How to Use Jev to Build a 24/7 HFT Trading System

0:00I don't care how long you've been

0:01trading, this is insanely cool. If

0:03you're someone who wants to get into

0:05trading or currently trading and you

0:06want to integrate AI into this process,

0:09well, we've just had something

0:10monumental happen in the AI world. A new

0:12AI has come out called Jev, and everyone

0:15swept this under the rug like it wasn't

0:17important, but for trading,

0:19this is a game-changer, no joke. What

0:21you're seeing on screen is every time a

0:23new dot flashes,

0:25my AI is buying and selling Bitcoin.

0:28Currently, it's doing paper trading, but

0:29you can convert this with a flip of the

0:31switch into live trading. I was

0:34scrolling through Twitter and I came

0:35across a post by Rone on chain. He's a

0:38quant analyst and I love his work, so so

0:41much credit to him for bringing this to

0:43my attention. But, I've seen so many of

0:45the smartest people I know online just

0:47completely enthralled with this system

0:49that I'm about to show you how to build

0:51today. I'm going to give it to you

0:52absolutely free. I'm going to give you

0:53the full one-shot prompt, so it's really

0:55just a copy and paste into your Claude

0:57and have this whole system up and

0:59running in just a few minutes. You'll

1:00have access to Jev, which is currently

1:02invite-only, but I'll get you in. It

1:04will allow you to build a strategy and

1:06implement that strategy in paper trading

1:08or live trading by the end of this

1:10video. So, the first thing to do is

1:11click the link in the top line of the

1:13description, and it will take you to my

1:15prompt page. This is the Jev loop, I'm

1:17calling it. So, you come down here, just

1:19type in your first name, and your last

1:21name with your email, and you can unlock

1:23the prompt. As soon as you unlock the

1:25prompt, you can just download the prompt

1:26or copy the prompt and simply bring it

1:29over to your Claude code and paste the

1:31prompt in. That's all you're going to

1:33need to do. We're going to follow the

1:34process together in this video. So, if

1:36you enjoy it, please hit subscribe and

1:38like. These one-shot prompts have been

1:39designed so that you have to do very

1:41little to make this thing work. I just

1:42want you to achieve the goal in the

1:44fewest amount of steps. So, let's get

1:46started and let's walk through what this

1:48build is going to do. So, as it's

1:50suggesting here, it's saying that the

1:52Jev key is likely a mock run because

1:54TypeSafe is a wait-listed gated thing.

1:57You're not going to be gated by the

1:58waitlist. I'll show you how to get round

2:00that in just a second. So, all we have

2:02to do is say go, and it will start the

2:04process of building this out for us. So,

2:06while it's doing some work here, I

2:07thought I'd explain a little bit what

2:09about what makes Jev so interesting

2:11versus the other models. And it's

2:13actually because comparing Jev and these

2:15other models uh as like apples to apples

2:18is not the right thing to do. They're

2:20actually not similar at all. So, the way

2:23that you would look at classic AI, which

2:25is LLM based, Claude and Codex, ChatGPT,

2:29you're typing to the AI, and the AI is

2:31typing back. Jev doesn't interact with

2:35language like that.

2:36Jev is making decisions at very fast

2:40rates. And so, what's really important

2:42here, and I really have to emphasize

2:44this because it's probably one of the

2:45more important things I'll say here

2:46today. There is a danger that you use

2:48Jev like uh

2:50a fancy calculator. Because if you if

2:52you engage with Jev with information

2:55that's nuanced and everything, it's

2:57going to slow down, and the benefit of

3:00Jev is that it can make decisions based

3:02on programmatic uh predetermined

3:05information very quickly.

3:08And so, of course, you have to look at

3:09what is the quickest way to engage with

3:11a computer. Well, it's probably binary,

3:13right? Like ones and zeros. This or

3:16that.

3:17This is the best way to get the most out

3:19of Jev. So, when you're applying that to

3:21trading, what we need to bring it is not

3:25the nuanced ideas of a trading strategy,

3:28but rather, you need to bring it the

3:30binary. This needs to happen, and if

3:32this happens, then this happens. And if

3:34this happens, then this happens. And

3:36ultimately, is this a trade? Yes or no?

3:39Two options at all times. Yes or no. You

3:43know, up or down, uh good or bad. That's

3:46what we're looking for. And if you can

3:47feed the information in with your

3:49trading uh strategy into Jev, it's going

3:52to go at 500 milliseconds.

3:55Uh the speed. It's insane how quickly it

3:58can make decisions.

4:00And so, if we're throwing a strategy or

4:01loads of trading data at it, and we want

4:04it to test out even like backtest

4:06strategies, we need to be very clear,

4:08crystal clear, about the programmatic

4:11element of what we give it. Meaning, how

4:14many of the data points can we assign

4:16yes or no or a one and a zero to it?

4:19And of course, like I know that's really

4:21all very complex stuff. If you're a

4:23veteran trader, you might have a bit

4:25more programmatic understanding. But, if

4:28you don't, you're likely using AI in the

4:30first place to try to figure out this

4:32programmatic stuff, right? How do you

4:34talk to a computer about trading? I'm

4:35just trying to get

4:37the hang of trading in the first place,

4:38right? I understand. And that's why, you

4:41know, it's really important to build

4:42strategies properly. I build strategies

4:44for a living. That's literally what I do

4:46all day with AI.

4:47And so, if you're interested, we've got

4:49stuff going on uh in the description

4:51here. You can actually help yourself

4:52build some amazing strategies. We've got

4:55a $9 thing that you can join. We have a

4:57trading competition every month where

4:59you can win $500 for winning.

5:01$9 subscription, $500 prize. Uh but, we

5:05also have a course there for building

5:07out your first ever trading strategy and

5:09iterating on it and developing more

5:11strategies and iterating and uh creating

5:14a self-improving loop. So, if you're

5:15interested in that, I would really have

5:17a look.

5:18Um I've put my heart, blood, sweat, and

5:20tears, and soul, and everything into

5:22this uh system for people just to build

5:24trading strategies properly.

5:26And so, obviously, we're going to need

5:27that element done. But, if we've built

5:30the strategy and it's all got the logic

5:32and it is programmatic, and now we

5:34engage with Jev,

5:36that's when you can create stuff that

5:38looks like this. Like, it's trading

5:39consistently, making decisions as we go.

5:42You can see on the side here that it's

5:43coming up with buy and sell. Each one of

5:45those flashes or each movement is

5:47actually a buy or a sell.

5:49Um but you'll also see this percentage

5:51and the percentage is the confidence

5:53score that Jev is giving that decision.

5:56So, Jev is never going to say 100% yes,

5:59100% no. It's going to have a a scale.

6:02It's a basically a spectrum of

6:04confidence. And if you're installing

6:07quant-like systems, quant analysis,

6:09quantitative analysis systems

6:11underneath, you're getting this really

6:13interesting mix of

6:15huge amounts of data blended with

6:17programmatic behavior blended with Jev.

6:20And Jev can make these decisions

6:22allowing you to do high frequency

6:24trading with Jev in real-life

6:27environments. It's magic. Like truly

6:30truly is. So, essentially this prompt is

6:32going to install for you my Jev loop

6:35skill. This skill is basically going to

6:37allow you to trade uh continually on

6:40loop using this high frequency trading

6:42method with all the quant principles

6:44built in underneath.

6:46And actually execute trades for you if

6:48you so desire. As it currently stands,

6:50it's just going to do paper trading cuz

6:51that's the safest way for me to do it.

6:53But seriously, linking it up via API,

6:56talking with your Claude or your Codex

6:58and saying, "Hey, I want to trade live.

7:00Here's my API for my exchange or

7:01something." That's all it takes to turn

7:04it live to like real-life trading. I

7:06would say you want to do paper trading

7:07and monitor the

7:09the um performance of it over time, but

7:11you know what I mean. I don't have to I

7:13don't have to tell you what to do.

7:14It's also going to be building out your

7:16dashboards. There's a couple of

7:18dashboards we're going to make. One

7:19looks like this, another one is more of

7:21a dark mode. It shows you all the data

7:22points that are making up a decision and

7:25then the ultimate decision in the end.

7:27It's quite a cool way to look at it. And

7:28these are This is currently a real

7:30reflection of what's actually buying and

7:32selling on my account right now. Okay,

7:34so this isn't just like a simulated

7:36dashboard. This is real-life and it's

7:38actually doing this.

7:39Um obviously a small amounts of money

7:41each time, that's what high frequency

7:43trading is.

7:44Um, but look, it's on paper mode and

7:47it's just it's incredible. And so just

7:49to clarify the distinction between Jev

7:51and what you would find with like Claude

7:53or something

7:54is that you're not interacting with Jev,

7:56you're purely using the other other LLMs

7:59or if you've got enough brain power

8:00yourself to do it, you're feeding

8:02information to Jev to make a decision.

8:04That's all it's doing. You don't

8:05interact with Jev, data interacts with

8:08it. And if you can interact with it in

8:10the right format, aka as close to binary

8:12as you can, you're going to get a way

8:14quicker output from Jev, meaning you can

8:16do this high frequency trading. Um,

8:19whereas otherwise if you didn't do it in

8:20that way, you simply wouldn't be able to

8:22operate this fast. And I think I think a

8:25lot of people find this very appealing

8:26to be able to trade all the time in

8:28small ways. So it's an absolutely

8:29monster prompt that I've given you and a

8:31massive system that it's creating. It's

8:33already been working for 9 minutes and 2

8:34seconds at this point. So so now the

8:36system has noticed that we haven't got

8:38an Alpaca key on this machine. All it

8:40means is we haven't registered with

8:42Alpaca, which is the free paper trading

8:45system that we're going to use here. So

8:47what it's done in the background is

8:48actually opened up Alpaca for me in my

8:50dashboard. I haven't had to find Alpaca,

8:52it just opened it up for me. Of course,

8:54I've already got an account, but once

8:55you get an account, it's all free. Get

8:58your paper trading account and this is

9:00this should be your home page. By the

9:02way, if you come on here, you can

9:03actually see

9:05uh, which is kind of like the

9:06fascinating thing for me. You can see

9:08the activity on this account. Every

9:10seems like every 3 seconds or so there's

9:13a trade happening. 3 to 5 seconds is

9:15happening. Oh, that was the second after

9:17it made that trade.

9:18And you can see everything that's

9:19happening all the time and I've got so

9:22much happening from this from this bot.

9:24There's just so much stuff happening.

9:26Anyway, you need to connect it via API.

9:28What's also come up on my screen for me,

9:30haven't had to find it in my files or

9:32anything,

9:33is this and this is an ENV file.

9:36And this ENV file is going to hold the

9:38the secret information that we have that

9:40we need to basically place in here to

9:42make this whole system work.

9:44So, let's actually set that up.

9:46So,

9:47on if you scroll down a little bit on

9:49the right-hand side, you'll see API

9:51keys.

9:52You're going to click generate new API

9:54keys. I'm going to click regenerate so

9:56that we can generate new keys to set

9:58this up. I'll rotate these so they're

10:00unusable after this cuz it's quite a big

10:02security risk. So, first we're going to

10:04copy the key, come over to our file, and

10:07we're going to see Alpaca API key, and

10:10we're going to paste it in there. We're

10:11going to come back. We're going to see

10:12the secret. We're going to copy that.

10:15Scroll over, find the Alpaca secret key,

10:18and enter it in there. And then we're

10:19going to save this file. So, you can

10:21come up to the top and click file and

10:24save, or you can do command S or control

10:26S, depending on the system that you're

10:28using. So, now we've entered in the

10:30Alpaca key and their secret key. This

10:32means that the system should now work

10:35with paper trading.

10:36Just like that. Um it's also started to

10:38open up here for us on the side. Um this

10:41is the dashboard that will actually be

10:42functioning when the trades are running.

10:44Okay? So, you can make that half the

10:47screen. That's all there, prepared for

10:49you, waiting to go. So, let's keep

10:50reading.

10:51Phase four, Alpaca paper key, takes

10:54about a minute, and there's no cost.

10:55We've already done that. It says Look

10:58like just like I said, it said that they

10:59both open. You can sign in, um generate

11:01the new keys, get both of those keys.

11:03That's all good. And uh there we go.

11:07It's also then going to check to see if

11:09you have a Vercel account, an AI Gateway

11:12Vercel. On your version of this prompt,

11:14it will actually auto open the Vercel AI

11:17page, but I'm going to show you over to

11:18it anyway. So, all you have to do is

11:20come over to vercel.com/ai-gateway,

11:24and it is as simple as it looks. You

11:26click get API key.

11:28And and there you go. You get the API

11:30key. Um you click create API key here

11:33and you're given your API key. So, this

11:35is the best way to do it. You've got

11:37your AI gateway API key. You can click

11:39copy on that.

11:41Come over to your file over here. You'll

11:44see you've got AI gateway API key. You

11:47paste it in there just the same way as

11:49you have right after this uh equals

11:51mark. And then again, you want to click

11:53save. And that's what we need to do for

11:54this stage. We're just going to tell it

11:56now we're ready to continue. Ready to

11:59continue.

12:00That is not how you spell continue, but

12:02doesn't even matter at this stage. It's

12:04now a verified my Alpaca account,

12:06meaning we can now trade with uh paper

12:08trading. You can It can see how much

12:10cash we have. We can see that it's

12:11active with crypto enabled, whatever.

12:13You can change a lot of this stuff if

12:15you want. We're not going to get into

12:16that into this video. You'll see on the

12:17files that it's also asking for a type

12:19safe API key. You don't actually need

12:22that because we this uh Versel API

12:24gateway key is actually the workaround

12:26for for you know, you don't need an

12:28invite to Jev. Um, if you use Versel AI.

12:33Um, so that's what we're doing, which is

12:35why we don't need to fill that out.

12:37Okay, so perfectly what's come up here

12:39is which asset as Oh my god, I can't

12:42talk. Too excited. Which asset should

12:45the loop watch for the first run? So,

12:46we're just going to say uh let's do a

12:48Bitcoin USD. It's like you can choose

12:50what you want really as long as uh

12:52Alpaca serve that market. So, it's

12:54saying it's found it. There's crypto

12:5624/7.

12:57Uh $10 minimum order. That's fine. No

13:00shorting.

13:02This is long only. You could change some

13:04stuff,

13:05but for the sake of this video, we want

13:07to get through this um

13:08and respect your time. So, phase seven

13:11is going to do its first run.

13:1330 ticks against the real Jev model on

13:15Alpaca paper. So, if we now come over to

13:19our Alpaca account, let's come back over

13:21to the Alpaca.

13:22You should stay see stuff start to move,

13:25but if you're not seeing stuff start to

13:26move, then it's likely going to be

13:28happening on your dashboard, which is

13:29why

13:30it's better to watch the dashboard than

13:31to watch Alpaca.

13:33Um but the dashboard is a real

13:34reflection of what's happening on

13:36Alpaca. So, we'll wait for that to start

13:38populating and um shouldn't be long.

13:41Yeah. Okay, so actually ran it. Looks

13:43kind of crazy. It just kind of updated

13:44the the dashboard here. It said it ran

13:46it for 30 loops. So, it ran it 30 times,

13:49which is why we're seeing it here, and

13:50it's also not moving at this point. So,

13:53uh it ran for the 30 times. It made a

13:55series of buys and sells. Um and what

13:58it's now asking me to do is a bit fiddly

14:00now. For you, it will just continue to

14:02work after the 30 ticks. I'm going to

14:04ask it just to, by speaking to it, "Can

14:07you make sure this is running all the

14:08time now? I'm ready to um start trading

14:11at scale." Just give it a nice little

14:12nudge along. For you, you'll have the

14:14finished prompt, but this is really like

14:15I just wanted to get this out to people.

14:17So, I'm using a I'm using the worst

14:20version of this. So, if you think this

14:21is cool, try it yourself with the prompt

14:23below in the top line of the

14:24description. Now, it goes without saying

14:26that the strategy is so important here.

14:29This is really like the harness for you

14:31to

14:32uh create to plug your strategy into and

14:34see it perform. Everything you're seeing

14:36here today has just been a generic

14:38strategy pulled off out of thin air,

14:41just applied to this, just so you can

14:42see that it's trading.

14:44Real good strategies are really hard to

14:46come by. You need to build them

14:47properly. You can see actually now,

14:50because of this this kind of temporary

14:51strategy that we've put in, we've

14:53actually got this down to 395

14:56milliseconds

14:58um

14:59between trades, between decisions.

15:01That's really crazy. Like, we're really

15:03talking to it at the most fundamental

15:05binary level here. And uh

15:07your performance is going to vary

15:09wildly, but the performance is going to

15:11depend on the strategy. So, uh I like to

15:14build my strategies with full kind of

15:16quant systems and quantitative analysis

15:19underneath, and I use AI to do that all

15:21the time. This is what I do all day. I

15:23even turned my own

15:24portfolio from $1,000 to $180,000 in

15:28about 3 years.

15:29Um it just completely has captured my

15:31mind. So, really pay attention to

15:34building strategies. If you want to

15:35learn to build your first strategy, I

15:37would suggest you join the $9 community.

15:40Um and if you can find something that

15:42works good, get in the competition and

15:44win yourself some money. Now, Verse Cell

15:46is the way that you're getting access to

15:48Jev. So, you're going to have to put

15:49some money into Verse Cell pennies. It's

15:52actually pennies. All you have to do is

15:53come down to your account. You'll see an

15:55alert down at the bottom where it says

15:58uh you know, uh

15:59add a card or something like that. It's

16:01down at this bottom left-hand side.

16:03It'll say add a card. Just add your card

16:05and put like $5 in. $5 will last you

16:08like a month. Like it really there's

16:10very It's very cheap to run Jev. Uh and

16:13and Crunch just this data. So, it's kind

16:15of insane. And there you have it. A full

16:18system built for you very easily using

16:20Claude or any LLM to programmatically do

16:23high-frequency trading for you. Uh paper

16:26trading at first, moving into live

16:28trading as well. Now, before you go,

16:30there's something really important. If

16:32you try this prompt and something didn't

16:34work for you because you're on Windows,

16:36you're on Linux, or some weird thing

16:38happened for you, put that in the

16:40comments below this video. Me and my

16:42team read every single comment. We take

16:45your feedback and then we improve the

16:47prompt, okay? So, make sure you actually

16:49the feedback and bug reports that you

16:51give in the comments here will actually

16:53impact the result of the prompt that we

16:55give other people. If So, if you want

16:56the best product possible, make sure

16:58you're putting in in your bugs and

16:59feedback in the comments. Also, if you

17:01want me to share some information with

17:03you about how to build proper

17:04strategies, you can join the $9 school.

17:06It's really cool to join. Or you can

17:08just message the word quant below and

17:10I'll send you to some resources and some

17:12help to get you along your way. Happy

17:14trading.