Nice looking new prediction tool from the HLTV devs! 💪 hltv.org/valve-ranking/teams… It will obviously, with HLTVs reach, fill the gap that I have been trying to help filling here on X, with the major race predictions. So long term, when HLTV gets the bugs in this tool ironed out, there is no point in continuing doing simulations when HLTV delivers such great tools for the community. Also they will understandably always be considered the authority by nature, so I'd rather spent my effort pointing out potential bugs/errors in HLTVs tool than keep posting competing predictions after this major race. I just hope they are committed to maintaining all the data going into making a serious prediction. At the heart of it, it is the validity of "what the machine spits out" that is important. You may backtest your model on previous seasons and find it to be very accurate, but that is a test with perfect information about event schedule and depending on how you test it, it is even a test with perfect information about which teams will play what events. The accuracy of a forward looking prediction rests upon the hard work of maintaining information about what events teams have signed up for which are not yet registered on HLTV. I hope resources are set aside at HLTV to doing that hard work of making the prediction a serious one. It would be a shame to make a great tool but present misleading data in it. I personally believe that Major predictions are in fact anchored best on HLTV, especially since they provide the data for Valve, so I am happy to see them make this tool, let there be no doubt, it will also free up a lot of time for me 😊👍 The page says it is based on 500 simulated seasons, which sounds like way too little though, but I am not a math professor myself and I am myself relying on what the AI's of the world tell me about how many simulations are needed. They tell me you have to simulate way more than 500 seasons. Think of it like this if you play out EPL only 500 times, 1win might randomly win in 0, 2 and 4 times if you repeat the 500 simulations 3 times. You would then get quite a different results for the more unlikely outcomes in events which can totally skew the prediction.. hmm.. I'm no expert, but it sounds like too few simulations from the description. I have a million other concerns about the correctness of the data shown.. (how would the chances of teams move as much from yesterday as shown, with yesterday being so quiet??? Alliance up 17%, 3DMAX down 39%, etc.. looks like a clear bug) .. but it is likely bugs that can be ironed out over time, you have to start somewhere. The only problem is that there are probably only a few people that can verify that this data is even correct and if no one objects it will just be taken as gospel and maybe even acted upon by teams.. that's my only concern, but it's also not my problem of course. This will be used by teams to make decisions with quite big economic impact for them and the disclaimers that the devs behind this tool would give the teams as to the validity of these predictions and calculations, if they were asked, will of course never surface. This is a great tool for the fans, but teams should be a little bit careful how they interpret the data it spits out in my opinion. Just like they should with my predictions and Mischiefs predictions as we will both tell any team asking. I don't think a hotline for the teams as to how to interpret this data and what exactly it is based on, is a service HLTV will be offering along with this tool :)