YT YouTube2Text

YouTube2Text → Transcripts → Claude + IBKR API: Complete AI Trading Bot Guide

Claude + IBKR API: Complete AI Trading Bot Guide — Transcript

by Humbled Trader · 3,935 words · 579 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Imagine this. Every morning you wake up,
0:02your AI trading bot has already scanned
0:05the market and send you a list of
0:07gappers that fit your strategy. After
0:09the market opens, it triggers buy and
0:12sell orders for you autonomously. And it
0:15continues to run the cycle every 30
0:17minutes through all the trading day.
0:20This is the power of connecting
0:21Interactive Brokers API with Cloud Code.
0:24And I'll show you how to do that in this
0:26video. So, this is what you're going to
0:28learn. How to set up a broker API and
0:31connect Cloud Code to it safely. How to
0:34create your strategy with the help of
0:36AI. The full pipeline I've built from
0:39universe scan, decision loop,
0:41executions, and exit logic. And finally,
0:44setting up Telegram alerts like this, so
0:47you get notifications every 30 minutes
0:50from your bot. Last but not least, we're
0:52also going to build a dashboard [music]
0:54that tracks the R multiples per trade.
0:56In my last video, I showed you how I use
0:59Cloud and TradingView together to build
1:02a pre-market trade planning process with
1:04a high win rate and backtested strategy.
1:07But the one thing that TradingView and
1:09Cloud connection couldn't do was
1:11actually pull the trigger to buy and
1:14sell. So, in this video, I'm going to
1:16show you how to go from a backtested
1:18strategy to an actual trading bot that
1:21places its own orders, manages all the
1:23stops, takes partials, trails the
1:26winners, and force closes everything
1:29before the market closes. All of this
1:31pipeline is running on my computer with
1:33Interactive Brokers API and Cloud,
1:35[music]
1:36and I'm completely hands-off. I want to
1:38give you the same disclaimer as last
1:40time. AI is simply a multiplier. You
1:43still need to know how to trade first.
1:46I'm not a developer. I have no
1:47background in coding, but I am a trader.
1:50>> [music]
1:50>> So, if I can do this, then so can you.
1:53So, buckle up. Make sure to like this
1:55video and subscribe down below to see
1:57more step-by-step video lessons on using
2:00AI to optimize your trading. Okay, let's
2:03get started with setting up Interactive
2:05Brokers API. There's three things you
2:08need to install. I'll keep this one
2:10really short. The full steps and the
2:12details are on my blog. You can use that
2:14article as a video companion to follow
2:16along here and copy all the prompts from
2:19that blog. The link is down below.
2:21Before we do anything with Claude and
2:23AI, you need to have these things to
2:25have installed on your computer. First
2:27is Trader Workstation TWS, which is
2:30Interactive Brokers trading platform.
2:32And you also need Python. Again, no
2:35coding experience is necessary here
2:37because we have Claude. So, I already
2:39have Python installed and I'm just going
2:41to start running my Interactive Brokers.
2:43I'm going to be using a paper trading
2:45account on TWS because we're not putting
2:48in real money on this yet until it runs
2:51clean for weeks. So, I'm going to log in
2:53to my paper trading account. Uh make
2:55sure you go through the process. You can
2:57download TWS on the Interactive Brokers
3:00website. I'll leave the link down below.
3:03And you can open up a free paper trading
3:04account even if you had don't have
3:07market data subscription.
3:09All right, we're in. You can see now
3:10we're in the Interactive Brokers TWS
3:12platform. I've done video lessons and
3:15tutorials on how to optimizing and make
3:18it look prettier. I tried my best, but
3:21it still looks like this. I know. So,
3:22what you want to do is to set up the API
3:24access. Go to file up here and go to
3:28global configuration.
3:30And once you're in here, you want to go
3:31to API, go to settings, and you want to
3:35enable ActiveX and socket clients.
3:39And over here, the next part is you want
3:41to make sure the socket port here is
3:43labeled as 7497.
3:46This is the paper port. If you're
3:48changing this up to a live account
3:50later, you just need to change a number
3:52to 7496.
3:55Next, scroll all the way down, um
3:57uncheck that, and create a trusted IP.
4:01You want to add 127001
4:04to the trusted IPs. I've already done
4:07this one, so I'm just redoing this again
4:09with a new paper trading account for
4:11this demo. So, once you have that ready,
4:13you want to click apply, but I want to
4:14check one thing. One more thing, up here
4:16on the general, you want to check off
4:18read-only API because you want the API
4:21to actually be able to
4:22execute trades for you. So, make sure
4:25you have these things enabled and the
4:27trusted IP, and you want to click apply
4:30and
4:32and okay and okay.
4:34Okay, and after that, you don't really
4:36need your TWS anymore, so I'm going to
4:39minimize it. Another thing you want to
4:40make sure you have is you need to have
4:42Python tools installed on your PC. I
4:45already have Python installed. If you
4:46don't, just download Python 3.12
4:50or newer, Claude or Codex can walk you
4:53through that easily. And oh, and I
4:54forgot to mention you obviously also
4:56need either Claude or Codex on desktop.
5:00Um so, if you don't, make sure you go to
5:01their website and at least get the I
5:03think the $20 per month minimum max
5:07plan, the paid subscription, um then
5:09that should be enough to get you
5:10started.
5:12So, after we're done with TWS and
5:13Python, we want to create a new
5:15environment for us to work in. So, I'm
5:17just going to copy this prompt um from
5:20my blog here. So, next we want to start
5:22a fresh project um in Claude. You can
5:25see I named this one Interactive Brokers
5:27Claude demo. So, I want to make sure
5:29that Claude can now talk to Interactive
5:32Brokers. I'm working on connecting
5:35Interactive Brokers
5:37Python and Claude together.
5:41I have an Interactive Brokers paper
5:42trading account in global configuration
5:45setup already.
5:47Can you walk me through the connection
5:50process?
5:55Okay, so you can see that Claude is
5:57walking me through the process. Um, do
5:59you want to place order placement label
6:01or read only? We want to do full
6:03trading, which is with paper.
6:07So now I think about it, you can start
6:09at the very beginning of the video just
6:10asking Claude to walk you through as
6:13well.
6:14And really important, I'd like to use
6:16auto mode
6:18because, you know, it allows Claude to
6:20kind of proceed without asking all the
6:23permissions. I hate that. But obviously,
6:25if you're more careful, you can use ask
6:27permissions, but you just got to be here
6:29and you'll have questions for you like
6:31every 10 seconds. I've been doing it
6:33doing a lot of projects and tasks on
6:35Claude, so that's why I'm comfortable
6:37using the auto mode.
6:40As you can see, Claude just did a whole
6:41connection pipe for us with Python and
6:45Interactive Brokers. Now we can test the
6:47connections.
6:49Like I mentioned at the beginning of the
6:51video, AI won't magically make you
6:53profitable. Claude can help you build
6:55scanners, analyze news, and even
6:58automate parts of your trading process.
7:00But if the underlying strategy has no
7:02edge, the bot will only help you lose
7:05money faster and more consistently. At
7:08the end of the day, the edge is still in
7:10your strategy and in your ability to
7:12identify opportunities in the market.
7:15For the momentum day trading strategies
7:17I personally trade, I tend to focus on
7:19sectors showing outsized relative
7:22strength. And this year, one of the
7:24biggest themes has been semiconductors
7:27and all the companies powering the AI
7:29boom. If your strategy also rides
7:31momentum in that sector, BetaPro offers
7:34a 3x leveraged semiconductor bull ETF
7:38that can potentially provide amplified
7:39exposure to the semiconductor market
7:42movements. On the other hand, if the
7:44strength fades and the sector turns
7:47bearish, there are also inverse
7:49semiconductor ETFs designed to provide
7:52inverse exposure to declines in the
7:54sector. Just remember, leveraged ETFs
7:57are trading tools, not investing
7:59products. They reset daily and in
8:01volatile markets, losses can add up just
8:04as fast. They're built for daily
8:06short-term trading and aren't typically
8:08meant to be held long-term. Make sure
8:10you understand the risks before trading
8:12leverage products. You can learn more
8:14about BetaShares lineup of leveraged
8:16ETFs using the link down below. Now,
8:19let's get back to building the trading
8:20bot with the help of AI.
8:27As you can see, Claude just did a whole
8:29connection pipe for us um with Python
8:32and Interactive Brokers. Now, we can
8:34test the connections.
8:36Now, I'm going to ask you some
8:37questions.
8:40Tell me if I have any open positions in
8:42this paper trading account
8:45on Interactive Brokers TWS.
8:49And also give me my buying power.
8:55All right, let's see if it can.
8:58This is how you test the connections.
9:01Oh, wow. Okay. So, you can see the
9:03snapshot for my paper account. There's a
9:06million in paper money.
9:08Um you know, the pro tip, don't start
9:10paper trading with a million. I would
9:12reset this to like 10,000 or something.
9:15Uh I just It's like I said, it's a brand
9:16new paper trading account I just set up
9:18yesterday, so I did I forgot to do that.
9:21Next, let's see. So, you can see in this
9:23that's This is the account
9:26that's pulling.
9:28Um let's see what else can it do.
9:30So, this is the account that is pulling
9:32a data from. You can see um the cash is,
9:35you know, the paper trading cash is a
9:37million.
9:39That's great. Let's see if we can try to
9:41place an order for us.
9:44Can you place a buy order on Nvidia with
9:48a limit price of $100?
9:54I want to buy just one share.
9:59All right. So, let's test the connection
10:02if and see if Claude can now place the
10:05order for us.
10:06So, this is the first paper trade I'm
10:09placing on Interactive Brokers TWS here.
10:12So, you can see there's a pop-up for
10:14simulated trading. You can enable API
10:16orders. There's a precautions. This
10:19basically just telling you that you are
10:21placing trades via an API, which yes,
10:23that's what we're doing. So, I'm going
10:24to click yes.
10:26So, you can see they just tried or
10:29Claude just tried to buy one share
10:32of, you know, quantity right here if you
10:34zoom in. Quantity one share of Nvidia
10:36with a limit price of $100. So,
10:38obviously that's not going to get
10:39filled, but you can see now the
10:40connection is working because Nvidia's
10:43trading at $219 right now. But, that's
10:45just a test.
10:47Okay.
10:48>> Order cancelled.
10:49>> So, I just cancelled the order. Now, we
10:51can move on to part two, building the
10:53Interactive Brokers API bot. And we're
10:56going to start with building the
10:57strategy. In the previous video, we
11:00already have a backtested strategy from
11:02TradingView and Claude. And these were
11:04the results on the last video. You can
11:06see this is pretty good. Profitable 64%
11:09of the time. Profit factor is 2.48.
11:13And the max drawdown is 7%. Total P&L is
11:1712%. And you can see the list of trades.
11:21Um you can see the entire detail
11:23actually in the last video. So, this is
11:25a simple um trend join long setup that
11:27we created in the last video using
11:30Claude and TradingView to do the back
11:32test and creating the Pine Script
11:35strategy over here. So, what what we're
11:38going to do now is to turn this exact
11:41trading strategy, the Trend Join Long
11:43setup, into one that can be read by the
11:45new Interactive Brokers API bot. So, the
11:48bot is going to read all the rules, all
11:50the entry and exit criteria that we
11:52created here, and it's going to run it
11:55on every single cycle, and we're going
11:57to build that right now. So, what we'll
11:59need to do now is to create a strategy
12:01rules file called rules.json, and we can
12:05ask Claude to do that for us.
12:08Hey Claude, build me a rules.json file
12:13for the trading bot. The strategy name
12:15is Trend Join Long. This is a long-only
12:18strategy. Um it's going to be traded on
12:21the 5-minute chart, and here are the
12:23daily filters. The stock needs to be
12:26above the previous day's high. The
12:28previous day's close also needs to close
12:31above the 200-day.
12:34Here's the exit criteria. The initial
12:36stop is 1% below the low of day. Partial
12:40take profit is at 0.75R.
12:44Then you set the break-even at 1R. Then
12:47trail under the 5-minute swing lows.
12:50Here is the risk. 1% account size per
12:54trade. Maximum 10% position size of the
12:58entire account. So, I know that was a
13:00long prompt, but you can just go to him
13:02on my blog [clears throat] again and
13:03just go to the strategy file here. You
13:06can see I listed all the daily filters
13:08here. You can just copy this here. You
13:10can just copy the prompt, so it'll be a
13:12lot less work for you. So, let me see if
13:14everything's correct. I'm just going to
13:15clean this up really quick.
13:18So, we're inputting the prompt into uh
13:22Claude right now to build that
13:23rules.json file.
13:25So, you can see Claude is working its
13:27magic right now.
13:30This is amazing. So, you can see it
13:32created the file um strategy name long
13:36only index. So, one thing I forgot to
13:38mention is that when did the strategy
13:40demo and the backtesting in the previous
13:42video, I only did it on the S&P 500
13:45stocks just to keep it simple.
13:47But, obviously when I trade day-to-day,
13:50the trend joining long setup, I'll trade
13:52it on all the stocks over a billion
13:54market cap and over $3 per share. So,
13:57just a caveat here. You can add any
14:00stock universe that you want. Before we
14:02move on, just a reminder that if you're
14:04enjoying step-by-step videos like this
14:07that are really easy to follow along,
14:09then make sure to drop a like down below
14:11and subscribe if you want to see more
14:13like this. Now that we have the strategy
14:15file ready, next we need to build a
14:17settings file that tells the bot our
14:20Interactive Brokers port, position
14:22limit, account information, um and
14:25Telegram tokens for later. This is
14:27really boring, but necessary. So, what
14:29I'm going to just going to do here is to
14:31go to my blog and copy the prompt here
14:35and then going back to Cloud and paste.
14:38So, what this prompt does is that it
14:40builds the bot that actually talks to
14:42Interactive Brokers, reads the rules,
14:44and decides if we should trade. And also
14:47orchestrates everything and places the
14:49orders on the back end.
14:52And the connection and everything is all
14:54firing correctly. That's what we want to
14:56see. Okay, amazing. Like this is so much
14:59easier than our last video with
15:01TradingView connection to Cloud.
15:04So, what we just did is we set up
15:06Interactive Brokers API and Cloud
15:08connection. We also built the trading
15:11bot with our backtested strategy. Now,
15:14we can actually have Cloud take the
15:16trades for us. So, what's next is the
15:18fun part. We're going to build out the
15:20rest of the trading pipeline and create
15:22a brain for the spot. So, now the bot
15:25will have a way to scan the market for
15:27you for all the potential stocks that
15:29fit your strategy. It'll filter through
15:32the gappers and repeats this process
15:34every 30 minutes. And if there are
15:37potential trades to take every 30
15:39minutes, it'll pull the trigger and
15:41execute on the entries and exits. So,
15:43we're going to identify the kind of
15:45stocks we want to trade. For the trend
15:47following long strategy specifically,
15:49like I mentioned earlier, I want to
15:51trade bigger cap stocks. So, for this
15:54demo, we're going to limit the stock
15:55universe to all the stocks in the S&P
15:58500.
15:59Hey Claude, create a Python list called
16:02um SP500_tickers
16:07and putting all the stocks, all the
16:09tickers from the current S&P 500
16:12listings in there.
16:15Okay, great. You can see Claude created
16:17the list for us of all the stocks in our
16:20universe that the bot is going to trade.
16:22But now you need to give that bot a
16:24filter. So, to create the filter, I'm
16:26going to enter this prompt here. What
16:29this prompt does is to further narrow
16:31down the list of the 500 tickers. So,
16:34what we're filtering for top 20 stocks
16:37that gapped up more than 3% on the day.
16:40Remember, that's the criteria for this
16:42trend following long strategy.
16:46So, you can see Claude is working its
16:48magic. You can see I asked me earlier to
16:50move the 500 tickers list we created to
16:53somewhere else. So, I said yes, and now
16:56it's working in the background. Okay, so
16:58now it created a morning pre-filter.
17:00Let's see what that filter looks like.
17:02Okay, that's what it looks like.
17:05So, you can see now it created a morning
17:07pre-filter based on the criteria we just
17:09told it. Um the top 20 stocks that's
17:11gapped up uh more than 3% on the day. We
17:15can run a test. You can see it's asking
17:17it run a test. So, I'm just going to do
17:19that. So, okay. So, let's run the test
17:22to see the pre-filter stocks that spits
17:24out.
17:26All right. Amazing. You can see Claude
17:28is doing all the hard work for us right
17:31now. Okay. So, here's the summary. They
17:33screened the 503 tickers from the S&P
17:36500 in 14 seconds and gave us the top 20
17:40stocks um that have gapped up above 3%.
17:43So, here are the results. Over here, you
17:45can see it listed it out. Uh and by the
17:48way, this is a list that we're going to
17:50wire through to our Telegram
17:52notifications later on. So, I want to
17:55see this list of the pre-filtered stocks
17:58every 30 minutes. So, I'll show you how
18:00to set that up later as well.
18:03So, this is pretty cool. You can see it
18:04also gave us some um observations.
18:07Heavy semiconductor and um memory stocks
18:11today, which is the case today. Today,
18:13all the all those stocks are flying to
18:15the moon. Speaking of watchlists, I send
18:18out a free weekend watchlist every
18:20single Sunday, which include three to
18:22four stocks I'm watching for the week
18:24for day trading and swing trading. You
18:26can sign up for free with the link down
18:28below. Okay. So, now that we have the
18:30API bot, the strategy, the scanner,
18:33everything in place, we just need to
18:35automate it and set up the alerts. Let's
18:38set up the brain cycle now. This is the
18:41key to automating the entire trading bot
18:43cycle every single day. And this is a
18:46prompt I'm going to use. So, I'm going
18:48to copy
18:50and paste.
18:51So, what this is doing that it's asking
18:53Claude to set up the cycle of scanning
18:55and taking the the order entries every
18:5930 minutes during the market hours. You
19:01can change this to 5 minutes if you
19:03want. Um but for trend during long, I
19:05think 30 minutes is fine. So, you can
19:07see we missed a step earlier. We need to
19:09create the watchlist file. Um and this
19:11is the file that's updated every 30
19:14minutes. So, Claude is going to do that
19:15for us.
19:20All right. So, that took about a couple
19:22minutes. Not bad, not bad. So, this is
19:25the cycle that's created. So, you can
19:27read through this over here.
19:30Uh, all different cycles and timing on
19:32with Eastern time and what it created
19:36with all the files and the project tree.
19:38Um, I just want to say that I have no
19:40coding or tech background. So, as long
19:43as it's doing what I told asked it to
19:45do, which is I'm the trader, I created a
19:48strategy, I gave it to Claude, um, and
19:51with API and Python it wrote me this
19:53cycle. As long as it's executing the
19:55trades and running the cycles and
19:57scanners like I asked it to do, then
19:59it's perfect. Like I already said, I
20:01came here with no technical background.
20:04Next, let's move on to Telegram alerts
20:06like the ones I showed you on screen
20:08earlier. As part of automation, I want
20:10to set that these alerts to be sent to
20:12my phone. So, you're notified during the
20:14scanning process every 30 minutes. When
20:17a bot takes a trade, you are notified as
20:19well or even when it gets out or take,
20:22you know, losses or profit. And I also
20:25wanted to create an end of the day
20:26summary. This can be done easily with
20:29Telegram notifications. I showed you the
20:31step-by-step process in the last
20:33TradingView video, so you can check out
20:36the full breakdown if you need more
20:37details. But for this demo, I just want
20:40to set up this prompt.
20:45All right. So, it looks done. Let's do a
20:47test right now.
20:53Okay. So, you can see that it sent some
20:57examples cleanly on our Telegram. You
21:00can see the pre-market filters, the
21:01pre-filters sent at 9:32
21:05um, a.m. You can see the top 20 list of
21:07stocks. You can see some examples of a
21:11buying a stock, selling a stock,
21:13partials, trailing stops. And uh it
21:16doesn't have a good example of the end
21:17of the day summary, but you can see
21:20Claude is explaining how you don't have
21:22a real P&L summary right now. But, if it
21:25does take some trades, it's supposed to
21:27look like this.
21:29I'm not going to lie, building this
21:31automated trading system hasn't been the
21:33simplest thing. Well, yes, the strategy
21:36was already backtested on TradingView to
21:39be really high win rate and really
21:41profitable. Once it's transported to
21:44Interactive Brokers API for actual
21:47executions, the results are just very
21:50meh. This is a dashboard I asked Claude
21:53to build. You can do the same as well.
21:55Just use the prompt on my website. I
21:57think there are many reasons why there's
21:59a huge discrepancy between the two
22:02trading performance. The backtest was
22:04limited. We ran an unlimited number of
22:07tickers with limited timeframe on
22:09TradingView. Second is, I haven't quite
22:12figured out how to translate all the
22:13executions I do manually as a human to
22:17be 100% defined as a code with
22:20indicators and all the technical
22:22parameters. That's the part I'm still
22:24testing out, and I'll make future video
22:27updates on this. The honest result so
22:29far, the backtested numbers don't
22:32perfectly translate to live execution.
22:35Right now, my own performance as an
22:37actual human being trading a trend join
22:39long setup is still much better than any
22:42of these bots um automations. But, I do
22:46think this TradingView and Interactive
22:48Brokers API system is a solid starting
22:51point. So, I'm going to keep working on
22:53this. It's not finished yet. So, make
22:56sure to follow along because I'm going
22:57to make a lot more videos on how I'm
23:00optimizing my strategy, performance, and
23:03trading executions with AI, cloud, and
23:07all the tools out there. Special thanks
23:09to BetaPro ETFs for sponsoring this
23:11video. Make sure to check out their
23:13leverage ETF products with a link down
23:15below. If you haven't seen the video on
23:18how I'm using cloud and TradingView to
23:20backtest my strategies and create a
23:22premarket scanner, then you should check
23:24out this video over here.

About this transcript

This page contains the full transcript of Claude + IBKR API: Complete AI Trading Bot Guide by Humbled Trader, generated from the public captions YouTube serves with the video. The transcript has 3,935 words across 579 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.