 70 in like this is what we're looking for right so now we can actually do this live so I'm going to ask let's ask here now you can already kind of see it here on the post here relays a bloomer report claiming half capacity restoration within days so yeah something happened okay so we just had a big spike in the price in the Saudi event so basically uh I want you to analyze the post again and look at the timing of the post uh I guess the last hour or last half hour should be fine. We just had this spike [snorts] and you should analyze check towards the price spike and was there any gap between the reports on X and the actual price spike. Did we actually have a chance to get in based on the expost? So again run the analysis and put of market event three in the in the dashboard. So let's just run that and let's see if we can actually see and the most interesting part was there a gap there.
¶14Could we check X first then go to poly market place a position and then uh then the price would spike. That's the interesting part right. So, let's just do this research. This was perfect timing because now we can kind of get some answers. Even if it doesn't work on this market, it could work in future markets.
¶15So, let's run this research. Let's take a look at it. And yeah, I guess I'll be right back. Okay, so I think I actually found something here. I don't know if I even I should make this video and just use this myself, but uh I guess we can just go through it.
¶16So, you can see here at 1647, right? uh we get this exposed and here you can already see the price starts to move but that is 42 seconds afterwards. So if we were pretty quick put up a position here we could catch this price rise. So if you scroll a bit further down here you can see uh there was a bunch of uh different uh accounts tweeting out around the same minute here and you can see the prices at this time. But uh this can kind of be confirmed now because if you look at the the post here that was at 164713 from Diego Bloomberg.
¶17So he says Saudis seek to resume half of key oil pipeline within days. And this gets kind of restated by this Zero Hedge account. This is a very known account. This is like a good source that kind of says the same, right? Zero Hedge.
¶18He has 3.3 million followers. I've heard about him. So basically I think we actually have something interesting here, right? So let's say we saw this post at 1746 if there was a market. I haven't checked but can we check this here?
¶19Uh maybe there was a market. Yeah, it looks like there were some shares here, right? So that's good. So we could have probably entered at least with the small size and in this window we could have catched going from like let's say 45 cents 47 all the way up to like 85. So that's very good.
¶20So yeah, should I even post this video? I guess I got to just do it anyway now. But this is definitely interesting and this is something I'm going to keep working more on. Uh this is by far the strongest evidence, but we did have one more one I could show you earlier today. Uh this was this one.
¶21We found a post uh that was posted here and a few minutes later, you can see this spike here. This was only five cent, but still uh another point on the data curve for interesting setups like this. So yeah, I'm really positive on this to be honest. And even though my pipeline is not perfect yet, you can see we are still pulling here. And yeah, it's it's far from perfect because we can't catch every single single tweet.
¶22But uh we are working on it trying to uh trying to find the best way to set this up. So if there is like following specific accounts to specific events or just having some other filter or stuff like that, I just got to work out. But so far I got to say this looks pretty interesting at least now that we saw this this change here and you can still see we are at 70 and the liquidity is pretty good. So I think there will be okay here we dropped that 65. So there's some volatility in this market right now.
¶23People are quite unsure what's happening probably but uh that's good. Volatility is good if you want to trade right and the liquidity looks pretty strong. So, I think there will be no issues actually trading this. But again, yeah, super happy how this turned out so far and yeah, looking forward to exploring this more. So, the point of this video not was not just to to make it like a really long video.
¶24I just wanted to share a bit about how I kind of think about trying to run new experiments on these new fight markets like hyperlquid um prediction markets Cali poly market and yeah uh all these kind of new uh markets that I call it and it's such a good setup with like AI and you can have the APIs the websockets and everything and you can build these different pipelines you run it on a VPS you can run this 24/7 you can and automate all of this process, right? That is kind of the point. We could have had like an automatic setup here, read this post. Uh maybe it saw a retreat from Zero Hedge, not a retweet, but a tweet and that was enough signal. We could have automated that, bought in at this.
¶25And yeah, maybe we could have made a good deal there. So yeah, just wanted to share a bit about kind of how I try to find up new ways to think about these markets. Uh this is not going to be the last one. So if you en enjoy this uh feel free to like the video, subscribe and hopefully I see you again when I have experimented more with this update update not update with this strategy and I can have something that is a bit more yeah interesting with some examples where we actually tried the pipe run it for a long time made some trades and entered some position and stuff like that. This is today is basically just theory and testing it in like a shadow environment, right?
¶26So yeah, hope you enjoyed it. Hope this gave you some inspiration to start exploring this. Check out my other videos if you want to support the channel. Follow the links in the description and yeah, I'll see you again soon.