I am writing a feed forward neural network that uses NEAT to learn to trade in the forex market. I've gotten a bit stuck trying to choose an activation function for my network. I am reluctant to use relu because of the dying network problem, but there are a lot of relu variations that I could consider using. I was thinking of using Leaky Relu or Relu6 (or perhaps a combination of the two) or Selu but I'm not sure which one would be best. Can anyone give me advice on which one would work best for my network or if I should just stick with normal relu or another not mentioned activation function? Also any comments or advice or critiques of the way I'm going about this would be appreciated as well. My neural net is going to be fairly shallow, with just one hidden layer to start, if that matters. I understand some activation functions work better on deeper nets and some on more shallow ones. submitted by
First off, let me state that I am not even close to a professional when it comes to trading. I got in a few months ago for passive income from copying 3.14fx and have come a long way since then, quadrupling my initial investment and losing half of it. I've watched traders such as cfdtrader, Lumyo, Robot, and crypto_chris lose several hundred percent after a fail from opening multiple positions. I got into 1broker to make money without monitoring it, but instead I learned a lot about trading and risk management, even profiting off several of my own trades. It's a valuable experience in itself even if you're not profiting and I wouldn't give it up for anything. If these losses are enough to make you quit, so be it. Investing comes with risks that some people can't handle. It's not free money. https://www.dailyfx.com/calendar
is the economic calendar that I use while trading. High importance events can easily trigger a 80% loss or gain depending on the direction you choose. It's highly risky to trade when someone of great importance such as Draghi or Yellen are speaking.
Even if you follow a general MAX
5% rule, you will still lose up to 16% of your account if somebody opens 4 of the same positions and they stop at 80%. Making back money is also tougher than losing it, as once you lose 16% of your account, 5% of your account is a lot less than before. Therefore, you have less capital per trade.
Also, be careful when changing your copy amount. I often see copiers saying things like "Great work, I'm upping my copy amount" and "Increased copy amount from x to x". In my opinion, increasing a copy amount should only be done when your initial amount is already low. Losses on a higher copy amount may wipe out the gains on a smaller copy amount. (-50% loss with 0.1 btc = +100% gains with 0.05 btc) Always stick to a 5% max rule unless you're feeling risky.
Then, there comes the gambling/greed phase that many new copiers often do. (Guilty of this myself). After extreme success, a copier may feel the need to upgrade their copy reward to maximize profit. Or after extreme failure, a copier may feel they need to upgrade their copy reward to make up for losses. All of these are mistakes.
1broker is not filled with market professionals. Most of us here are either self taught or complete novices. Professionals would not be sharing their trades for about $70-80 for each trade (at best). They won't be asking for copiers on other traders' profiles. They won't be using a Pikachu as their profile picture. They won't be using a broker that isn't heavily regulated and insured. They would be using their own capital to make millions off of trades.
Remember, any newbie can easily accumulate winning trades by gambling with high leverage. As long as they have around $1300 as of now, they can easily create a profile that suggests that they are a professional, when in reality they are entering at random points and exiting when a position turns into profit, rather than using technical analysis and watching economic calendars.
And even the best of traders will have their ups and downs. I've stuck with 3.14FX even when he reached -100% this month because he's had a great history on this site. I feel that he can make the money that he loses back. And even though he has doubled up on a position yesterday (not sure why, probably was extremely confident), it was a success.
Can you really trust anyone? No way! Unlike regular trading, 1broker is more unregulated. Signing up requires no personal information so any user with malicious intent can build up a steady reputation and perform an exit scam (or have a massive failure) without any reparations. Robot has no link to any social media or anything in his profile. For all we know, he could own another account that has -100%, and he is depending on luck while opening multiple positions to accumulate followers. (I just used Robot as an example, my intent is not to accuse him of multiple accounts)
Then there are potential exit scams (from a trader, not 1broker itself) that will drain a decent portion of your account. There's a reason why you have a choice to choose how many trades maximum you can copy per day. This hasn't happened yet, but it will definitely happen in the foreseeable future. Somebody will set up an order for 50 shorts and 50 longs and set the take profit and stop loss the opposite of each other. Then after closing, they'll withdraw their bitcoin never to be heard of again.
When you put your trust in a trader, you should trust them to carefully monitor a trade. Unfortunately, there's currently no way to tell if your copied trader is online or not, so you'll never know if they're in a coma and won't be back for another 6 months. My suggestion is to either take profit when you think that the conditions are correct or
just trust the trader. Nobody can see the future. If you think that you'll rather close the trade before the weekend, it's your choice. If you think upcoming news will destroy the trade, feel free to close early. However, be prepared for regret if it goes up, or a great feeling that you dodged a bullet if it goes down. It's all a part of trading.
1broker's copy system is seriously flawed at the moment. Of course, there's no easy way to fix it. Why would a great trader want to share one of their trades if they're not getting much out of it? This encourages opening multiple positions to maximize copy rewards, which can result in massive losses. Robot is one of the traders exploiting this.
So how can you prevent massive losses? There's really no way. You're putting your trust in random people without an identity, who can easily be a scammer. When it comes to people like Robot, I put 1-2% of my funds because I know that he opens multiple positions. This is why I'm always sticking with 3.14FX, he established himself a long time ago and he knows what he is doing. Somebody who has been on the platform for over 3 years with several losses is preferable to an anonymous newcomer who just registered but appears to be good at trading.
Also, the percentage on 1broker is misleading. You may think "Wow, I'm going to get an 500% of my initial investment if I copy Lumyo!" In reality, you should only be using 5% max of your capital per trade. If you copied him from the beginning (I started copying at around 90%), you should have only gained 25% rather than 500%. But still, 25% of your initial investment is huge.
My opinion on several traders vits2015
: If you watched vits2015 from the beginning, you would know that their style of trading is... off. 15 positions on UK100, all short, some of them at -30% when I first saw him as a successful trader. What does that tell me about him? He can open up to 6 positions on the same trade at once, and is willing to hold them as long as possible to get a profit. (Average holding time 8 days) gtfann
: Even with recent losses, he still appears to be a decent trader. It seems that he upped his usual leverage due to the crowds of traders flocking to copy him though. Multiple positions with a lower leverage isn't really something that I like either, but I'm sticking with him for now until there's a drastic change. vaiono
: He lets his losses play out and even though he has a decent track record,it's still risky to play with. Silver is extremely volatile and due to leverage, a small move in any direction can either be a huge loss or huge gain. Snortex
: Pretty much a meme on 1broker. He acknowledges his trading style and warns his copiers. I like him as a person due to his warnings,
but still wouldn't recommend copying him unless you can afford to lose a lot
. Edit: After examination of his trades, I feel like he's not only gambling like his description suggests. His entries are planned out carefully (Although that has hurt him when there was a flash crash). You'll take several 80% losses but you may take several 400% gains. He seems to have a habit of chasing a trade, which can lead to multiple 80% losses. However, once the trend reverses, his profits go through the roof. When you're copying, copy for the long term! Of course, feel free to uncopy if you feel that the bottom is still far away. noIDea
: He has had bad stretches in the past, but still makes his way back. I think he's a good trader and even though he opens multiple positions, he's one of the best at setting stop losses so the risk is not as high as others who open multiple trades. Gold_Gangsta
: Name change from Crypto Chris for some reason? Be wary of multiple positions as the USDJPY fiasco shows. Seems to be doing fairly well with gold as of now. 1monk2
: Multiple positions fairly often, even says that he's drunk in the description. This is gambling. knightlife999
: The description definitely shows promise. There is no proof to those claims on the site, but I feel it's safe to allocate some of your funds toward copying him with his track record. HedgeCryFx Risk 5
: Decent trader, pays attention to economic calendar as well. The only problem is that he lets losses play out to 80% boogi
: I would be wary about the higher losses, but then again, there's a good track record. sergiomc
: Seems to be decent at trading stocks. With an average holding time of 14 days and leverage of 10, you should be expecting to lose about ~3.92% of your gains to financing, which is not actually that much. Cool Hand Luke
: Low leverage trading. If you were to copy him, I would recommend only using 1% or 2% of your account max per trade if you plan on copying others as well. He's a great trader for slow steady gains, but if you're looking to get rich fast or go broke trying, this is not the guy for you. eylemc
: Quick trades with minimal profit and no losses so far. As of now, it may be too early to judge, but I think that he's somebody that might be worth copying. Edit: Seems to let losses play out to 80%. Be wary. 3.14fx
: Back in the game, doing well with stocks and USDJPY recently. High leverage, but usually stops losses within a reasonable range. SunnyNet
: Small gains, huge losses. Be wary as your first copied trade could easily be a -80%. SatoshiReport
: Trading using a neural network, after looking deep into the trade history, I'm not so sure about it. Correct me if I am wrong, but the bot doesn't take into account important news and events. Edit: This bot has too many flaws to continue copying in my opinion. Even with the previous gains, it opens the same position as soon as one closes, negating the 33% stop loss AND
forcing a loss due to the spread. The only thing that keeps it out of the negative is the rare 80% gains that you might find once in a while. CryptoMessiah
: The image being shown on his twitter has weird numbers on it (USDJPY at 100-103 in the matter of minutes), I think it's a simulator so it isn't actually "proved". Also, asking people to copy for "free money" is misleading as anything can go wrong in the forex market, there is no guaranteed money. I copied with a minimum 0.001 btc and will update this post if the bot proves to be successful. Edit: Tons of losses trying to get the right direction and then huge wins. I would say it's ok, but you're better off with a human capitalizing on gains. The only advantage to this bot is 24/7 hour trading. kosanet
: His description says it all. Be careful while copying, but don't be discouraged to place an amount you can afford to lose. He seems to have a great history of monitoring losses (positions never get below 20%) but it's still a new profile who clearly states that he's not a pro. May open multiple trades and trading with USDJPY a lot. His scalping strategy means that overnight fees won't be an issue. Edit: Now he's starting to be a little more risky with his trades as more copiers arrive. Be careful, he never reached liquidation at 80% yet but he could at any moment. google
: A bit late to the party, but what can I say? I honestly can't believe he accumulated 190 copiers but he seems to have faded out quickly. Golgo13 is having a fun time on all of his trades KillerWhale
: Extremely high risk with all of those multiple positions. Like google and robot, don't be fooled by performance recently and look through their whole account. People who saw the 220% recently may have missed when he was in -475% a few days ago. SoontobeWW3
: Great trader in my opinion. However, I think emotion plays a role in his trading as every huge loss is often followed by more. APPoh
: Seems to know what he's doing. However, there is a very short trading history and we're never sure. Positions can reach 50% without closing, so it's very possible that he might let losses play out to 80%. dingo
: Not much to say. Good with 1 position at a time, and even with the 80% loss last month, still ended in profit. Be careful as he might sometimes not stop a position and instead wait for it to recover and a 80% loss is huge compared to his gains.
Edit August 12: Will stop adding new traders now. Before copying someone, remember:
- Check their trading history, ALL OF IT. You're entrusting them with your money, you should be 100% sure.
- Wait until they've established themselves. Sure, you can be frustrated about potentially losing 200% profit, but it sure beats 700% losses.
I already expressed my views on Robot and 3.14FX above. Lumyo is currently inactive.
Last tip: Don't uncopy people if you feel like they can make it back. If you choose to copy someone, you're in it for the long run. Now this may contradict some of my earlier statements, but if you have somebody that you believe in, don't uncopy them after a loss. Eventually, they will make their way back up and after you see their success again, you'll be tempted to copy again. Of course, if you are copying somebody who you have no faith in, feel free to drop them. Cutting your losses short is important to learn in trading.
Hi all again! If you’re reading my blog regularly you know that I have published a bunch of tutorials on financial time series forecasting using neural networks. If you’re here for the first ... Neural network processing is mainly divided into a learning phase and an inference phase. In the learning phase, data is input to the neural network and trained. In the inference phase, you actually enter unknown information and make predictions. Implementation. This time, I will use TensorFlow as a library for building neural networks. Neural networks with three hidden layers of ReLU neurons are trained as RL agents under the Q-learning algorithm by a novel simulated market environment framework which consistently induces stable ... Foren-Übersicht ‹ Forex Strategien, Analysen and Ressourcen ‹ Forex Markt; Internet Marketing & SEO by www.seoline.de; Ändere Schriftgröße; Druckansicht; FAQ; The Beginner's Machine Learning and Neural Networks Bootcamp . Allgemeines und Aktuelles zum Thema Devisen- und Währungsmarkt. 1 Beitrag • Seite 1 von 1. The Beginner's Machine Learning and Neural Networks Bootcamp. Beitrag ... Neural networks in trading = waste of time. The general rule with neural networks is that if you want to extract valuable results you have to feed the system with valuable data. Such data does not exist in the Forex space because this market is decentralized. So with technical NN applied to Forex you will get the classic GIGO. (garbage in ... I want to implement trading system from scratch based only on deep learning… Get started. Open in app. Alexandr Honchar. 8.2K Followers · About. Follow. Get started. Neural networks for ... Finally, there is a last parameter to know to be able to control the way the neural network learns : the “learning rate”. The name says it all, this new value determines on what speed the neural network will learn, or more specifically how it will modify a weight, little by little or by bigger steps. 1 is generally a good value for that parameter. Perceptron. Okay, we know the basics, let ...
Testing the artificial neural network P-net, included in the Expert Advisor for the Forex market - Duration: 16:58. Ivan Negre 2,460 views This Quantra video on Deep Learning in Python will give you an overview of a course for quants and traders to implement neural network and deep learning in financial markets. Offered by Dr. Ernest ... This is the first video in a series I'll be doing called Neural Networks in Tradingview! This first one is going to be focusing on making the MACD portion of a 3 indicator strategy, that we're ... Crypto Trading With Neural Networks: Machine Learning & Markets - Duration: 9:16. Proof Suite 19,601 views. 9:16 . 10$ to 124$ in 5 minutes - IQ Option Live Trades Starting With Only $10 ... Try Atera! If playback doesn't begin shortly, try restarting your device. An error occurred. Please try again later. (Playback ID: PvUNcn_meEgWTet9) Learn More. You're signed out. Videos you watch... Crypto Trading With Neural Networks: Machine Learning & Markets - Duration: 9:16. Proof Suite ... How to Build a Winning Machine Learning FOREX Strategy in Python: Introduction - Duration: 14:32 ... Forex Robot TFOT http://iticsoftware.com/mt4-robot-tfot full code with description here: https://mikepapinski.github.io/deep%20learning/machine%20learning/python/forex/2018/12/15/Predict-Candlestick-patterns-with-... How to Build a Winning Machine Learning FOREX Strategy in Python: Introduction - Duration: 14:32. ... Convolutional Neural Networks And Unconventional Data - Predicting The Stock Market Using ...