Offshore
Video
Moon Dev
If you are not using bots to trade in 2026
You’re probably not going to last https://t.co/cKga0OMJhw
tweet
If you are not using bots to trade in 2026
You’re probably not going to last https://t.co/cKga0OMJhw
tweet
Moon Dev
From Hundreds of Thousands Lost to a 24/7 AI Bot: My TCN Model for High Frequency Trading
most people think you need a basement full of liquid cooled servers and a phd from mit to run high frequency trading bots but i just proved that you can build a predatory hft system right from your kitchen table with nothing but an internet connection and some ai. there is a specific reason why 99 percent of manual traders end up as exit liquidity for the big banks and it has everything to do with the temporal convolutional network model i just finished cooking up
the secret sauce starts with a machine learning architecture that has a 73.3 percent area under the curve which basically means it has a massive statistical edge over the rest of the market. when you combine that with a 56 percent win rate you are no longer guessing which way the price will move because you are playing a game of pure probability
i believe that code is the great equalizer because i spent years losing money with liquidations and over trading before i realized that my human emotions were the only thing holding me back. in the past i spent hundreds of thousands of dollars on developers to build apps for me because i was terrified that i would never be able to code myself
it was only after i decided to learn live on youtube and iterate through the pain that i finally reached the point where fully automated systems are trading for me instead of me getting liquidated. you might think the hard part is coming up with the strategy but the real trap is actually in the technical execution of the bot itself
the engine of this hft bot is built on order flow imbalance data which we pull directly from the binance websocket to see exactly where the big orders are hitting. we are not just looking at price action but we are calculating things like ofi velocity and acceleration along with multi level weighted ofi using harmonic waiting
by looking at depth imbalance and volume spread we can predict where the market is going to move in the next few seconds with scary accuracy. this model is so much more powerful than traditional correlation based strategies because it actually understands the micro price structure of the order book
the implementation of these bots requires a strict entry and exit loop where we only use maker orders to avoid paying taker fees and losing our edge to the exchange. we make sure to always fill on the bids or the asks and cancel any orders that do not get filled immediately before putting in new ones
one of the biggest adjustments we made was increasing the max hold time to 300 seconds so the bot has enough room to let the trade play out without getting chopped up by noise. tracking stats across multiple sub accounts is essential because it allows us to incubate different versions of the strategy at the same time
you will run into bugs that seem impossible to fix like when your bot throws a signature error or fails because your anaconda environment is missing a specific package. i spent an hour chasing a ghost in the machine only to realize that the tensorflow environment needed a simple pip install for the extended exchange sdk
this is why most people quit because they expect the path to be easy when in reality you have to be willing to sit in the trenches and debug your code until it works perfectly. building is compounding and every time you solve a technical problem you are adding another layer of armor to your trading system
the rbi system which stands for research backtest and incubate is the only way to move from a theoretical idea to a production ready bot that actually prints. once the model shows a strong edge in backtesting we move it into the incubator phase where it trades small size in the live market to prove its worth
we have different bots running simultaneously where one might use a liquidation filter to avoid high volatility periods while the other trades through everything. it is a constant game of testing and optimizing because the [...]
From Hundreds of Thousands Lost to a 24/7 AI Bot: My TCN Model for High Frequency Trading
most people think you need a basement full of liquid cooled servers and a phd from mit to run high frequency trading bots but i just proved that you can build a predatory hft system right from your kitchen table with nothing but an internet connection and some ai. there is a specific reason why 99 percent of manual traders end up as exit liquidity for the big banks and it has everything to do with the temporal convolutional network model i just finished cooking up
the secret sauce starts with a machine learning architecture that has a 73.3 percent area under the curve which basically means it has a massive statistical edge over the rest of the market. when you combine that with a 56 percent win rate you are no longer guessing which way the price will move because you are playing a game of pure probability
i believe that code is the great equalizer because i spent years losing money with liquidations and over trading before i realized that my human emotions were the only thing holding me back. in the past i spent hundreds of thousands of dollars on developers to build apps for me because i was terrified that i would never be able to code myself
it was only after i decided to learn live on youtube and iterate through the pain that i finally reached the point where fully automated systems are trading for me instead of me getting liquidated. you might think the hard part is coming up with the strategy but the real trap is actually in the technical execution of the bot itself
the engine of this hft bot is built on order flow imbalance data which we pull directly from the binance websocket to see exactly where the big orders are hitting. we are not just looking at price action but we are calculating things like ofi velocity and acceleration along with multi level weighted ofi using harmonic waiting
by looking at depth imbalance and volume spread we can predict where the market is going to move in the next few seconds with scary accuracy. this model is so much more powerful than traditional correlation based strategies because it actually understands the micro price structure of the order book
the implementation of these bots requires a strict entry and exit loop where we only use maker orders to avoid paying taker fees and losing our edge to the exchange. we make sure to always fill on the bids or the asks and cancel any orders that do not get filled immediately before putting in new ones
one of the biggest adjustments we made was increasing the max hold time to 300 seconds so the bot has enough room to let the trade play out without getting chopped up by noise. tracking stats across multiple sub accounts is essential because it allows us to incubate different versions of the strategy at the same time
you will run into bugs that seem impossible to fix like when your bot throws a signature error or fails because your anaconda environment is missing a specific package. i spent an hour chasing a ghost in the machine only to realize that the tensorflow environment needed a simple pip install for the extended exchange sdk
this is why most people quit because they expect the path to be easy when in reality you have to be willing to sit in the trenches and debug your code until it works perfectly. building is compounding and every time you solve a technical problem you are adding another layer of armor to your trading system
the rbi system which stands for research backtest and incubate is the only way to move from a theoretical idea to a production ready bot that actually prints. once the model shows a strong edge in backtesting we move it into the incubator phase where it trades small size in the live market to prove its worth
we have different bots running simultaneously where one might use a liquidation filter to avoid high volatility periods while the other trades through everything. it is a constant game of testing and optimizing because the [...]
Offshore
Moon Dev From Hundreds of Thousands Lost to a 24/7 AI Bot: My TCN Model for High Frequency Trading most people think you need a basement full of liquid cooled servers and a phd from mit to run high frequency trading bots but i just proved that you can build…
market is always evolving and what worked yesterday might not work tomorrow
if you are still trading by hand you are basically bringing a knife to a gunfight where the guns are powered by ai and they never sleep or get tired. the beauty of learning to code is that you no longer have to chase the market because you have built a machine that waits for the perfect setup to appear
i am here live every single day because i want to show you that this level of automation is possible for anyone who is willing to put in the work. take the code and play with it yourself because the more you subtract the noise and focus on the logic the closer you get to true financial freedom
tweet
if you are still trading by hand you are basically bringing a knife to a gunfight where the guns are powered by ai and they never sleep or get tired. the beauty of learning to code is that you no longer have to chase the market because you have built a machine that waits for the perfect setup to appear
i am here live every single day because i want to show you that this level of automation is possible for anyone who is willing to put in the work. take the code and play with it yourself because the more you subtract the noise and focus on the logic the closer you get to true financial freedom
tweet
X (formerly Twitter)
Moon Dev (@MoonDevOnYT) on X
From Hundreds of Thousands Lost to a 24/7 AI Bot: My TCN Model for High Frequency Trading
most people think you need a basement full of liquid cooled servers and a phd from mit to run high frequency trading bots but i just proved that you can build a predatory…
most people think you need a basement full of liquid cooled servers and a phd from mit to run high frequency trading bots but i just proved that you can build a predatory…
Moon Dev
Beyond Generative AI: Building a High-Precision LSTM Trading Bot from Scratch
most people are chasing the shiny objects of generative ai while the real wealth is hidden in the machine learning engines that actually power them. i am going to show you exactly how to build a system that is fundamentally better than any ai assistant you have ever seen.
before we dive into the code i need to tell you about the one year lead i have on the rest of the market. if you start today you are already behind unless you understand the specific data structure i am about to reveal.
i believe that code is the great equalizer because through losing money with liquidations and over trading i knew i had to automate my trading. i spent hundreds of thousands on devs for apps in the past thinking i would not be able to code myself.
with bots you must iterate to success so i decided to learn live and now i have fully automated systems trading for me instead of getting liquidated. everyone is talking about ai agents right now but they are mostly just general purpose tools that lack mathematical precision.
machine learning is actually the precursor to what we call ai today and it is much better for finding the repetitive patterns in the market. if you use the same models as everyone else you will never have an edge which is why the data you feed them is everything.
i have been storing proprietary tick level liquidation data for over a year now which gives me a massive head start on the competition. most traders use open high low close data that a million other people can access easily.
the real alpha is found in the data that others simply do not have or are too lazy to collect themselves. if you can find patterns in the liquidation flow before they are absorbed into the price you can predict the next move with scary accuracy.
we are going to focus on lstms or long short term memory networks because they excel at detecting sequential patterns in order flow. these models can remember what happened sixty minutes ago and use that context to predict what will happen in the next five minutes.
there is a specific feature that most people overlook when they try to build these models and it has nothing to do with price. i call it the whale liquidation ratio and it is one of the five loops i use to keep the model focused on high impact events.
while most people are trying to predict price we are actually trying to predict the probability of a specific move based on liquidation intensity. if you only look at price indicators like rsi or macd you are just doing regular technical analysis which has no edge.
i focus purely on usd liquidation volume and the velocity of those liquidations to see when a cascade is building up. the reason this is better than ai is because we are handcrafting the features to look for specific market anomalies.
one of the biggest mistakes i made early on was faking my data with forward filling or back filling to make the charts look pretty. in the real world that creates data leakage that will blow up your account the moment you go live.
you have to be willing to drop rows with missing information and keep your data as raw and honest as possible. if your model shows ninety five percent accuracy it is almost certainly overfit and you are in for a rude awakening.
there is a rumor that wall street is trying to suppress this type of information because it allows the retail trader to compete on their level. i have seen my own content being pulled because it shows too much of the internal plumbing of the markets.
i am not going to stop sharing this because code belongs to everyone and it is the only way to escape the trap of manual trading. if i do not show up one day it is because the powers that be have finally decided i am too much of a threat.
to build a proper lstm you need to split your data into eighty percent training ten percent validation and ten percent test sets. the validation set is critical because it tells you when[...]
Beyond Generative AI: Building a High-Precision LSTM Trading Bot from Scratch
most people are chasing the shiny objects of generative ai while the real wealth is hidden in the machine learning engines that actually power them. i am going to show you exactly how to build a system that is fundamentally better than any ai assistant you have ever seen.
before we dive into the code i need to tell you about the one year lead i have on the rest of the market. if you start today you are already behind unless you understand the specific data structure i am about to reveal.
i believe that code is the great equalizer because through losing money with liquidations and over trading i knew i had to automate my trading. i spent hundreds of thousands on devs for apps in the past thinking i would not be able to code myself.
with bots you must iterate to success so i decided to learn live and now i have fully automated systems trading for me instead of getting liquidated. everyone is talking about ai agents right now but they are mostly just general purpose tools that lack mathematical precision.
machine learning is actually the precursor to what we call ai today and it is much better for finding the repetitive patterns in the market. if you use the same models as everyone else you will never have an edge which is why the data you feed them is everything.
i have been storing proprietary tick level liquidation data for over a year now which gives me a massive head start on the competition. most traders use open high low close data that a million other people can access easily.
the real alpha is found in the data that others simply do not have or are too lazy to collect themselves. if you can find patterns in the liquidation flow before they are absorbed into the price you can predict the next move with scary accuracy.
we are going to focus on lstms or long short term memory networks because they excel at detecting sequential patterns in order flow. these models can remember what happened sixty minutes ago and use that context to predict what will happen in the next five minutes.
there is a specific feature that most people overlook when they try to build these models and it has nothing to do with price. i call it the whale liquidation ratio and it is one of the five loops i use to keep the model focused on high impact events.
while most people are trying to predict price we are actually trying to predict the probability of a specific move based on liquidation intensity. if you only look at price indicators like rsi or macd you are just doing regular technical analysis which has no edge.
i focus purely on usd liquidation volume and the velocity of those liquidations to see when a cascade is building up. the reason this is better than ai is because we are handcrafting the features to look for specific market anomalies.
one of the biggest mistakes i made early on was faking my data with forward filling or back filling to make the charts look pretty. in the real world that creates data leakage that will blow up your account the moment you go live.
you have to be willing to drop rows with missing information and keep your data as raw and honest as possible. if your model shows ninety five percent accuracy it is almost certainly overfit and you are in for a rude awakening.
there is a rumor that wall street is trying to suppress this type of information because it allows the retail trader to compete on their level. i have seen my own content being pulled because it shows too much of the internal plumbing of the markets.
i am not going to stop sharing this because code belongs to everyone and it is the only way to escape the trap of manual trading. if i do not show up one day it is because the powers that be have finally decided i am too much of a threat.
to build a proper lstm you need to split your data into eighty percent training ten percent validation and ten percent test sets. the validation set is critical because it tells you when[...]
Offshore
Moon Dev Beyond Generative AI: Building a High-Precision LSTM Trading Bot from Scratch most people are chasing the shiny objects of generative ai while the real wealth is hidden in the machine learning engines that actually power them. i am going to show…
to stop training before you start memorizing the noise.
you also need to scale your features using a standard scaler or a min max scaler so the model does not get confused by large numbers. if you feed a model usd volume of five hundred thousand next to a time stamp of twelve it will think the volume is infinitely more important.
the rbi system is how i stay disciplined in this game which stands for research backtest and implement. you cannot just work harder at trading to get better because trading requires patience and a total lack of greed.
most humans are not built for this which is why we must remove the human element entirely through automation. i am not a guy who is afraid to die on a treadmill and if we get on one together you are getting off first.
discipline is doing what you hate to do but doing it like you love it every single day. i spent a decade being scared to code because i thought it was only for geniuses but i was wrong.
you just have to start with the first line and keep iterating until your bots are doing the heavy lifting for you. the prize is not just the money but the freedom from the emotional roller coaster of the charts.
i have documented every step of this journey and provided the roadmap so you do not have to spend years wandering in the dark. if you follow the heart and intuition you have they already know what you truly want to become.
everything else is secondary to the pursuit of your own edge in a world that wants you to follow the pack. stay hungry and stay foolish because that is how we win against the incumbents who have grown fat and lazy.
the fully automated life is waiting for anyone who is willing to learn the language of the machine. through code we can finally achieve the equality that the financial world has tried to keep for itself.
once you close the loop on your first profitable bot you will never want to look at a manual chart again. keep pushing and keep coding until the systems are working for you instead of you working for the systems
tweet
you also need to scale your features using a standard scaler or a min max scaler so the model does not get confused by large numbers. if you feed a model usd volume of five hundred thousand next to a time stamp of twelve it will think the volume is infinitely more important.
the rbi system is how i stay disciplined in this game which stands for research backtest and implement. you cannot just work harder at trading to get better because trading requires patience and a total lack of greed.
most humans are not built for this which is why we must remove the human element entirely through automation. i am not a guy who is afraid to die on a treadmill and if we get on one together you are getting off first.
discipline is doing what you hate to do but doing it like you love it every single day. i spent a decade being scared to code because i thought it was only for geniuses but i was wrong.
you just have to start with the first line and keep iterating until your bots are doing the heavy lifting for you. the prize is not just the money but the freedom from the emotional roller coaster of the charts.
i have documented every step of this journey and provided the roadmap so you do not have to spend years wandering in the dark. if you follow the heart and intuition you have they already know what you truly want to become.
everything else is secondary to the pursuit of your own edge in a world that wants you to follow the pack. stay hungry and stay foolish because that is how we win against the incumbents who have grown fat and lazy.
the fully automated life is waiting for anyone who is willing to learn the language of the machine. through code we can finally achieve the equality that the financial world has tried to keep for itself.
once you close the loop on your first profitable bot you will never want to look at a manual chart again. keep pushing and keep coding until the systems are working for you instead of you working for the systems
tweet
X (formerly Twitter)
Moon Dev (@MoonDevOnYT) on X
Beyond Generative AI: Building a High-Precision LSTM Trading Bot from Scratch
most people are chasing the shiny objects of generative ai while the real wealth is hidden in the machine learning engines that actually power them. i am going to show you exactly…
most people are chasing the shiny objects of generative ai while the real wealth is hidden in the machine learning engines that actually power them. i am going to show you exactly…
Offshore
Photo
Moon Dev
starting soon
our zoom call is getting started in about 30 minutes
you are not going to want to miss this one
we are diving deeper into the war trade
see if there are still tickets left
join here https://t.co/JbJdIbW2p9
moondev
tweet
starting soon
our zoom call is getting started in about 30 minutes
you are not going to want to miss this one
we are diving deeper into the war trade
see if there are still tickets left
join here https://t.co/JbJdIbW2p9
moondev
tweet
Offshore
Video
Moon Dev
If you are currently trading by hand you are no different than a slot player
They actually have better odds https://t.co/LGanBgyivC
tweet
If you are currently trading by hand you are no different than a slot player
They actually have better odds https://t.co/LGanBgyivC
tweet
Moon Dev
Beyond Manual Trading: Building a Cross-Exchange Arbitrage System with AI Agents
i used to think that the only way to beat the market was to have a faster terminal or more inside info than the guys in the suits. then i realized that the real secret is hiding in plain sight within the code of decentralized exchanges where prices move at the speed of chaos.
most people are chasing the next shiny coin while the real alpha is sitting in the price gaps between two exchanges that can't stay in sync. i'm going to show you how to build a system that hunts these gaps for you but first you need to understand the hidden mechanism that triggers the biggest arbitrage opportunities in crypto.
i believe that code is the great equalizer because through losing money with liquidations and over trading i knew i had to automate my trading. i spent hundreds of thousands on developers in the past because i didn't think i could do it myself but now the bots do the heavy lifting.
with bots you must iterate to success and that's exactly what we're doing by connecting systems like habachi and pacifica. these platforms offer unique point systems that lead to massive airdrops but most traders are too lazy to automate the interaction.
the real magic happens when you realize that liquidations aren't just a sign of a bad trade but a signal for a profitable entry. if you can build a bot that watches for a half million dollar liquidation it can act as a trigger to enter a position before the rest of the market reacts.
most people fail because they try to build everything from scratch without realizing there is an ai agent ready to write ninety percent of the code for you. i used to spend weeks debugging simple order functions until i started using agentic coding tools to handle the repetitive plumbing.
you need to build specific functions to handle buys and sells and getting your current position in real time. if your bot doesn't know exactly where it stands it will just keep adding to a losing position until your account is empty.
there is a specific trap with account wide leverage that will blow up your entire bankroll if you don't account for it in your logic. most exchanges use a global setting which means if your bot thinks it is using two times leverage but the exchange is set to fifty you are one small wick away from zero.
i found out the hard way that manually adjusting your leverage on the exchange interface is the only way to stay safe when the api calls fail. once you have the leverage locked in you can start looking at the spread between exchanges to find your delta neutral edge.
the goal is to find an exchange with less capital and match the size on the secondary exchange to stay perfectly balanced. i use a logic that looks at ninety eight percent of the smaller balance to ensure every trade is fully covered and risk is minimized.
the industry wants you to believe that this level of automation is only for hedge funds with millions in funding. they never expected a solo developer with a belief in open code to build the same systems that they use to front run the retail crowd.
the airdrop season is a gold mine for anyone who knows how to keep their volume high without actually taking directional risk. by trading back and forth in a delta neutral state you are essentially mining points for free while others are gambling on the next meme coin.
i've seen traders spend their whole lives looking for a winning strategy only to give it all back because of emotional fatigue. automation is the only way to remove the human element and stay disciplined when the market gets rowdy and unpredictable.
there is a rumor that certain exchanges are suppressing api access to prevent retail from taking all the arbitrage profit. this is why being connected to multiple dexes is critical so you can pivot your capital the moment one server starts giving you five hundred two errors.
building these systems isn't just about the money but about the freedom to walk [...]
Beyond Manual Trading: Building a Cross-Exchange Arbitrage System with AI Agents
i used to think that the only way to beat the market was to have a faster terminal or more inside info than the guys in the suits. then i realized that the real secret is hiding in plain sight within the code of decentralized exchanges where prices move at the speed of chaos.
most people are chasing the next shiny coin while the real alpha is sitting in the price gaps between two exchanges that can't stay in sync. i'm going to show you how to build a system that hunts these gaps for you but first you need to understand the hidden mechanism that triggers the biggest arbitrage opportunities in crypto.
i believe that code is the great equalizer because through losing money with liquidations and over trading i knew i had to automate my trading. i spent hundreds of thousands on developers in the past because i didn't think i could do it myself but now the bots do the heavy lifting.
with bots you must iterate to success and that's exactly what we're doing by connecting systems like habachi and pacifica. these platforms offer unique point systems that lead to massive airdrops but most traders are too lazy to automate the interaction.
the real magic happens when you realize that liquidations aren't just a sign of a bad trade but a signal for a profitable entry. if you can build a bot that watches for a half million dollar liquidation it can act as a trigger to enter a position before the rest of the market reacts.
most people fail because they try to build everything from scratch without realizing there is an ai agent ready to write ninety percent of the code for you. i used to spend weeks debugging simple order functions until i started using agentic coding tools to handle the repetitive plumbing.
you need to build specific functions to handle buys and sells and getting your current position in real time. if your bot doesn't know exactly where it stands it will just keep adding to a losing position until your account is empty.
there is a specific trap with account wide leverage that will blow up your entire bankroll if you don't account for it in your logic. most exchanges use a global setting which means if your bot thinks it is using two times leverage but the exchange is set to fifty you are one small wick away from zero.
i found out the hard way that manually adjusting your leverage on the exchange interface is the only way to stay safe when the api calls fail. once you have the leverage locked in you can start looking at the spread between exchanges to find your delta neutral edge.
the goal is to find an exchange with less capital and match the size on the secondary exchange to stay perfectly balanced. i use a logic that looks at ninety eight percent of the smaller balance to ensure every trade is fully covered and risk is minimized.
the industry wants you to believe that this level of automation is only for hedge funds with millions in funding. they never expected a solo developer with a belief in open code to build the same systems that they use to front run the retail crowd.
the airdrop season is a gold mine for anyone who knows how to keep their volume high without actually taking directional risk. by trading back and forth in a delta neutral state you are essentially mining points for free while others are gambling on the next meme coin.
i've seen traders spend their whole lives looking for a winning strategy only to give it all back because of emotional fatigue. automation is the only way to remove the human element and stay disciplined when the market gets rowdy and unpredictable.
there is a rumor that certain exchanges are suppressing api access to prevent retail from taking all the arbitrage profit. this is why being connected to multiple dexes is critical so you can pivot your capital the moment one server starts giving you five hundred two errors.
building these systems isn't just about the money but about the freedom to walk [...]
Offshore
Moon Dev Beyond Manual Trading: Building a Cross-Exchange Arbitrage System with AI Agents i used to think that the only way to beat the market was to have a faster terminal or more inside info than the guys in the suits. then i realized that the real secret…
away from the screen while the bots fight for every cent. when you finally see the orders filling on both sides with perfect precision you realize that the old way of trading is dead.
the final piece of the puzzle is the cooldown period which prevents your bot from over trading during periods of high volatility. if you don't give the market time to breathe your bot will just eat itself alive through fees and slippage.
most people will read this and go back to their manual charts because they are scared of the learning curve. the reality is that the tools are already here and the only thing standing between you and a fully automated life is the willingness to start.
through code we can finally achieve the equality that the financial world has tried to keep for itself. it is time to stop being the exit liquidity for the big players and start being the one who provides the efficiency that the market demands.
once you have your first profitable bot running you will never look at a price chart the same way again. the systems are working for me now instead of me working for them and that is the ultimate goal for anyone who values their time and their freedom
tweet
the final piece of the puzzle is the cooldown period which prevents your bot from over trading during periods of high volatility. if you don't give the market time to breathe your bot will just eat itself alive through fees and slippage.
most people will read this and go back to their manual charts because they are scared of the learning curve. the reality is that the tools are already here and the only thing standing between you and a fully automated life is the willingness to start.
through code we can finally achieve the equality that the financial world has tried to keep for itself. it is time to stop being the exit liquidity for the big players and start being the one who provides the efficiency that the market demands.
once you have your first profitable bot running you will never look at a price chart the same way again. the systems are working for me now instead of me working for them and that is the ultimate goal for anyone who values their time and their freedom
tweet
X (formerly Twitter)
Moon Dev (@MoonDevOnYT) on X
Beyond Manual Trading: Building a Cross-Exchange Arbitrage System with AI Agents
i used to think that the only way to beat the market was to have a faster terminal or more inside info than the guys in the suits. then i realized that the real secret is hiding…
i used to think that the only way to beat the market was to have a faster terminal or more inside info than the guys in the suits. then i realized that the real secret is hiding…
Offshore
Video
Moon Dev
today i put gpt 5.4 to the ultimate trading bot test for ww3..
cooked up a bot and shared all the code https://t.co/rhdVrCIyve
tweet
today i put gpt 5.4 to the ultimate trading bot test for ww3..
cooked up a bot and shared all the code https://t.co/rhdVrCIyve
tweet