Full transcript
0:07I get five to 10 DMs every single day on
0:09my Instagram of people who are
0:11struggling with using AI in day trading
0:12or they're having issues with back
0:14testing or live prices or really any of
0:17these five topics in day trading. And
0:19this video is to address the 90% of
0:22issues that you're having with day
0:24trading so that you can figure out what
0:26you have going wrong, make a change,
0:28implement it, and actually achieve your
0:30goals, actually get to where you want to
0:32be. If you don't know who I am, my name
0:35is Luke. I'm known on the internet as
0:38little fish. I have been day trading for
0:39over three years, and I have been using
0:41AI for around four years. This year, I
0:43started using AI in my day trading. For
0:46the first half of the year, I was doing
0:47AI assisted day trading where I wasn't
0:49watching the charts anymore. AI was
0:51watching the charts for me, sending me
0:52signals of trades, and it led me to my
0:54biggest month ever. I've made over five
0:57figures in payouts, and this summer I
0:59started my own live account, risking my
1:01own personal capital on my training.
1:03I've also started doing completely
1:06autonomous training. So, that was AI
1:07assisted. I'm now in the process of full
1:09AI automation. So, Claude built the
1:12strategy. Claude did the back test.
1:15Claude is placing the trades on proper
1:17and just yesterday I passed two accounts
1:20and those are now funded accounts and
1:22this morning Claude placed a trade and
1:24won a trade on the funded account. So
1:26hopefully we can keep making progress on
1:27those and get that money printer that
1:30everybody's after and that's going to be
1:32the first question probably the most
1:35common is can I have your bot? People
1:39want the money printer. Of course we do.
1:41Nobody gets into day trading because
1:43they're passionate about the way that
1:45stocks move. You're here to make money.
1:47We all are. That's why I started. That's
1:49why you started. That's why we're here.
1:51We're trying to make as much money as we
1:53can in an easier route. The idea that
1:57you can now have this money printer that
2:00runs around the clock is so appealing.
2:02It's less friction. It's easy. Why
2:04wouldn't you want that?
2:07But AI amplifies you.
2:11AI pushes your personality to its
2:14extreme. If you are a lazy person, AI is
2:17going to make you lazier. If you are a
2:19person with high agency, AI is going to
2:22allow you to do even more. And the first
2:25thing I actually want to talk about here
2:26is a parable from the Bible in Matthew
2:2925. And it's the parable of the talents.
2:32In this parable, there's three men.
2:36The first guy gets five talents. The
2:39second guy gets two talents and the last
2:42guy gets one talent. Now the five guy
2:44who has five talents, he goes out and he
2:46trades in the market and he gets five
2:48more talents and he brings them back.
2:50The second guy does the same thing. He
2:52goes and he trades his two talents and
2:54he brings two more talents back. Now the
2:57last guy, he went and he buried his
2:59talent because he was afraid and he
3:01didn't want to work. And when he came
3:04back, he only had the one talent he
3:06made. Now when the master spoke to these
3:09servants, the one who had made five
3:11talents, he gave five more talents. The
3:15one who made two talents, he gave two
3:18more talents. And the guy who made one
3:21talent, he said, "Why did you only wake
3:23one talent?" Because he was afraid. He
3:26was afraid what the master would do if
3:28he lost. And so the master took his
3:31talent away for not taking the risk.
3:33Now, that's a story about risk, but it
3:35applies the same here in the sense of
3:38working. AI is not going to give you the
3:41magical money printer that you want. If
3:44you're here because you want to know how
3:45to build the magic printer that you
3:48click a button, you wake up and you make
3:49money and you never had to do any work
3:50and now you just live for free and do
3:52whatever you want all day, I want you to
3:54leave the video now because this just
3:56isn't for you. This is just that's not
3:58how this works.
4:00I personally am working harder than ever
4:04right now, but it's different work. It's
4:08work that anybody can do if you
4:11understand the tools and use them to
4:13your ability.
4:16AI is your five talents. If you look at
4:1950 years ago, 100 years ago, the status
4:22of life that we're living now is so far
4:25beyond. And the amount of opportunity is
4:28unreal. Ridiculous opportunity. The
4:31amount of blessing we have in
4:33opportunity from God should make you so
4:36excited to work. You should look and
4:38say, "Wow, if I just work hard, I can
4:41get really rich. I can make a lot of
4:43money." And what should be an even
4:45bigger driver for you is I can make so
4:48much money. And that money can be a tool
4:50to change other people's lives.
4:53that money can be a tool to bless the
4:56people around me because if that money
4:58is just for you to consume and enjoy and
5:00be selfish with that is going to be an
5:02empty pit. It is going to run dry. That
5:05well will be dry.
5:08You need to be driving for the right
5:09reasons and you need to be inspired by
5:12the opportunity that God has given you.
5:14AI is your five talents. This is new
5:17opportunity. We're seeing what people
5:18were doing. You're probably on this
5:20video because you're seeing what I'm
5:21doing with it. I'm out here working
5:23really hard and I am making money
5:27and you can do the same thing. Now the
5:30difference is the kind of work has
5:33changed. The barrier to entry has
5:36dropped. You no longer have to be this
5:39super genius expert in technical
5:41analysis. You can speak English and
5:43speak your ideas and go so much further.
5:46You don't need to be some quantitative
5:48that knows how to code in Python. You
5:50just need to speak English about your
5:52ideas to an AI agent or AI system that's
5:55going to do all of the work for you. The
5:57barrier to entry is here. You just need
5:59to work hard, be inspired by what God
6:01has given you, and do it for the right
6:03reasons.
6:07Number two, my strategy isn't profitable
6:10or my win rate is 60%.
6:15When somebody tells me their win rate is
6:1860%.
6:20Your initial reaction to that might be,
6:22"Oh, that sounds pretty good." Or, "I
6:25win rate 90%."
6:27Oh, wow. That's pretty good. That means
6:30nothing on its own. That is irrelevant
6:33on its own. You need more context than
6:36just my run rate is 60% or my strategy
6:39isn't profitable.
6:41And the best example of this is the 2021
6:44memecoin peak.
6:47The win rate of the most successful
6:49traders in the 2021 memecoin peak
6:54was probably somewhere between 1% to 5%.
6:57Maybe even less because remember this
7:00was a time where you could put $20 in a
7:02coin, wake up overnight, and have 100k.
7:06But you had invested $20 into 500
7:08different coins and you just won once.
7:12So your win rate when you were the rich
7:14guy winning was 1%.
7:1899% of the time you lost. But the one
7:22that you won, the riskto-reward was so
7:24huge
7:26that you could kill it. Data in
7:29isolation is useless. If you tell me
7:31your win rate is 60%. I can't do
7:33anything with that. Okay, great. or how
7:34much are you making per trade? Because
7:36if you're risking 500 to make 50 bucks
7:38and your win rate's 60%, you suck.
7:40Sorry. Sorry. The numbers don't lie. But
7:43if you're making that much, risking 500
7:45bucks to make 50,
7:48got to figure something else out here
7:49because you're not going to get very
7:50far.
7:53And this goes into the proper structure
7:56because profs are structured in very
7:58unique ways. They force you to trade a
8:01certain way because there are profitable
8:03traders in live markets who will have a
8:0510% win rate. But when they win, they
8:07win big. They win 20 times the amount
8:09they risk. And you can do that on a live
8:12account. And it is profitable. And it's
8:14so profitable that it's not able to be
8:17done on most prop firms. That is why
8:20they are structured the way they are.
8:22You have to meet certain requirements.
8:23You have to follow consistency rules.
8:26Consistency rules are a pain and they
8:29have only showed up in the past one to
8:31two years because people were getting
8:33rich without them. People were blowing
8:35these accounts to the moon and taking
8:37huge payouts. So they added consistency
8:40rules. So the way you need to approach
8:42proper structures is you need to reverse
8:45engineer. You don't say let me find a
8:47strategy and then let me go train on a
8:49prof. You need to say what are the rules
8:52for me to pass the prof and get paid.
8:54Now, let me find a strategy that fits
8:56those rules.
9:01Number three,
9:03this is a bit of just free sauce, if you
9:06will. Live prices and executing trades.
9:11A lot of people ask me, how do you get
9:13live prices? You use a Trading View MCP
9:16bridge. An MCP bridge has been something
9:18created by Anthropic and it allows
9:20claude code to read external resources.
9:23And what this is doing, this is dipping
9:24into the HTML of Trading View because
9:27Trading View is an electron coded app
9:29and the live prices of assets are stored
9:32in the HTML of the Trading View app. If
9:34you've ever clicked inspect element on a
9:36website, that's where the live prices
9:38are at. The MCP bridge reads those live
9:40prices and then allows cloud code to
9:42read those and then you can do whatever
9:43you want from there. That's how I read
9:45my live prices. That's how I get my
9:46signals. That's how I place my trades.
9:49One of the ways that I place my trades,
9:52I can think of at least four different
9:55ways to place trades.
9:58And Claude is probably not going to
9:59place them for you immediately.
10:02You can do APIs, you can do Chrome
10:04control interfaces, you can do so many
10:05different ways of placing trades, and
10:06I've done them all.
10:09But odds are, if you've gotten that far
10:12and you've said, "Hey, Claude, place
10:13this trade." Claude has probably said,
10:16"I can't place this trade for you
10:17because it's against my policy." So, how
10:20do we get around that? How do we get
10:21around this pain of an issue?
10:24Claude will place a trade if you give it
10:27a formula to place it.
10:31If you say place a trade right now, it
10:33will say no. If you say place a trade
10:36when X, Y, and Z meet with this
10:39criteria,
10:41it is going to build a code that is
10:45going to just be an automatic trigger
10:47and will place it when that happens.
10:50That way, Claude is no longer making a
10:52decision, but it is building a solution
10:55for you that just runs. And that is how
10:58I place my trades. That's how I've
10:59passed my accounts, and that's how I'm
11:00working towards a payout right now. I
11:02back tested all these strategies and all
11:04these systems and now Claude has built
11:06these presets. So when price meets XYZ
11:11criteria, trade automatically executes
11:14with stop-loss here, take profit here,
11:16this amount of sizing only within this
11:18window, only within this asset. Boom.
11:20It's extremely clear. But if you try to
11:23tell Claude to place it right now, it's
11:25probably not going to do it.
11:28>> [snorts]
11:30>> Number four, this is a big one. This is
11:32this is so huge. You are not using the
11:35full potential of AI. And this is
11:38something that's changed in the past few
11:40years because
11:42[clears throat] when we look at AI a few
11:44years ago, I mean I like probably many
11:48of you who are still in school at least
11:50have used AI when you were in school.
11:51And if you were one of the earlier
11:53users, you can probably remember a time
11:55when you asked AI to do a task and it
11:57failed and it did a crappy job. It
12:00couldn't get the right answer. It
12:01couldn't figure it out or it kept
12:02getting it wrong.
12:05There was a joke for a while, but it was
12:07true. You could ask ChatGpt, "How many
12:09Rs are in strawberry?" And it would say
12:11zero.
12:13One, two, three. Chat GBT would say that
12:15there's zero Rs in the word strawberry.
12:18AI was the bottleneck. We would ask for
12:21these things and it couldn't complete
12:23them. It was not a good enough solution,
12:26a powerful enough model. Context window
12:29wasn't big enough. Training data wasn't
12:32big enough. Too much halluc too much
12:35hallucination and in some cases now
12:38ethical constraints, which there's ways
12:40to get around those.
12:44That's not the case anymore.
12:47It is no longer the case that AI is
12:49holding you back.
12:51Remember, AI is an amplifier and we're
12:53here now where your ability to
12:56communicate your ideas to artificial
12:58intelligence will dictate the degree of
13:00solutions that you get. If you can
13:02communicate well, you will get good
13:04solutions. If you communicate lazy,
13:07you're going to get lazy solutions. And
13:10the statistics of prop firms are that
13:1290% of people aren't passing and that
13:14another 90% of those people aren't
13:15getting paid. So you need to take that
13:18extra step and work harder with them.
13:21And that applies here to use the full
13:24potential of AI. I did this this past
13:28weekend on the holiday Labor Day with
13:30one of my students in Ocean AI. We did a
13:34back test
13:36and he came to me with a strategy and he
13:38said, "All right, let's back test the
13:39strategy now. All right." I said,
13:40"Sweet. Let's do it." So, we're going to
13:43take this one strategy. We're going to
13:44hook it up to historical data and we're
13:46going to run some simulations here.
13:47We're going to see what's our
13:49performance of the strategy and we're
13:51going to ask some extra questions
13:53because why not? AI strength is pattern
13:55recognition. Your strength is judgment
13:57and discretion. Use the pattern
13:59recognition part for your judgment and
14:01discretion.
14:03So, let's ask some additional questions
14:05to see if we can find some more because
14:06AI is only going to do what you tell it.
14:09So, if we have one strategy that we want
14:11to test, we might go to AI and we might
14:14ask for things like how does it perform
14:16monthtomonth,
14:19why not test it on another asset, too?
14:21It only takes an extra sentence. If
14:23we're testing on NQ, what's the harm in
14:25testing it on gold? All you have to say
14:27is test it on gold, too. Okay, that is
14:30such a big deal because we can just do
14:32that now. We can just speak and get that
14:34done.
14:36I remember what back testing used to
14:37look like. I would come home and I would
14:39be on my computer and I would spend an
14:41hour or two hours every night after
14:44school or after work and I would go
14:45trade by trade dragging my charts and
14:48drawing out the position tools and
14:49writing down notes and writing out the
14:51results of RR and you can get a 100
14:53trades in an afternoon
14:58which is unreal.
15:01So we have this strategy right and we
15:04say let's test on another asset too.
15:06that. So, we're going to take the
15:07strategy and we've already doubled our
15:09output by just saying NQ and gold. We've
15:12already doubled the amount of results
15:14and information we're going to get,
15:15which is awesome. Why don't we do all
15:18the futures? Now, we're up to 10.
15:22Now, we have 10xed the amount of output
15:25we're going to get simply because we
15:27asked another question.
15:31Now, let's go a step further, right? We
15:33have our strategy.
15:35Can't we ask AI to make strategies of
15:37its own and make variations? How does it
15:40perform on different time frames? How
15:43does it perform during news events? What
15:46if you add different filters of the most
15:48popular confluences? VWOP, EMA,
15:53trend lines. Why not? It takes one extra
15:56sentence and you're going to do what
15:58would take hundreds of hours.
16:02All because you asked another question.
16:06So this weekend we did this back test
16:09and what we ended up doing for context.
16:11I want to highlight this so much because
16:14this is what I mean when I say that AI
16:16is your five talents because this is
16:17unreal. I've back tested manually in my
16:21life probably around a thousand trades
16:24and it's taken hundreds if not thousands
16:26of hours to do so. It's taken a lot of
16:29work. This weekend with one prompt that
16:32took 20 to 30 minutes to write. Fable
16:345.1 on Claude Code Ultra Code. We back
16:38tested over a 100,000 trades on 500
16:43strategies.
16:49You're bound to find profitable edges.
16:50If you do that,
16:53you're bound to get the information
16:54because that's the work that nobody's
16:56willing to do. All you had to do was
16:58think a little bit harder and ask the
17:00right questions. Keep asking. AI is not
17:04the bottleneck. Your ability to ask and
17:07give AI the work. You can say come up
17:11with your own strategies. That's fine.
17:14You can say be creative and think
17:15deeply. That's fine. You should say be
17:18agentic because being agentic means that
17:20it's going to find its own problems to
17:22solve and take it further in ways that
17:24you might not even be able to think. If
17:26you think you're smarter than AI,
17:28you're probably not watching this video
17:30anymore.
17:35Ask more questions. Ask for more from AI
17:38because you're going to get more.
17:41AI is a super genius. You just need to
17:43ask it to do that for you.
17:48Number five, cloud versus cloud code.
17:54There's a lot of different models and
17:56it's changing super fast. So odds are if
17:58it's not within the two weeks that this
17:59video is released that you're watching
18:01it. Things have changed. Fable 5.1
18:04that's my CEO. I use that in Claude
18:07chat. I ask for it to build prompts for
18:10Claude code which I use for Opus 4.8
18:12because Opus 4.8 does not have as many
18:14of the ethical constrictions that Opus 5
18:18does. Opus 5 has such an ethical issue
18:21that the advertisement for Fable 5.1 was
18:24that it had 60% fewer false positives
18:28because Fable 5 and Opus 5 in particular
18:30were giving so many issues to people
18:32saying things like, "Oh, I'm not going
18:34to do this for you because this is a bad
18:35idea." Or, "I'm not going to back test
18:37because it's giving poor results."
18:43What do you mean? What do you mean
18:45you're not going to do what I ask? Opus
18:474.8 8 does a lot less of that. So Claude
18:51chat fable 5.1 CEO to manage at the top
18:54code monkeys opus 4.8 and lower because
18:58they're still going to give you really
18:59good results and they're not going to
19:02stop you from doing the things you want
19:03to do.
19:08Analysis paralysis. I do a lot of things
19:12with AI
19:14and the one thing that makes me
19:15successful with getting what I want is
19:18taking one thing and bringing it to the
19:20end. The first thing I do with my
19:22students is I say, "We're going to pick
19:24something and we're going to bring it
19:25home. We're going to get there before we
19:28do anything else." Because you've
19:31already thought of so many things you
19:33could do with AI. Notifications, back
19:36testing, live testing, automated
19:38executions, self-healing models. I've
19:40done all of those things and I did none
19:43of them at the same time. I picked one
19:46and I finished it and then I moved to
19:49the next thing.
19:51Golden egg syndrome is going to be
19:54exactly that where you're working
19:57towards something and you you wrote your
19:58first prompt and you're waiting on the
20:00results and then you're like, "Oh, what
20:01if I could do this thing also?" And then
20:04you totally forget and drop the one
20:05thing that you were doing and went and
20:06did something else.
20:09The most successful students in my
20:11community, the ones who have gotten
20:12payouts, the one who are passing
20:14accounts,
20:16they have picked a project and brought
20:18it home. That is how you win.
20:22And the final point which comes back to
20:25this point as well,
20:28data without implementing is
20:30entertainment.
20:34You can run a 100,000 trade back test
20:37today and you can have a blast doing it
20:40and you can love reading in the numbers
20:41and you can see the green and the red
20:43squares and the cool navy background and
20:45all these lines and all these charts and
20:47you wow this is awesome. Look at all
20:49this work I just did. Okay, great. Nice.
20:57What was it for?
21:01What was the goal?
21:04If you want to entertain yourself, you
21:06can do that with AI. You totally can.
21:10But if you are not going to take the
21:11next step,
21:14you are wasting your time.
21:19And you can do that. You can you can
21:21entertain yourself all you want. It's
21:22fun. I enjoy it.
21:25But if you actually want to succeed when
21:28you are done with a test, when you are
21:30done with the system, the objective is
21:33to make a profit.
21:36And if you read what AI gives you and
21:38you say, "Wow, that's cool, bro."
21:41And you do nothing with it,
21:44you were merely entertained.
21:48If you want results, you need to work
21:51for them. If you are lazy, AI will
21:54amplify that because it's going to do
21:55the work you want and then it's going to
21:57leave you on your own. If you have high
22:00agency and you bring things to the
22:02finish line, AI is going to amplify that
22:05by squashing the amount of time it
22:07takes.
22:08If you wanted tonight, you could grind
22:11out in 6 to 12 hours strategy back test,
22:15building a live testing system, and even
22:17hooking it up to trade on real accounts.
22:21You could grind it out fast.
22:24Would it be hard? Sure, of course it is.
22:29But you can do it. You just have to
22:32work.
22:34And the only way that I can understand
22:36to motivate you to work is to give you a
22:38purpose outside of yourself.
22:42Go for the car. That's lovely. I'm after
22:45a car for myself right now. I would love
22:47to get a nicer car. I drive a Honda
22:49Civic 2014. I would like to upgrade that
22:52car. I could, but that's another
22:54conversation.
23:00We are blessed to have an opportunity in
23:02our hands.
23:05We are blessed to be able to use AI to
23:08build amazing things and do things
23:10faster than anyone ever in history. That
23:14should motivate you in itself.
23:17And what should motivate you even
23:18further is you have the ability to bless
23:21the people around you. You can change
23:23the lives of the people around you.
23:27That is a purpose worth grinding for.
23:33I teach in my oceanfront AI mentorship
23:37how to fill in these blanks, what I
23:39would do in certain situations, how I
23:41connect the dots and the best decisions
23:44I make. and I hold people accountable
23:46and I say, "All right, let's take these
23:48next steps. Let's keep going. Let's keep
23:50pushing for the finish line.
23:53Let's get paid. That is the goal." If
23:57that's something you're interested in,
23:58the link down below is there to apply.
24:01But otherwise,
24:03if you take away nothing,
24:06take away that AI is your five talents.
24:10You have an opportunity. You've seen
24:12people using it.
24:14Take advantage of it.