Predicting Stock Prices in Python

[@_ScyIIa_]

you're underrated man, you deserve a lot more

[@NeuralNine]

Patience is the key ^^ Thank you

[@sachinr1929]

Totally agree

[@baskarpalani2014]

True

[@cansaruhann]

Yeah

[@christoflers]

Hey man great video, only thing I'd recommend is moving your camera overlay to the top right corner though, as it often blocks the current line of code that you are writing/discussing.

[@mihirshahpro]

I agree!!!

[@thecodfather7109]

Your content is amazing! I can't find one video that sucks or that's useless. Keep it up dude, your channel will hit big numbers 👌🏽

[@NeuralNine]

Thanks brother :)

[@bud384]

That's for sure i believe..

[@c0dakw0lfgaang48]

@@NeuralNine which yahoo api is that ? Please tell me the package name and the author name.

[@martinwestin4539]

​@@c0dakw0lfgaang48 there are several yahoo finance apis in python that you can use, but i recommend using yahoo_fin (pip install yahoo_fin)

[@c0dakw0lfgaang48]

@@martinwestin4539 thanks mate

[@akramelomrani8728]

I am so happy that I am one of the oldest channel members you taught me a lot about sockets and also your project ideas are amazing cant wait to see much more I think you are in the road of becoming one of the biggest channels

[@Nvidia-Clan]

Very nice video, learned a lot from it. Only one advice, you could every now and then print the data we are working with so that we see exactly how the data should look. I am guessing that for someone that advanced as you, visualizing the data in your mind is pretty simple, I for example, print it pretty often, just to make sure that I am on the right track. Apart from that, really awesome video, I'm glad the algorithm recommended your channel.

[@GabrielBauerr]

yeah, theory behind would be great to know how to apply the knowledge in different projects and situations. great work man

[@danajanastova8550]

Instead of reshaping with reshape: x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1], 1)), there is a shortcut syntax: x_train = x_train[..., None], where None stands for the new axis added as the last axis. Maybe you want it to be understandable for begginers though, which I totally understand :).

[@greenpumpkin172]

Thanks for the nice tip.

[@devsunny10]

Hello, can you please help me

[@msa3533]

thank you for this amazing content. I now went bankrupt

[@dmitriyvoznyuk7142]

LOL (sorry if it wasn`t a joke)

[@yeweishi6157]

hahaha

[@alexleban6823]

lmao

[@TokyoWasTaken]

RIP... i invested using this .. im kinda profitable

[@Anaskhan-kv5qi]

😂😂

[@delalias5754]

Those muscles.... damn nerds are jacked now... thanks brother

[@dabbopabblo]

21 skinny guys liked your comment

[@makdavian3567]

lol

[@doublegdog]

love these very specific python videos! there are too many tutorial videos but not enough videos on specialized programming topics. I have been meaning to get started on algorithmic trading but had no idea where to start. It is as if you read my mind. Thank you!

[@philippebaillargeon5204]

Here is a tip for everyone that is new to programming: Learn clean coding as you learn how to code. You're gonna thank me later. Cause I'll garantee if you don't know how to clean code and you just copy pasting stuff and adding lines of code your code is gonna be horribly hard to read and understand. Especially algorithms with heavy maths involved.

[@regeleionescu935]

Line 60: model_inputs=total_dataset[len(total_dataset)-len(test_data)-prediction_days:].values

[@muhammadvaliallah2781]

Took me a while but the purpose of this line is to get from 60 days before test and include the test set

[@Kotavedavyas]

does anyone know this error ValueError: Input 0 of layer "sequential" is incompatible with the layer: expected shape=(None, 5, 1), found shape=(None, 4, 1)

[@Sturmbutcher]

​@@Kotavedavyas I am on the same boat... did you find how to fix it?

[@ishan_shrestha]

same... help

[@ramach668]

Anyone able to resolve this error? "expected shape=(None, 5, 1), found shape=(None, 4, 1)". Please help with a workaround

[@helenoliver4838]

The stock market is still a fantastic tool for building wealth, however, so it's wise to consider investing even if you don't have much money to spare.

[@greenquake11931]

Money is a tool that can help you to achieve your goals. It can provide comfort and stability for your family, make it easier to plan for the future, and allow you to save towards important milestones. But to achieve these things, you need to know how to make your money work for you by investing with the right signal.

[@mayacho4910]

@@greenquake11931 Hello, what signal do you invest with ? I'm new here.

[@greenquake11931]

@@mayacho4910 'BRIDGET MARY TUROW"

[@mayacho4910]

@@greenquake11931 I'll like to connect with her. I want to invest.

[@greenquake11931]

@@mayacho4910 look with her name online for her page.

[@dynoagency1323]

thank u man u are the only youtuber who do this things

[@sunhound817]

Great stuff! Don't stop posting, man. This is very cool.

[@_Loki__Odinson_]

Man believe me you would surely get a lots of subscribers in a very few time You are siriously awesome dude

[@sachinr1929]

Great content as always. Keep going !!!

[@NexGenSlayer]

FB Stock the next day from his prediction was 261.10 (1/19/21). The model seems to be pretty good at predicting momentum at least.

[@SylvioSaraiva]

These videos recorded while coding are the best! please keep producing them

[@macel1602]

Great video! And i would love to see some theorical mathematics about what's behind the functions you were talking about

[@antoinepoisson4791]

Look for Denuit, Hainaut & Truffin « Effective Statistical Learning for Actuaries Vol. 3 - Neural Networks & Extensions » 😉

[@DilpreetSingh-sw3ei]

How to predict for next 90 or 100 days if we want to ? For other situations in which we need to predict data for a whole day for every 15 mins intervals

[@nallurivenkat7425]

We need more and more videos of this kind. Just loved it.

[@kaushiksprabhakar9088]

Awesome work dude🔥

[@NeuralNine]

thanks :)

[@kaiprzadka6896]

Very cool! I did the same, but tried it with linear regression. I really recommend changing the Programm a bit, so it tells you If the price will Go Up or down, NOT the exact Number. This is way more accurate and even though it's Not as informative, it probably is more usefull. I did the same with my linear regression attempt and that improved the average accuracy by quite a few percent.

[@amarmemane2583]

hey, kai is this possible to share your code with me?

[@ShakTMT]

Hey ,what did you change to make that edit? Thanks

[@kaiprzadka6896]

@@ShakTMT Well, the easiest solution would be to compare the current price and the predicted price.

[@timeshiftbr]

I’d love to see a more technical video. I like to understand what is happening that actually takes me to the results I get, so I can then understand how to make it better. Nice video, man, congrats for your work

[@marcoscully7021]

you should check out coding trains neural network playlist it goes into a lot of depth into how neural networks function

[@smalltribes88]

MIT posts all of their lectures in algorithmic models on KZbin, more technical than you probably even want hahah

[@kamilzielinski593]

this is actually my 6 months engineers degree condensed in 30 minutes

[@giancarlo9866]

You got an engineering degree in 6 months ?

[@jsj6665]

Lol

[@bardocks72]

Bs

[@zulfadlizein5192]

I immediately hit the subscribe button. Nice content, keep up the good work.

[@punyagalundia9160]

I absolutely love your videos and they have really helped me get started with coding as a beginner. Would love it if you could put your video in another corner so code is visible at all times. Thankss

[@KingCAMBO]

I used the code from this lesson yesterday 5/12/2022 to predict today's close on DIA. The prediction last night was 322.19 - Today's 5/13/2022 real-time close was 322.24!

[@jakob4371]

So?

[@lizardmtb6349]

@@jakob4371 so that's pretty close > 5/12/2022 > 320.82 > prediction > 322.19 real-time close was 322.24

[@milepivo]

I would like to see a detailed video where you explain the tensorflow implementation of the LSTM, because I'm not really sure how they work when you combine them with Dense layers.

[@ProGuitarUA]

Wow man that’s good! Can’t wait to hear more about optimizers and loss functions :) Thanks

[@AdoukaStalks]

Man you do be grinding!

[@NeuralNine]

always ^^

[@timothymagsino]

I’m a CS student and I gotta say, you really have fantastic helpful content

[@annaarbuz3277]

Just an example of NN overfitting. Predicted curve should be shifted to the left (to the past) from the actual curve - then it will mean the prediction. Actualy NN just repeating curve from the past to the future with some bias and curve approximation, and this is the opposite of the meaning of prediction. Whatever if you just need to smooth any curve and shift it to the right, then there are many simple ways, not a neural network.

[@ViralKiller]

You need to train it differently...find historic data where a very large move to the upside or downside has occurred, and then take the 120 candles before that, and label them either "bullish" or "bearish"...then show it the current chart and ask it what will happen next

[@intfamous4001]

and it still wouldn't change anything. This approach is no better than human guessing

[@media1critter]

Even though the models aren't perfect, it's amazing how close they are! They could definitely be handy in assisting day traders with determining how long to hold a position for :) Thanks so much dude!!

[@GianCampos]

I don’t know if you’re going to see this comment, but I must say it: The way you teach the content in details explaining why and what that section does is amazing! Very good content, I’ll see all your videos now that I understand more about machine learning stuff than 3 months ago when I found your channel. Thank you so much! A hello from Brazil!

[@RemorfChuket]

24:26 shouldn't that be len(model_inputs)+1 with the parenthesis before the + 1? also I'm pretty sure that it doesn't even do anything since I tested it and it returns the same array

[@ryanreddy8450]

Great stuff. Could you possibly do a video on how to deploy these into production where newer data is automatically fed in, so we do not have to keep training the models again and again?

[@dvdplays8444]

if that was possible then the stock market would be pointless cause everyone would use it

[@stephenhobbs948]

Great video. Can you move the code up (hit the return 4-7 times) bc your camera is blocking part of the code to the right. If you press return, it will shift the code up so we can see entire line of code. Looking forward to seeing more. Do a video on Q learning.

[@loganathansiva7063]

Hai sir. Thanks for casting your videos. Learnt how to make predictions on stock prices watching your video. I was in search of videos like this. Luckily i got it. Very well narated and explained. Hats off to you sir. May god bless you. .

[@balloney2175]

Learning python at the same time predicting stock success. Fantastic!

[@RAJAT2372]

Thank you so much! This is exactly what i had been looking for! 🙏🏼

[@timmothysanders1324]

THANK YOU for making this.

[@Juanpablete12]

I have this error : ValueError: Input 0 of layer "sequential_1" is incompatible with the layer: expected shape=(None, 60, 1), found shape=(None, 59, 1)

[@zomaareenkijker286]

@Juanpablete12 Have you found a fix to this? I'm running into the same problem.

[@doruksrtoglu408]

Hey, thank you so much for the video, very smooth and understandable. I agree the fellow commentators, please make a separate video to describe the mathematics behind the model. Thanks!

[@Itschotsch]

Yes please, theoretical videos would be of great help!

[@Cristian-bg1pr]

Isn't TimeSeriesSplit Cross-Validation similar to your type of split? It kind of does the same thing. For example: "Train on first 3 days, predict on 4th, then train on first 4 days, predict on 5th" and so on, thank you!

[@damiangames1204]

I am interested in the theory and maths! Also, top quality video.

[@anant.jain11]

man, you're amazing. just amazing, no words i calculated the r2_score for FB dataset prediction it's 0.86 you're predicting stock market with that much accuracy. i might just trade on that 😁 love your content man, keep up the good work❤👍

[@srkk3360]

could you tell me the code for it

[@angelgabrielortiz-rodrigue2937]

Great videos. Love this type of content. Please make more

[@loganathansiva7063]

Thank you very much. I was looking for predictions made like this.. May god bless u.

[@joeyb8455]

A scatter plot and/or calculating the correlation of the predicted price change vs actual price change would give much more insight in wether or not the model has some predictability

[@timothysiallagan3886]

I get the ValueError: input 0 of layer "sequential" is incompatible with the layer: expected shape =(None, 60, 1), found shape=(None, 59, 1) . Despite the warning you received at 26:22, you still get the prediction, unlike me. Does anyone have the same issue?

[@ishan_shrestha]

same.. did u solve it?

[@ZachDiesel]

did you ever figure it out?

[@elvischen1590]

Thanks for sharing, good work! For whom is watching this video, again, the previous data does not correctly predict the future!

[@Krumplyn]

Hi, thank you for the quality content! I just wanted to let you know, that I would be also interested in the math behind the whole AI thing. Have a nice day and be safe :)

[@press1to]

I’m looking for someone to adapt this code with my parameter and formulas

[@albertenstein5965]

Amazing channel! Thanks for the tutorial dude!

[@Sturmbutcher]

Very well explained! I am trying to figure out if I can predict the price of certain currency and then calculate the "value at risk". My doubt is, I have to train this model every day to get a better result? I plan to use it on production to compare it with the actual state of art of "value at risk", using the covariance or the historical method.

[@PurpleGDuffy]

This was great! Thank you!

[@mceiche6626]

I would love to see some theory about the opimizer. Really interesting but hard to understand on the Mathematical point of view

[@Henrychc19970911]

Awesome content! Keep up going

[@cebeb16kmrdayaasaagar87]

dude thank u !! u are really great !!! Thank god i found your video !!! This helped me in the project thanks.

[@bhuvanaggarwal7387]

Awesome content man!!

[@g.paudra8942]

This is more simple code than i use, if you try to use epochs up to 1000 and batch size 64, maybe the accuracy will be same as my model, mine is always above 99% predicting the next day closing

[@sketchychillandchill]

Is your test data in your train dataset?

[@jaredfiacco1107]

Thank you, @NeuralNine! The people who gave you dislikes need a neural network to predict better taste in tutorials.

[@CodeWithJoe]

What was the actual success rate you had in predicting?

[@dipikabudhathoki00]

💯% true n actual content u r on extremely right way

[@vsp540]

thanks for the video man it's great! and yeah it would be super cool if you made a video explaining the math!

[@stevenduyck5204]

Could I say i need to change the last value from (model_inputs+1) to let us say '+5' to see price prediction in 5 days? real_data = [model_inputs[len(model_inputs) + 1 - prediction_days:len(model_inputs+1), 0]]

[@ebz7486]

Hey were you able to figure this out I’m trying to do it now

[@_M_643]

You do the best programming tutorials. Easy to understand and learn. I have a question. Can you make a tutorial about how to make a Discord Bot written in Python. If you don't know what Discord is, it is one very popular chat platform.

[@my-financial-wealthblog4423]

Wow! As a trader that's amazing thing to learn. Thanks for the video

[@CodeWithJoe]

what comes after value in line 60? model_inputs = total_dataset[len(total_dataset) -len(test_data) - prediction_days:], value

[@NeuralNine]

nothing it ends with .values ^^

[@CodeWithJoe]

@@NeuralNine danke, ich hatte value deswegen der series error. Geiler kanal junge danke für die ganzen videos, mach weiter so

[@shinrei19]

wow! nice! i can learn phyton! awesome content!

[@Josephus_vanDenElzen]

after running line 75 (at 19:39) I get an error: ValueError: Found array with dim 3. Estimator expected

[@rayanebouayad920]

I'm getting the exact same error and it's bugging me. Will let you know as soon as I find the answer.

[@Josephus_vanDenElzen]

@@rayanebouayad920 great

[@jitan-art]

@@Josephus_vanDenElzen check line 42, it fixed my issue with the same error

[@enjigaming11]

Can someone tell me what he wrote in .value in line number 60??

[@julesancillon275]

Super nice content! A video about optimizers would be super welcomed !!

[@faridparvez]

Great video, very clear. Please make a video on roadmap to learning python from scratch, specifically for stock analysis, chart analysis, getting trade signals using charts and statistical analysis of stocks. I mean create a roadmap on the course tailored cut for only stock trading. Regards Farid

[@Kotavedavyas]

can any one help? ValueError: Input 0 of layer "sequential" is incompatible with the layer: expected shape=(None, 60, 1), found shape=(None, 59, 1)

[@joehuang1889]

Same error here, not sure either

[@ZachDiesel]

did you figure it out?

[@das250250]

Good videos , notes of possible ways to improve .. A. Lower keyboard sound relative to your voice ..maybe a noise cancel lapel microphone ? Maybe AI noise cancelling B. Some commentary on the commands used and what library they are from to connect the resources used and what goes on in each command . Thanks

[@CodeWithJoe]

How can I fix this? Traceback (most recent call last): File "main.py", line 38, in model.add(LSTM(units=50, return_sequences=True, input_shape=(x_train.shape[1], 1))) AttributeError: 'list' object has no attribute 'shape' Stackoverflow says that the issue is a 2d lis while it should be 3d, but i dont know what to do with that information?

[@robertchu4434]

Are you sure x_train is a numpy array and not a list?

[@bielvv9148]

I am trying to reproduce your model and I get this errors: 'Traceback (most recent call last): line 21, in data = web.DataReader(company, 'yahoo', start, end) line 210, in wrapper return func(*args, **kwargs) line 370, in DataReader return YahooDailyReader( line 253, in read df = self._read_one_data(self.url, params=self._get_params(self.symbols)) line 153, in _read_one_data data = j["context"]["dispatcher"]["stores"]["HistoricalPriceStore"] TypeError: string indices must be integers Does anybody now what is wrong? I tried to fix it in many different ways but nothing worked.

[@giuliamancini753]

Same for me. I don't know what to do. Can some one help us?

[@nippi86]

Same for me

[@atasozuvesiirler]

Same for me..

[@bielvv9148]

@@atasozuvesiirler Hey, if you're trying to get a project like this to work, I recommend trying to understant the basics and then building up the code with ChatGPT for example. By doing so I got it to work. Just so you can have some reference, I used yfinance to fetch the data. Good luck!

[@atasozuvesiirler]

@@bielvv9148 thanks I already have the solution in same way too. 👍

[@ottobormann]

I am interested in why are you choosing that topology of the neural network. Why 3 LSTM layer each with 50 neurons. It is clear why the last layer is of one neuron. But the previous ones are a mystery.

[@YundongCai]

Line 87 should be real_data = [model_inputs[len(model_inputs) + 1 - prediction_days: len(model_inputs) + 1, 0]]

[@srijanpareek4563]

mate i am having error ValueError: Input 0 of layer "sequential_1" is incompatible with the layer: expected shape=(None, 60, 1), found shape=(None, 59, 1) THIS ISSUE IN LAST3RD LINE prediction=model.predict(real_data)

[@karthikeyan-kc2py]

Thanks for your efforts and collection of Good informations 👍👍

[@erictovbin4200]

Can you make a video where you explain the mathematics?😄

[@martinsmolik2449]

Three blue one brown does an excellent job of this!

[@mw7185]

At 10:46 we can t see the code since it is covered by your picture )

[@stansuen8072]

Thanks! Love your videos. Wonder if there is a good way to detect pattern (instead of predict future prices) such as inverse head-and-shoulder and cup-and-handle?

[@davidef2789]

I wrote the whole code and installed Pandas, Anaconda, Tensorflow, etc. but I can't see anything because it gives so many errors

[@PeterPetahPetah]

great video! subbed!

[@ananyaiyer6049]

Great video!! thank you so much

[@sobhanmovassagh6792]

for create x_train you can use sliding_window_view in numpy library

[@25_FpS]

Very informative video, can you provide the link for the code ?

[@sonyh2so4]

Hi ! When i tried to run the last part of the code in spyder, i had this error (not a warning like the video): ValueError: Error when checking input: expected lstm_4_input to have shape (60, 1) but got array with shape (59, 1) How can i solve this?

[@parthbansal2775]

I ran into the same problem, did you get a solution?

[@kyleyoung4974]

​@@parthbansal2775 Look at your x_train = np.reshape lines.. those errors are regarding the reshaping usually

[@kyleyoung4974]

Look at your x_train = np.reshape lines.. those errors are regarding the reshaping usually

[@parthbansal2775]

@@kyleyoung4974 Thank you very much I fixed that

[@marcmolinsdeamaya7189]

@@parthbansal2775 Can you share how did you fixed pls? Im with the same issue, thanks!!

[@mr.sweaty7862]

When running (22.00) i get this error. How to fix? Traceback (most recent call last): File "C:\Users\soubi\PycharmProjects\tensorflow\Code\linear regression.py", line 19, in data = web.DataReader(company, 'yahoo', start, end) File "C:\Users\soubi\PycharmProjects\tensorflow\venv\lib\site-packages\pandas\util\_decorators.py", line 211, in wrapper return func(*args, **kwargs) File "C:\Users\soubi\PycharmProjects\tensorflow\venv\lib\site-packages\pandas_datareader\data.py", line 379, in DataReader ).read() File "C:\Users\soubi\PycharmProjects\tensorflow\venv\lib\site-packages\pandas_datareader\base.py", line 253, in read df = self._read_one_data(self.url, params=self._get_params(self.symbols)) File "C:\Users\soubi\PycharmProjects\tensorflow\venv\lib\site-packages\pandas_datareader\yahoo\daily.py", line 153, in _read_one_data data = j["context"]["dispatcher"]["stores"]["HistoricalPriceStore"] TypeError: string indices must be integers Also, I can't do from tensorflow.keras.models Import sequential or tensorflow.keras.layers... So instead i've done: from tensorflow import keras from sklearn.preprocessing import MinMaxScaler from keras.models import Sequential from keras.layers import Dense, Dropout, LSTM Is this okay?

[@_applejuice_]

same

[@melihaksoy1129]

did you guys find the solution.Im stuck too

[@mr.sweaty7862]

@@melihaksoy1129 Nope. Not yet

[@jihyeonku]

same

[@morto1826]

tensorflow.python.keras.models import Sequential worked for me,as for the string indices must be integers, im struggling to figure that out now

[@tacowashington717]

I can’t wait to fully comprehend this

[@adebayoognibo4211]

hi thanks for the video, I have a little problem on line 68 ---> AttributeError: 'numpy.ndarray' object has no attribute 'append'

[@alikellaway3901]

Cool video bro 🙂

[@Apapa88657]

Is there a way to predict not only next day’s price, but also next few days’ prices?

[@marbentraje3245]

Optimizer and loss model could be a brain-breaking topic, but I'm curious. Very informative and interesting contests sir! GoodJob!

[@randalltockes2247]

Really like your videos, I've watch many at this point. Only thing I would comment is that as much as its nice to see you at the beginning/end, I think it would help to turn off your video while actually writing code. I was frustrated a few times trying to keep up when you were coding along the bottom of the window and the right extents were hidden by your pip which is kind of unnecessary. Anyway that said, enjoy the content. Keep it up.

[@serbrad6426]

Please do make more theoretical videos. It really helps to learn and understand the fundamentals of the underlying. We're gonna learn them fast and more accurate. Please do and let us join the channel. I would like to be a paid subscriber.