Это видео недоступно.
Сожалеем об этом.
Polarity ATI July 5th Auto Arm
HTML-код
- Опубликовано: 4 июл 2024
- July 5th afternoon session into close.
Auto Arm Regression mode.
Chicago time cst
Polarity ATI - ATM Template +16 / -16 stop trigger +11 to -8. 1:1 RR
Snails pace into close today.
Will start working on Polarity ES soon. Haven't spent much time on it but accumulating data to start the process.
If you like the channel content, subscribe and give the video a like please.
Thanks for watching.
Thanks Brett, your comment on understanding how Catboost works is a much-needed nudge in the right direction for comprehending the best approach to injecting quality data into the model and the methodology behind it. Keep the videos coming. Your efforts are much appreciated
Very welcome!
Hey brett, so followed all the steps you showed, created a master incremental file and dumped everything into the ML file after deleting everything from there and then hitting train the model . First test run was decent 14 wins and 8 losses. I can do better by tweaking some settings. We'll see... Thank you for showing this technique.
stay with it. tweak based on the trading day values. ninjascript output will lead the way
Well done Brett R! It's really wrong that PolarityATI does not have adequate technical documentation and requires users to try to figure it out on their own as if it was a puzzle to solve instead of a trading tool.
we all bought a model and a starter kit engine for $500 it's up to us to customize that engine as to how well you want it to run. No smart business person is going to sell Connors NQ or ES model raking $2k - $10k pots like that on a daily basis for $500. His models are worth 6 figures fine tuned the way they run.
There is a lot of work under the hoods of his personal trading models. He should be congratulated for his willingness to get someone started with the tools necessary to build a winning model to their liking and trading style and get to where he is. Unfort, most try a little fail, and give up. some try alot and fail, because they don't grasp the concept of how catboost works.
The settings for fine tuning engines are sensitive. why we build and test test test. take what we learned that showed promise, use on next and tweak more settings.
@@kb_rod I'd share your perspective if the Futures Analytica products were disclosed and marketed as a DIY starter-project with obsolete, incomplete, and inadequate documentation and instruction, but they weren't. That's deceptive by omission. After a purchase, I have absolutely no problem doing whatever extensive work is necessary as long as I know beforehand what will be required and that there will be full documentation and instruction sufficient to allow fully-informed accomplishment of the work. It's a normal buyer's expectation, especially with technical products, that there will be documentation and support to fully explain product functionality and enable the buyer's proper operation of the product. That is an implied warranty. With Futures Analytica products, however, the documentation fails and the support fails, and that's why Brett R and everybody else is spending countless hours discovering functionality, creating their own documentation and knowledge base, and establishing best practices and procedures. All of which information should already be in Futures Analytica documentation and instructional resources, from which users could proceed with a solid base of product understanding. Thanks again to Brett R for sharing your discoveries.
My reply to kb_rod was deleted. Why?
no reply's have been deleted from any of my videos. I see your replies on my end. maybe browser cache needs dumping
@@BrettR-Trades It's not browser cache. I checked RUclips documentation about removed comments, and it seems that when comments are sorted in Top Comments mode, RUclips removes comments that "RUclips detects viewers are not likely to value or interact with". Not so when comments are sorted in Newest First mode. So that's what happened. The RUclips algo's judgment prevails over ours and silently censors.
Brett, my friend I have the polarity, I send emails ti Connors asking questions but they don’t answer to much.
To train the model in playback, I have to do it in 1x (speed) or I can do it in 10x for example?
I have been trying with NQ and ES with Connors Model but still don’t make it, loose all the time, I dont know what Settings are you using.
ES settings on the new PrePack models, I used Connors settings and that didn't go well. With that said, those models need some machine learn time in current market conditions. NQ settings / parameters information is given through several of the posted videos on my channel. Paramaeters can change daily based on current market conditions. Ninjascript output can help with those adjustments on the fly.
😂 Looks like Ron is re- emerging. These guys never stops to fascinate. Honesty is a very RARE commodity nowadays. Really unfortunate. Be carefully folks.
glass isn't half empty for everybody.
Hi Brett, What you have tried and tested with ur time is of immense help to so many who are trying polarity. It would be of immense help to ppl like us who have no technical knowledge if you could pls upload a video on how to go about deleting those files and the other technical terms you have mentioned in video . Thanks
I want you to tune into Connor's live session this morning on Futures Analytica YT scheduled next hour or so. There is a questionnaire you can fill out now that he will receive when he goes live. Ask these questions to him about making a new video showing the process. He answers the questionnaire questions while he is live.
Thank you for publishing this video
Thank you for stopping by
One more question, in your strategy settings, do you use static or the Min/Max values?
Min Max boundaries. and fyi Min Imb floor boundary on Imb Volume does not work. FA team is well aware and have been since R9B version. It has yet to be fixed.
@@BrettR-Trades Thank you
Hello Brett.. Can you upload a video on how to do it.. i am unable figure out from quite sometime..
I want you to tune into Connor's live session this morning on Futures Analytica YT scheduled next hour or so. There is a questionnaire you can fill out now that he will receive when he goes live. Ask these questions to him about making a new video showing the process. He answers the questionnaire questions while he is live.
Hello Brett, I am trying to figure out from a long time to get this working, it would be of a great help if you could upload a video.
I want you to tune into Connor's live session this morning on Futures Analytica YT scheduled next hour or so. There is a questionnaire you can fill out now that he will receive when he goes live. Ask these questions to him about making a new video showing the process. He answers the questionnaire questions while he is live.
Very nice, Brett! Are you using a VPS? Thanks!
no vps.
@@BrettR-Trades Even more impressive!
What IMB VOL, IMB Ratio and ADV DET have you got ?
ninjascript output window I'm showing in bottom left corner of some of the videos should show you the way.
Hi Brett, I resolved an issue I mentioned last week, in which optimization for both trend and regression was being stored in the 'trend' incremental learning file. As it turns out, this was due to my own stupidity. I figured it out and was able to get both models to train and to optimize. The training run was on only one week's data (last week in June) which is short but avoids 19 June as well as 4 July. I optimized on 1, 2 and 5 July, skipping 3 and 4 July. This coming week I'll continue to optimize both models using Market Replay, watching as I do to see what kind of P/L shows up. I'll be interesting to see whether the models do in fact optimize toward 'best fit' values....
@@kb_rod Thanks for that info. Do you use the data file for training? I ask because this may be where I've been going wrong: i've been using 'chart data' to do my training--which includes 24 x 7 data for times when it's very quiet--rather than optimization data. Personally, I think using optimization data as the basis for training a new model is extremely clever!
@@kb_rod This answered my question, so deleted it. Thanks
hi Brett.. when you say you keep good data and delete the bad ones and make sure bad ones dont go to catboost, how do you know which is good and bad? is it based on winrate and are all the ones with zero's are bad data and we need to delete it and not let it go to catboost?
winrate, min imb volume, Min Imb Ratio, Min adv detection
@@BrettR-Trades so are the win rates zero bad ones and needs to be deleted?
personally, I would. If you have Pol_ATI checked ON when you collect the data to train a new model, you will see a landslide of zero's. leave that Pol ATI off. With it ON, 90%+ will be zeros
@@BrettR-Trades ok.. Thank you
Hey Brett, any chance of you taking it on live? or do you feel it is not stable enough. thanks for the transparency
zero i don't trade NQ market. I dipped my toes in that water several years ago and lost a couple of toes. lol. ES ZB ZN more my style. Occasional GC CL when I am swinging a long one on 240 min and daily chart.
I'm researching this software for my son in his late 20's and a couple of his buddies that found Connor and Ron on YT and asked me to take a look. I was intrigued. I have time because I am in front of screens in markets all day long. They are working and don't have time to mess with it. They are interested in the Auto Arm part of it. They asked me to check it out and see if it's legit. I am still in testing phase and have been for several months. I'm intrigued by catboost.ai. At least I am learning something new.
Appreciate your openness Brett.
@@BrettR-Trades It would be great if this works on Treasuries. Have you given it a try?
Hi Brett, I'm lost in this whole thing, I bought the full version last week but can not import the PolarityML zip file into NT8 so I can not load the training models, I wrote to support and they sent me information I already have, so not sure maybe you can help me with it I will send you a private email. i know you are also a customer but you seem to understand the software more than most of us
check your version of NT installed. If it is the latest release, I have heard others having issues installing. What kind of error are you getting?
Do you have incremental learning checked? After the new update Heuristic Parameter wont appear anymore, any reason why? I deleted the previous version then installed the new one
Heuristic should be lighting up if you trained both models correctly. Incremental learn is for collecting ML data after the model is built. You can choose at that point what Incremental learn data you want making it's way into your catboost engine. Incremental learn checked ON is not needed to make Heuristic ML light up. Enabling Machine Learning and optimization is.
Also make sure you have a trained model in each section for Active for Training or Trend Model and in Active for Regression Model. Otherwise the Heuristic Parameter will not show up. You can use the same model in both if you are just testing.
hi brett there are people like me who arent technically sound and trying from a long time to get this working, it would be of great help if you could upload a video of how to do it
Hi Brett, It would be of immense help to ppl like us who have no technical knowledge if you could pls upload a video on how to go about deleting those files and the other technical terms you have mentioned in video . Thanks
I want you to tune into Connor's live session this morning on Futures Analytica YT scheduled next hour or so. There is a questionnaire you can fill out now that he will receive when he goes live. Ask these questions to him about making a new video showing the process. He answers the questionnaire questions while he is live.
@@BrettR-Trades I have tried asking him many times but he never responds which are technical or show his settings.. Not once
I’ve had months of failer and frustration. Can’t figure this out from training models. You videos and information are awesome thank you. Is it even worth trading the prepackaged model in the final folder? I haven’t found success with it yet
I am currently using a new out of the box version of the prepack to build ML database Master file to build my own ES models. I am using their prepack settings stack 3, to gather full cash session (daily). It takes plenty of trades to gather the traded data I want to build models with. But it's time consuming gathering the data for each day in trend and regression mode. I have 6 days of data complete so far. I am gathering from june 17 thru july 5. then I'll build the ES models and put them into ML testing.
I would rather build my own model with my own settings, data input, etc to 'know' what went under the hood to drive the engine. I don't know what went into Connors build or how much ML went into it. I do know that I'm not a fan of 30 minute windows to gather data. .65% winrate, or target, stops of +10 -10 when when i'm going to use it for Hyperscalping 5 ticks on a 1:1 RR.
with that said, I did trade with it in ML mode for a few more hours after the original trade session. I was able to show a small profit in the 2 sessions I ran it following the trend / regression rules. It was a little over 50% winrate. I am sure over time, ML would make it a viable model to trade. But I would rather spend that time building my own models.
@@BrettR-Trades thanks so much for the information Brett I appreciate you and your time to explain. I’m going to try an attempt to make my own model, when you train your model do you turn on your autoarm for it to learn to take trades or your just running it in trend/regression to collect market data and after the data has been collected you then optimize it by arming your autoarm.
Hope your having a good day! Thanks
1st half of each session in arm both. that's the time consuming part. 2nd half of same day in auto arm. all on same file. each day gets ran arm both and auto arm in trend. remove that file to your congregated folder you are building for trend. then rinse repeat same day on regression. it takes time.
I don't know if its the right way, wrong way, for sure the long way. but I know i get good results doing it this way. trial and error is how we learn.
and I do NOT cleanse the daily files. I did in R9B version but it's not necessary now. I use all the gathered ML data. Connor and his team fixed that ML gathering data in this last Polarity update. It's a major improvement. The biggest one in my opinion. I wish they would have adjusted the MLTrain process the same way so I wouldn't have to be jumping through these hoops to get good data to train new models.
@@BrettR-Trades oh ya that’s time consuming but if it’s giving you good results long way yes but a good way since it’s working. I’ll give it a shot and see if I can figure it out along the way. Thanks so much for the info. So many of us are struggling but thank you for your insight. Awesome!
Around 1:26:13, you outline your traIning regime. Just so I'm clear and don't make more silly errors, does the master Incremental file contain the outputs of MR optimization runs for the date range of17 June-2July, aggregated into one file? And does this mean that you are using the optimization outputs as the basis for training (presumably WITHOUT using the 'overwrite training feature)?
Thanks! I'm slowly beginning to understand how PATI works (I think)....
the aggregated incremental files I accumulate daily are all piled into one main file. All Machine Learn info goes into that file. I don't cleanse / cherry pick data out of it. The last Polarity update, Connor's team did a nice job cleaning up what Machine Learn collects into the Incremental file now.
The purpose of dumping everything out of the MLTrain file is so that file info does not get over 'Written' to the catboost engine, you insert your collected Incremental file data into the MLTrain file and let that better data get over 'Written' into the catboost engine. when you run 'train the model', this action takes place. rinse repeat for the regression model with your specific gathered Machine Learn file while running regression mode.
if building a winrate model, load days on to your chart and turn on POL-ATI and when its ready to 'train that model, take a time out and open that MLTrain file and look at the data 'train the model' is about to burn into your catboost engine. You will see a ton of 00000000000 winrates packed in there. I don't want that bad data making it to a catboost engine for predicting markets. engine wont run well with that much bad gas.
gathering the machine learn data takes time. especially getting it from trend and regression mode. One can get this collect this data live while trading the markets or get it off replay when the day is complete.
It is important to set the parameters for target / stop that you will be trading with. If trading NQ and hunting +16 target with -16 stop, I set target to +18 and stop to -14 for train the model. Sampling interval is important. NQ is a fast paced whippy market, 60 seconnds over 30 minute windows to get data for Incremental learn is a little stretched. I prefer 20 second grabs over 4 - 10 minutes.
ES market runs slower. i prefer 30 seconds over 7 - 12 minutes. break it down to seconds. just test build model and see what works best.
@@BrettR-Trades Thanks, Brett. That's very helpful and I appreciate the detailed explanation as well as the pointers about the parameter settings!
@@BrettR-Trades Cleaning up the MLTrain file before "Train the Model" is something I was not doing. I will try this and see how it goes. Thank you.
with POl ATI on during train the model data collection, that data collection is bad and I dont want that bad gas feeding my engine. it's a common sense thing.
use ninjascript window to point you in the right direction as bars are populating. then try building a model using those paramaeters with soild incremental data you have been collecting and saving.
Can you post your settings for training and then when doing replays? I know you mentioned before you train it with a higher win rate but do you then change that for doing replays?
It's not just about training, you also have to know a lot more about it and change the files.
I'm in two forums and no one has any success with it.
Its also playback mode not Live.
@@alikalka I am very familiar with how it works, and what all the files do individually, how to train and optimize my models, along with incremental learning for continuous learning in the ML. I just wanted to know his settings for how he has done it and compare to mine.
@@ice-tim9796are you profitable with Pati? All the ones I know are in SIM mode, nobody makes real money so far.
I never change the winrate. If I train a model for .75%, I am trading for .75%. I have left some starter settings in vids and comments to get someone on the right path. pay attn to ninjascript output when your chart is populating bars. That will give you some nice boundary information.
hey there alikalka. I have live data CQG Auto Arm vids on my channel and Arm Both. Check em out. There is even a clock at the top of the chart showing date, actual time matching bar for bar. Are they being traded on one of my actual brokerage accounts? No, they are trading in sim mode on live data in real time. You will never see Polarity traded on a live brokerage account on this channel. There are liabilities that come along with trading live accounts on social media in real time. so I will keep it in sim 100% of the time. I am not selling anything. I am not calling trades. I am testing a trading software and sharing what I have found.
Recording live is not a priority for me nor is this YT channel. I don't get paid a dime by YT or anyone else for trying to help people out. I'm sharing some things that I have learned about Polarity since so many are struggling with it. Hopefully it benefits some. The majority won't get it and that's okay. Recording YT vids is actually a conflict with my personal trading since recordings are being conducted on my brokerage accounts connected trading pc. but you are welcome to stop by and check things out anytime.
How did you train the model for NQ? Csn you make a separate video on that?
time consuming and I dont have time. Ive dropped enough hints . notes, answers in comments to get someone on the right path.
Your question is basically answered in the video, thanks for sharing
video training instructions are in the FA pdf.
@@BrettR-Trades Right but following those instructions makes it perform nothing close to how it performs for Connor. I am looking to identify what makes it different.
question: how do you open MLTrain file ? I cant find it
I have the folder shown in many of my videos that shows exactly where to find it on your pc. skimming videos and you miss things.
So this and Ron Klemp are all just Connor trying to get a few extra views…prove me wrong
so you maybe spent 1 - 2 mins skimming this video? maybe spend some time actually watching several of the videos and reading the info shared in descriptions and shown in the videos. then get back to me and tell me if you still think I am Connor, Ron or anyone else affiliated with FA.