YT YouTube2Text

YouTube2Text → Transcripts → I Built an INSANELY Profitable AI Hedge Trading Bot (Full Guide)

I Built an INSANELY Profitable AI Hedge Trading Bot (Full Guide) — Transcript

by Trading with DaviddTech · 5,536 words · 774 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Allow me to introduce myself. I am
0:02Jarvis.
0:02>> Hey there Jarvis, how are my strategies
0:04going?
0:06>> All five strategies look healthy and
0:08nothing is flagged. The strongest is
0:11momentum BTC, net up 36.6%
0:15over the last 30 days.
0:16>> That was my real AI hedge funds running
0:20from this little computer here. It was
0:22built by Claude, managing real money,
0:25multiple strategies, one single brain,
0:28and it even sends me data reports to my
0:30phone. Today I'm going to be showing you
0:32a step-by-step how you can build the
0:34back end so you can have your own AI
0:36hedge funds step-by-step. And the fact
0:38of the matter is you don't need to know
0:39how to code and you don't need to be a
0:41strategy genius. And guys, if you want
0:43to skip ahead of everybody else and
0:45learn how to build your own full AI
0:47hedge fund that has been fully forward
0:49tested, you can find the links to my
0:52school down in the description. This AI
0:54hedge fund is not one brilliant tool. It
0:57is a mix of tools all under one single
1:01roof. And by the end of this video, I'm
1:03going to be sharing with you a chance
1:05that I hope will help you change your
1:06trading forever. I've been backtesting
1:09strategies here on YouTube for over five
1:12years. I've backtested over 2,000
1:14strategies and have previously been the
1:17top trader on multiple exchanges.
1:19Everything we talk about today will be
1:21linked in the description and most of it
1:23is for free. Okay, so the full setup is
1:26comprised of six steps to get your AI up
1:30and running. We're going to be building
1:32strategies, risk allocation, and
1:34reporting so you have everything set up
1:37inside of one single AI agent. Part one,
1:40we're going to be going through the
1:41tools and the cost. I'm going to do
1:43everything transparently and where
1:45possible we're going to be using free
1:47tools. In part two, we're going to be
1:49giving the AI agent a brain and we're
1:51going to install the tools all together
1:53so you can go away and build strategies,
1:55discover strategies, backtest
1:57strategies, and actually live trade
1:59them. In part three, we're going to give
2:01that brain a little bit more knowledge
2:02so it can go and look for strategies
2:04that actually work. So, you don't spend
2:06weeks or months at just waiting for
2:08strategies that were going to fail most
2:10probably anyway. In part four, we're
2:12going to be connecting to the broadcast
2:15so this AI hedge fund can actually make
2:17trades. Part five, we're going to be
2:19giving the AI agent hands so that it can
2:22actually activate and deactivate
2:24strategies when they start losing their
2:26edge or start becoming more profitable.
2:29This one is essential because we don't
2:30want to spend hours in front of our
2:31screens. AI can keep an eye on the
2:33strategies to make sure that we still
2:35have an edge. In part six, we're going
2:37to make it send you a report daily so
2:39you know exactly what is going on. And
2:42at the end, we'll just do some real
2:44talk. We'll talk about really what how I
2:46use this in my day-to-day trading and
2:49I'll try to give you a couple of tips so
2:50you don't just walk away with a loaded
2:52weapon and no safety. And guys, if
2:54you're finding value in this video and
2:56you are building along with me today, if
2:58you don't mind just doing a quick favor,
2:59don't forget to give the video a good
3:00old like and don't forget to subscribe.
3:02I'm going to be doing videos every
3:03single week about building your AI
3:06agents and how you can actually trade
3:08with them to increase your edge and
3:10become more profitable. Right, let's get
3:11over to my computer and have a look at
3:13the tools and their actual costs. Okay,
3:16here we are on my computer. Step one,
3:18the tools that we actually need and
3:20their costs. We are going to need a
3:22Claude Desktop. There are multiple
3:24levels of a pricing including a free,
3:27pro, and max. Personally, I think you do
3:30actually need a pro subscription if
3:32possible because we're going to be using
3:34Claude Code and Claude Code is essential
3:37for this product unless you're using
3:39something like Codex from ChatGPT, which
3:41will work equally as well. So, we will
3:44need to go over to claude.com/download
3:46and download the app on your computer
3:49just like this. We're going to click on
3:51code and we're going to set up a brand
3:52new session like this, which will allow
3:54us to start building this AI hedge fund
3:57without actually coding a line of code.
3:59Next thing we're going to need is a
4:01TradingView account. Now, TradingView is
4:04our charting platform that looks a
4:06little bit like this, and we're going to
4:07be building our strategies with our AI
4:09hedge fund that looks like this, and
4:12eventually making them at live trade.
4:15Now, TradingView can be a little bit
4:16more costly than the actual AI agents
4:19themselves, but there are free packages.
4:22The problem with the free packages is
4:23you won't be able to live trade, but you
4:25will actually be able to follow along
4:26with the trades on the charting
4:28platform. If you actually want to
4:30manually trade these things yourselves.
4:32So, an essential package is fine unless
4:34you want more data, which you can get at
4:36the $29.99 level. So, those are the two
4:39platforms that we're actually going to
4:41be using, and now we're going to be
4:42looking at the tools that we actually
4:44give to our AI agents, their costs, and
4:47how we actually use them. Guys, I would
4:50love to know, would you feel comfortable
4:52giving your real cash to an AI agent?
4:55Let me know in the comments down below.
4:57Personally, I've been algo trading for
4:59over 5 years, and love the fact that I
5:01don't have to sit in front of screens.
5:03So, once we have installed Cloud Code,
5:06we will need what we call MCP servers.
5:08Now, MCP servers are described as USB
5:11keys. They are the connector that give
5:13tools to your actual AI agent so that it
5:16can do things like backtest strategies.
5:19As you can see, this is a community
5:21project, and we have backtested almost
5:24600,000 strategies on this platform, and
5:28it is 100% free to use. Now,
5:31essentially, that is all of the tools
5:32that we actually need. I'm also going to
5:34be giving a couple of GitHub repos that
5:37you can actually get for free, which I
5:38made myself, but I'll show you them as
5:41we go along, and I'll show you exactly
5:43what they do. Okay, so let's get
5:45started. Hopefully, you've already got
5:47Claude a coded downloaded and you have a
5:50Claude coded open on your desktop. We
5:52will need to open a new folder just
5:55here. I'm calling my AI hedge fund and
5:58we're going to go over and install the
6:00MCP servers and these tools that we
6:02actually need. Remember, MCP servers
6:05aren't just tools for Claude to be able
6:07to back test, take live trades, and do
6:10everything that an AI hedge fund should
6:12do. So, the first tool that we're going
6:14to be installing is Trading Kit. Trading
6:16Kit is free of charge and as you can
6:19see, we've actually hit the 600,000
6:23back test since I'm live here on
6:25YouTube. So, the first thing we're going
6:26to do is go over to my account on the
6:28top here. Once you're logged in, you'll
6:30be able to see this code here. We're
6:32just going to copy that over and take it
6:34over to Claude and type, "Hey there,
6:37buddy. Could you kindly install this
6:39globally for me so it'll work on Claude
6:41coded and in Claude chat desktop." You
6:44will hit enter. I've already got it
6:46installed, so I'm not going to reinstall
6:48it and it will go away and install the
6:51MCP server. Once that's installed, it
6:53will ask you to authenticate. You just
6:55go over back over to your account on
6:57tradingkit.com,
6:59click on regenerate key, copy that key,
7:01and just paste it into Claude and it
7:04will save it into a file for you so you
7:06can do your back testing. Once that is
7:09done, you'll be able to say something
7:10simple like, "Hey there, buddy. Could
7:12you just test a very simple strategy on
7:16BTC USDT to make sure that the back test
7:19that is actually working." And boom,
7:21within a couple of seconds, we have a
7:23fully back tested strategy. Minus 22%
7:26isn't very good. 88 closed trades. It
7:29was a back tested as a simple SMA cross
7:32on a BTC USDT. So, very quickly, that
7:36has already built a very, very powerful
7:38system where we've given access tools to
7:41Claude so he can back test strategies.
7:43So, now when you're out on the run and
7:44you can think of a strategy, you can
7:46actually back test that idea just in a
7:48plain English. So, absolutely amazing.
7:51And so, now here is where the real fun
7:53comes. We're going to be giving these
7:55tools to an agent so that you can
7:57actually back test strategies, build
7:59upon those strategies, find the best
8:01settings, and actually even take them as
8:04far as live trading. It's going to work
8:06in an environment using these cool
8:09skills here which I've put on GitHub.
8:12Now, these skills may look pretty
8:14simple, but this is the full plan. The
8:18skills turn that back testing tool into
8:21a full AI hedge fund, where the AI
8:24agents first of all onboard you and ask
8:26you all the questions needed. Then, they
8:29go through a researching campaign. They
8:32search the markets, work out what kind
8:34of strategies will actually work on
8:35those markets, come up with ideas for
8:38trading looking at things like trend
8:41following, market structure, cross
8:43markets, and different ideas. The AI
8:46agent will then go away and do some
8:48indicator research looking for full
8:51systems, trend momentum indicators,
8:54volatility indicators, volume
8:55indicators, you name it, it has gone
8:57away and found those indicators. Now,
8:59for this part here, we will be
9:01installing another MCP server, which
9:03gives your MCP server access to the
9:06indicators on TradingView. So, what this
9:09means is the list of indicators that you
9:11can find here, hundreds and thousands of
9:14indicators that haven't been downloaded
9:17by thousands of people, actually can now
9:20be put into strategies. A lot of them
9:23are open source, and we can use those
9:25indicators to create a full strategy.
9:28Next, it will create an architecture
9:29around those strategies understanding
9:31where the long and short entries are,
9:34creating exit conditions, etc. All of
9:38that plan will then be passed on to a
9:40Pine Script engineer who will build a
9:43Pine Script strategy ready for us to
9:46upload to TradingView. But, when I say
9:49us, I don't mean us because we don't
9:51actually do anything manually here. This
9:53agent then passes it on to another agent
9:55that's actually going to backtest that
9:57strategy using the famous tradingkit.com
10:01MCP server. We will backtest the
10:04strategy on multiple time frames, also
10:07using in and out of sample research, a
10:10full validation framework. So, we're not
10:12just throwing ideas at the wall. It's
10:14going to download all of the information
10:16and create you a dashboard that will
10:18look like this, where you'll be able to
10:21see all of those strategies that it has
10:23built, and it will highlight the ones
10:25that it is actually interested in, and
10:27then put them through a verification
10:29system. That verification system will
10:32make sure that the code is actually
10:33running runnable on TradingView, and
10:36goes through rules like it doesn't hit a
10:38maximum drawdown. It will ask you in the
10:41onboarding process the minimum net
10:43profit that you're interested in, and
10:45things like that just to make sure that,
10:47like I said earlier, we're not just
10:49trading rubbish. We are looking for
10:51something that actually is going to give
10:53us a real edge long-term on the market.
10:56Once it's gone through that, it's going
10:57to go through a robustness validation,
10:59and finally a judge. So, as you can see,
11:02all of the strategies that I've been
11:04building over the past 5 years on
11:06TradingView have gone through all of
11:08these steps. It takes a month to build
11:11one single strategy as AI is going to be
11:14building it in the background for you,
11:16so you don't have to lift a finger. It's
11:18finally going to put it into a forward
11:20testing period. What that means is that
11:23it will highlight on the control panel
11:25that it has built for you as forward
11:28testing. So, it's going to go away and
11:30make sure that the trading strategy is
11:33working as it is. You will update those
11:35trading results every single day. Or if
11:38you don't have the compute power, it
11:39will do it once a week or whatever you
11:41decide in the onboarding process. This
11:43is a bit like a paper execution engine.
11:46It's just literally going to make sure
11:48that those fees and everything are in
11:50the bank test and make sure the equity
11:52curve continues to grow the way it
11:54should on data that it has never seen.
11:57Once that's all done, it's going to send
11:58you a report. It's going to tell you
12:00exactly how many indicators and
12:02strategies that it has backtested every
12:04single day. And then when you're ready,
12:06you'll be able to choose the ones that
12:08you actually want to for it to live
12:10trade and set up our final MCP server
12:14which will allow this full system to
12:16keep an eye on those strategies. Now
12:19last year in December, I got a bit ill
12:21and I went through quite a lot of
12:23drawdown on my account. Now I could have
12:26stopped all of my bots, but I didn't
12:28keep to all of the rules. This MCP
12:31server will allow you to set rules, tell
12:34your AI when you want to stop a bot. For
12:37example, if you feel as though it's
12:38losing an edge on the equity curve or it
12:41goes through a maximum drawdown in one
12:43day, you can actually tell your AI to
12:45turn off those strategies. Okay, now
12:47that we know exactly how it works, let's
12:49get this thing installed. So we're going
12:51to go over, we're going to copy the link
12:53which you'll be able to find all of
12:54these links once again down in the
12:56description. You'll be able to find a
12:57resource link which will take you to a
12:59page where you'll be able to download
13:01all of these resources. Right, okay,
13:02we're going to copy that. We're going to
13:04go back over to Claude and we're going
13:06to go, "Hey there, buddy. Could you
13:08please download this GitHub into this
13:11same folder?" We're going to hit that
13:14there. I actually already have it
13:15downloaded. As you can see, I already
13:17knew that it was actually that one
13:18there. "Hey there, buddy. Could you
13:19please get a full understanding of the
13:21actual skills, the actual Claude.md file
13:25and make sure you understand what we're
13:27actually going to be doing. Once it's
13:29done, it's going to fully understand
13:30everything we spoke about just here.
13:32It's going to ask you to do an
13:33onboarding and it's going to ask you to
13:35set up a couple more of the MCP servers.
13:37Now, to do that, we're going to go over
13:38to the two other GitHubs and we're just
13:41going to copy that and we're going to
13:42paste that here as well. Hey everybody,
13:44could you please install these two
13:46GitHub MCP servers globally? And boom.
13:50Making sure that they are available in
13:52both the code and also in Claude the
13:55desktop. And we can hit enter. That's
13:57going to go away and install them.
14:00You may have to restart Claude
14:01afterwards. I already had them
14:02installed, so I should imagine it would
14:03not be a problem. Okay, so let's have a
14:06just a very quick check-in. So, what
14:08we've done so far is we've installed
14:10Claude. We've given Claude its rules and
14:12its baseline, so it knows what its job
14:14is. And we've given Claude its tools, so
14:17that it can actually backtest and find
14:19that those wonderful strategies with an
14:22edge which will hopefully make us more
14:23profitable on the markets. We've also
14:26finally given it a tool, so that it can
14:27actually manage your strategies once
14:29you've decided to actually take them
14:31live. Our next step is to launch Claude,
14:34so that we'll start a loop on a cron.
14:36What that means is that it has a
14:38schedule and it's going to go in and
14:40start building strategies every 15
14:43minutes. You can change that to 1 hour.
14:45You can change it to whatever you like.
14:47I personally like a low time frame
14:49because it makes more strategies. But,
14:51if you have the smaller package on
14:54Claude's subscriptions, I'm going to set
14:56it to something more like 1 hour, even 4
14:59hours, which will eventually help you to
15:00find those strategies. Claude will take
15:03you through a full onboarding process,
15:05where it should actually send that
15:07report to via Telegram. It will also ask
15:10you for some things like data from your
15:12TradingView account, so that you can
15:14actually go away and turn on and off
15:17your strategies. Now, if you don't want
15:19to share that data with it. you don't
15:21have to do that part, and everything
15:23else doesn't need any data from
15:26TradingView as it is all using MCP
15:28servers. Right back to our AI hedge fund
15:31repo on GitHub. We're going to go over,
15:34and we're going to go over to this file
15:36just here. This is the the build prompt
15:39where we have all of our loops that I
15:40was talking about earlier. We're just
15:42going to go over here, and we're going
15:43to copy that raw file.
15:46And we're going to go back to Claude.
15:47We're going to paste it in there, and
15:49we're going to hit enter, and it's going
15:51to start building strategies. Now, the
15:53first thing that it will do is it will
15:55go away and build a local dashboard like
15:58this one here where it's going to be
16:00recording the backtesting results to a
16:03dashboard so that you can find them
16:05easy. It does also record them to
16:08TradingKit as well, but there are
16:10600,000 strategies on there, so it will
16:13be a little bit hard for you to find
16:14your own ones. So, I set it up so that
16:16it builds your own dashboard. There you
16:18can improve on this dashboard, add
16:20features, delete features, and all of
16:23that kind of thing. It's just a base for
16:25you to get started. The first step is is
16:28it's going to look for strategy ideas.
16:31It's going to go away and build those
16:33strategies, backtest those strategies,
16:34and find them here. When it finds
16:36something that is potentially worth
16:38trading, it will put a star just here
16:41and put it into the candidate strategy.
16:44This will then give you the backtest
16:46results, all of the optimizations that
16:48it has done, and all of the trades as
16:51well. If you scroll down a little bit,
16:53you'll be able to see all of the
16:54settings it used, and you'll be able to
16:56find a way to copy those best settings
16:59and the Pine Script just here. Now,
17:01before we go any further, just a quick
17:03heads-up. Next thing it's going to do is
17:05put it into forward testing. Forward
17:07testing means that it's going to freeze
17:09that strategy and come back and backtest
17:11it every single day. Personally, I'd
17:14leave the strategy for at least 20
17:16trades, or generally speaking around 3
17:19months depending on the actual time
17:21frame that you optimize that strategy.
17:23This gives that the strategy time to
17:25actually evolve, see data that it may
17:27have never seen in the back test, and
17:30we'll be able to judge for ourselves
17:32whether we want to trade that strategy.
17:34Now, I've been doing this for 5 years on
17:39my platform called Strategy Factory AI,
17:41where you can see the back test in green
17:43and the live trading in pink for over
17:46400 strategies. Now, if you want to skip
17:49ahead of all of the forward testing, you
17:52can get access to all of those
17:53strategies on my school.com/davidtech
17:57site, where I'm actually building a
17:59community and we're all working together
18:01to try and find the best strategies that
18:03using AI. You'll also learn how to build
18:07your own Jarvis like this, will take
18:10care of all of your strategies and act
18:12like an AI hedge fund. You'll be able to
18:14find links to that in the first pinned
18:16comment or in the actual description.
18:19Okay, now we're going to have to skip
18:21ahead in time. Imagine that we are 3
18:24months later. Our strategy has actually
18:27performed as we expected in trades. It
18:30is profitable, hasn't made a new huge
18:33max drawdown, and has actually traded in
18:36the forward testing time. We can
18:37actually decide ourselves to deploy this
18:40strategy. Now, unfortunately, this is
18:42the only manual part of the whole of the
18:45system because I haven't found a way
18:46personally to automate it completely
18:48just yet, but I am doing my best to find
18:51a way. So, we're going to go back to our
18:53control panel. We are very happy with
18:55the back test and forward testing
18:57periods we've gone through. We can see
18:59have 110 trades. You can even make your
19:02equity curve look like this just by
19:03asking Claude to change the colors on
19:06the forward testing periods. Now, it's
19:08time to import this over to our
19:11TradingView account. Why do we do this?
19:13It's because we're going to be using
19:14TradingView to execute the trades on our
19:17exchange directly. So, we don't have to
19:19spend hours looking at charts.
19:22Everything is going to be automated by
19:23TradingView and our AI agent. So, first
19:26things first, we go all the way down
19:28just here. We're going to go to the
19:30bottom and we're going to copy the Pine
19:32Script code. We're going to go over to
19:35Pine TradingView. We click on the Pine
19:37Script button just here. Create new
19:40strategy. Paste that strategy just in
19:43there and save and add to chart. Okay,
19:46so on the backtest uh we should see
19:48something very, very similar to what we
19:50saw on our control panel just here. We
19:53have a great backtest results. We know
19:56that it's been forward tested for at
19:57least 3 months. Next thing we need to do
20:00is connect this over to our exchange.
20:03Now, to do this, I'm going to be using a
20:04tool called trigger.trade. But, inside
20:07of TradingView, you don't have to
20:08actually do this step. You can actually
20:10connect your brokers directly inside of
20:13TradingView itself. I'm just going to
20:16show you another tool, which is actually
20:18free as well. It allows you to connect
20:20to any of these exchanges. If you want
20:22to do something with Forex or gold like
20:24this one just here, I suggest you just
20:26connect it in directly in with the
20:29TradingView brokers. Okay, so we're
20:31going to go over to triggertrade.com and
20:34log in. As I said earlier, it is a free
20:36tool. You just log it in and it will ask
20:39you to set up your exchange. I'm going
20:41to use just the Telegram bot here, so it
20:43will alert today, but these steps are
20:45exactly the same. Now, one thing that is
20:48very cool about TriggerTrade is it is
20:50anonymous. None of the API keys or
20:53secrets are ever stored on the server.
20:56Everything is encrypted in a JSON file
20:58that I sent over to you and then you add
21:00that to your TradingView. So, back to
21:02the TriggerTrade website, you can
21:04actually install this directly inside of
21:06Claude by going here and installing
21:09their Trigger Trade skills by literally
21:12copying the link over to Claude itself
21:15and asking it to generate a code. I'm
21:18going to take you step-by-step through
21:19what Claude would have actually done
21:21just because it helps you to understand
21:23the full process. So, we're going to
21:24click on Telegram just here, press next.
21:27They're going to ask us for our Telegram
21:29ID and a bot for us to send it. It gives
21:32you the David Tech bot by default at the
21:35top here. You can just click here. You
21:37can just press start. As you can see,
21:39I've already had tons of trades already
21:41sent through to mine. But, if you press
21:43start, you should end up with a blank
21:45screen. Next thing you need to do is go
21:47and get your UID. You click on the user
21:50info bot, open that in Telegram as well.
21:53It will give you back your ID. We just
21:55copy that to our clipboard, go back over
21:58and paste it in here. Press continue.
22:01You can test the connection if you like.
22:03And then you can see I've just had a
22:04test connection here and it's sent you a
22:06sample of what it will be sending. We're
22:09going to go back over to Trigger Trade,
22:11press continue. It's going to ask you to
22:13set up the trading strategy if you want
22:15manually or you can just follow
22:17everything off the charts. I personally
22:19prefer following everything on my
22:20charts. It's going to keep going along.
22:22This is going to give you a JSON file,
22:26which you're literally going to copy
22:27over to TradingView. Go over to
22:30TradingView. We're going to set up an
22:32alert. So, the alerts on TradingView
22:35allow us to a webhook. The webhook being
22:38a third party that will connect you
22:40either over to your exchange or over to
22:42your broker or in our case over to
22:44Telegram. To activate the alerts, we
22:46click the three buttons here. We press
22:49add alert BTC. We're going to change
22:52this first thing to fields only,
22:54otherwise we'll end up with two trades.
22:57We're going to go over to the message
22:58just here and we'll paste in that
23:01wonderful JSON code that we just got.
23:03Give it a name, test. Apply. We're going
23:07to go over to the notifications and
23:10we're going to copy this webhook URL
23:12just here. Paste it in underneath that
23:15webhook URL and press apply. Press
23:18create. And Bob's your uncle, you are
23:21now trading on your exchange or
23:23receiving alerts directly through
23:26Telegram. Now, there is one step that
23:27you will have to do if you want to set
23:29up with your broker and that is set up
23:31your API key and secret. It is very
23:34simple. It will encrypt them and then
23:36put them directly onto your TradingView
23:38so you're in control of everything.
23:40Okay, so now we've managed to connect
23:42our bots over to our exchange. It's
23:44taking live trades for us, which is
23:47absolutely amazing. We've come so far
23:49from being a manual trader staring at
23:51charts all day, every day, now to full
23:54automation, discovering of strategies,
23:56and actually live trading. But, there is
24:00one big problem. No matter how much
24:02people would tell you that these
24:03strategies or any strategies will last
24:06forever, they will trade on every single
24:08coin or any asset, this is just not the
24:11truth. Trading strategies work for a
24:15certain amount of time before they start
24:18to lose their edge. Now, this is
24:20something I had to learn the hard way
24:22myself and unfortunately have lost money
24:25to trading strategies that did extremely
24:27well. And if I would have stuck to my
24:28rules myself, probably would have lost a
24:31lot less. So, this is going to be the
24:33final part of your AI hedge fund that is
24:36actually trading for you and that is
24:38giving it the power to be able to turn
24:40off trading strategies when they go
24:42against your rules. Couple of rules that
24:44I put in place myself is maximum
24:47drawdown over a certain period, whether
24:49that's a week, day, or even a month. You
24:52have to set those rules up yourself.
24:54Personally, I also asked Claude code to
24:57code Bollinger Bands around my actual
25:00equity curve. And if my equity curve
25:02hits the lower Bollinger Band, so a
25:04standard deviation, I asked it to turn
25:06off those bots that are actually
25:08trading. So, this is the final skill
25:11that we're going to set up. This skill
25:13is actually an MCP server, so it's given
25:16us the tools to be able to turn off a
25:18strategies. So, as you can see, once
25:20it's installed, you can prompt things
25:22like pause every single BTC strategy, or
25:26even create me an alert on RSI using the
25:3015-minute time frame webhook to my bot.
25:34So, very, very powerful. Imagine now
25:36giving this to your AI hedge fund and
25:39asking it to keep an eye on your
25:40strategies. This is where the real power
25:42comes in, and it's pretty simple. So,
25:45we're going to copy this link over to
25:48Claude. We're going to say,
25:50"I have live setups now on TradingView.
25:54I want you to keep an eye on those
25:55trades that are actually being pinged
25:57out. I've got an MCPC server above.
26:00You'll be able to see all of the live
26:02trades from the log. Keep an eye on
26:05those trading strategies that I'm
26:07running. And if a strategy hits 4% max
26:10drawdown in a single day, I'd like you
26:13to pause that strategy. If you feel a
26:15strategy is actually losing its edge
26:17with some tools that you have actually
26:19built, please pause that strategy." And
26:22boom, there you have it. It is now going
26:24to go away. It has full control of your
26:27alerts on TradingView. So, technically
26:30speaking, so Claude can now turn these
26:32on and off depending on its max drawdown
26:34in a day, or whatever rules you set in
26:38to make sure that it is not losing its
26:40edge. Now, for me, this one here has
26:42been an absolute game changer. I've been
26:45trying to build this actually on my
26:46website before, but now it's integrated
26:49directly in TradingView, which makes my
26:52whole life so, so much easier because I
26:55no longer have to worry about trading
26:57strategies failing because I know that
27:00Claude has got my back. Another very
27:02cool thing you can do is also ask Claude
27:05to go away and use the Trader Dev plugin
27:09to send you directly alerts every single
27:11day with all of the effects and
27:14backtested strategies, which just gives
27:16you a very quick overview of how your AI
27:18hedge fund is actually performing,
27:21whether it is actually finding
27:22strategies, whether or not it is
27:24actually making real money live on your
27:26accounts, or whether it is just spending
27:29tokens and crates and throwing ideas at
27:31the wall. And guys, don't forget you can
27:33get access to all of these prompts and
27:35all of the tools you'll be able to find
27:37the links down in the description. Okay,
27:39just a little bit of real talk here,
27:41guys, before I leave you to your day.
27:43These strategies are absolutely amazing,
27:46but you will find a big number will
27:48actually fail. It used to cost me
27:50thousands of dollars per strategy just
27:53to get a backtesting results or multiple
27:55days of me coding in a dark room trying
27:58to find these strategies. The AI agent
28:01will save you loads of time and loads of
28:03money coding strategies, but always
28:06remember to forward test strategies with
28:09the fees and slippage included. No
28:11matter how much AI you add to the
28:13situation, you will be very disappointed
28:15with the live trading results. Now, like
28:18I said, I've been doing this for 5
28:19years. I've backtested over 2,000
28:22strategies myself manually, and I'm
28:24currently running seven bots. That's
28:27just to show you how how much pain and
28:30time you have to go through to actually
28:32find something that actually is worth
28:34trading. On the bright side, we live in
28:36a wonderful world where we have AI that
28:39can do all of that hard work that I've
28:40been doing over the past 5 years and can
28:43probably do it in less than a week,
28:45which is absolutely insane, blows my
28:47mind, and I'm so happy that these tools
28:49are now available to everybody, whether
28:51you know how to code or not. And guys,
28:53finally, if you're interested in
28:55building the voice agent that I showed
28:57at the beginning, I'm going to be doing
28:58a live stream on how to build your own
29:00AI Jarvis. [snorts] You'll be able to
29:02find the links to that down in the
29:04description as well. Without further
29:07ado, guys, take care, keep safe, trade
29:08like a boss, and I will see you in the
29:10next one. Have a wonderful day. Take
29:11care. Goodbye.
29:12>> Come on, David, [music]
29:14and once again he's gone to code. A new
29:16strategy.
29:18He'll be back in just 5 minutes. RSI
29:20[music] and MACD. [singing]
29:22If he doesn't run into any bugs, there
29:25will be a new [music] strategy.

About this transcript

This page contains the full transcript of I Built an INSANELY Profitable AI Hedge Trading Bot (Full Guide) by Trading with DaviddTech, generated from the public captions YouTube serves with the video. The transcript has 5,536 words across 774 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.