Full transcript
0:01So, what you're about to see is me
0:02taking a live call with one of my
0:04students and doing an entire prompting
0:07breakdown to get a back test. And I'm
0:10doing this with him to get as much
0:12details and information as we can, so he
0:14can take the best next steps and also
0:17understand the thought process of why we
0:19prompt in certain ways because I can
0:22just copy and paste the prompt and give
0:23it to him, but if he doesn't understand
0:25how to think through and what to do with
0:27the information, it's kind of a waste of
0:29time.
0:30If you've never seen me before, my name
0:32is Luke. I'm known on the internet as
0:34Lil Fish. I've been day trading for over
0:363 years. I've gotten over five figures
0:38in payouts. And um I use AI
0:42very frequently in my trading. We all
0:44have heard of this technology. We've
0:45seen it around before. I started using
0:47it back in 2022 and just this past year,
0:51I have used it for my trading and it's
0:53helped me get my biggest payouts ever
0:55and end up starting my own live account.
0:59AI is incredible.
1:01It's absolutely ridiculous and
1:05you can skip through and copy and paste
1:07the prompt and you'll get a lot of
1:09information and that's fine and that's
1:11great, but if you don't understand why
1:13we do things the way we do,
1:16you won't get anywhere with this
1:17information. You'll have the craziest
1:20back test that you're going to see
1:22results on here from this back test that
1:24are going to blow your mind. It's
1:26actually insane what we're able to do
1:29with 20-30 minutes of writing a prompt
1:32and a little bit of context and data
1:33providing.
1:35We have done an outrageous amount of
1:37work. I used to back test manually
1:40before all of this stuff was around and
1:43I would get home from school or I would
1:44get home from work and I would spend an
1:47hour
1:48to do 10 trades and I would document all
1:50of them on my Google sheet and I would
1:53put a little green box and a little red
1:55box for every trade and I would put down
1:57the RR and I would calculate my win
1:59rates using Google Sheets formulas, and
2:02then I would hop on FX Replay and use
2:04the free trial for as long as I could,
2:06and then I even started using
2:07TradeZella.
2:09And now we can test 500 strategies in an
2:12afternoon at the same time.
2:14And get more data than you know what to
2:16do with.
2:18And it's just absolutely unreal.
2:20Now, the purposes of backtesting can be
2:22completely
2:23It depends on your situation, it really
2:26does.
2:27If you're searching for just a
2:28mechanical edge, backtesting can be a
2:30phenomenal tool. If you're searching for
2:31a discretionary edge,
2:33I built discretionary trainers before
2:34because backtesting itself may not be
2:36applicable.
2:38But what we're able to do with this
2:40technology, with the right thought
2:42processes, is insane. And it can really
2:45expedite the process and skip out like
2:47all of this hard work. You don't have to
2:49spend so much time hunting anymore. You
2:51can just say like, "Okay, these are my
2:54These are my systems to work with. These
2:55are my profitable mechanical edges
2:57because they were tested on 1,000 trades
2:59and they came out profitable.
3:00And they were profitable in the past 2
3:02years, too. It's not just, "Oh, it was
3:04profitable in 1 month and then it was
3:05red the whole time and just averaged out
3:07profitable." No, it's
3:09a steady increase.
3:12And then with that information,
3:13then you can go and choose what you want
3:16to do. Notifications, live executions,
3:18prop firms, more testing, you name it,
3:20whatever you want.
3:23I run an AI mentorship.
3:26That's what the call is. This is one of
3:27my students, Nick. He has been
3:29absolutely crushing it, and he's going
3:31to use this information for taking on
3:33his testing to the next level and
3:35continuing to work towards passing. He
3:37counts he's only been in here for a
3:39month, and he's just
3:40been smashing it.
3:42If that's something you're interested
3:44in,
3:45there's a link down below to apply, but
3:47this is going to be free sauce. This is
3:49completely raw how I backtest, how I get
3:53absurd amounts of data and information.
3:57>> [clears throat]
3:59>> And to give you a little preview of some
4:02of the results that we're going to end
4:03up getting, this is something that I
4:04forgot to show in the actual back test.
4:08But,
4:09this is ridiculous. So, we have an HTML
4:13as one of our deliverables, and then one
4:15of the other deliverables that we have
4:17is this PDF
4:18of all
4:20these variants
4:21of strategies.
4:26This is 33 pages.
4:30It's absolutely unreal. So, here's the
4:33raw one-on-one client call with the Nick
4:37on how I back test my strategies.
4:41Everything's good to go.
4:42Cool. So,
4:45I'm going to hop right into it. So,
4:46we're going to run a back test using the
4:49strategy that you've created. This is a
4:50strategy on your own. There's a couple
4:52of different reasons that we run back
4:54tests. One of them, basic, is just to
4:57see how it plays out. We want to look
4:59for things like a win streaks. We want
5:00to test different risk-to-rewards. We
5:02want to see maximum losing streaks,
5:04especially in prop firm environments.
5:07And this is a way to historically test
5:10the performance of strategies. Most
5:11people know what back tests are. I don't
5:13really need to go into depth in
5:14explaining it, but
5:17it used to be the case where I would
5:19spend and a lot of time on my back
5:22tests. It would take me an hour to do 10
5:25different trades. And I would get home
5:27from school, get home from work, and I
5:28would spend my time and, you know, walk
5:30through trade by trade by trade, doing
5:32it on my own. And we don't have to do
5:34that anymore because of artificial
5:37intelligence and the new technology that
5:38we have. And it's so easy and it's so
5:42accessible because one of the core
5:44principles that AI has in place is its
5:47ability to recognize patterns. And all a
5:49strategy is is a pattern. We're looking
5:52for patterns in the market and we're
5:53looking to capitalize on those to
5:56see if we can extract a profit. And
5:59we can take a different a few different
6:01approaches of okay, I have a pattern and
6:04I want to see what happens when I run
6:05this pattern and what kind of results it
6:08yields. And we can also take the
6:09approach of I want you to test patterns
6:12on your own. This is something new. It
6:15we've been able to code strategies for a
6:18very long time. But what we haven't been
6:20able to do is to say, "Make your own
6:23variations. You fill in the blanks. Use
6:26your critical thinking. Use your
6:27judgment." And that's one of the new
6:28perks of using artificial intelligence
6:31and that's what I'm going going to be
6:32doing with this back test and
6:34essentially do exactly what I do when I
6:36test my strategies, when I'm building my
6:37automated systems on whether it's crypto
6:40accounts, whether that's prop firm
6:42accounts, um scalping, NQ, gold, you
6:45name it, whether that's my live account
6:47testing.
6:48>> [clears throat]
6:49>> Back testing is really easy.
6:52And it's a valuable thing to do to help
6:54you make the next decisions of should we
6:56live test, should we implement on a
6:57funded, should we
6:59implement on a live
7:01account, whatever the case might be,
7:02notifications, executions, you name it.
7:04This is a good starting point for we
7:06have a strategy and we want to see how
7:08it works before we take our next steps.
7:11So, when I start with my prompts, I like
7:13to specify with the deliverable. Well,
7:15that's one of the most basic things, so
7:17I want a back test results
7:22of the strategy
7:24given.
7:28That's what I want. What kind of format
7:30do I want it in? I want
7:33a
7:34PDF and a dashboard HTML format. The PDF
7:40is easy to transfer and the dashboard is
7:42easier to read itself. So, that's my
7:44specific deliverable. And you could
7:46click enter there, but you're going to
7:47get crappy results. Crap in equals crap
7:50out. And so,
7:52we're going to add context. And
7:55this is something that's changed a
7:56little bit with the development of
7:58recent models. It used to be the more
8:01context you give it, the better your
8:03results are. But as we've seen with some
8:05of these frontier models like Opus 5 and
8:07Fable 5 in particular, some of these
8:09models are giving more restraint.
8:11They're giving more false positives and
8:12more restrictions to where
8:14some of my students that I know you and
8:16other people as well have
8:18encountered where they're asking AI to
8:20do something for them, and then AI is
8:23not doing that thing claiming that it is
8:25a poor choice. It's overriding the human
8:28decision, which is a big problem. We
8:30can't have that happening. So, in some
8:33cases, not providing context that would
8:35lead lead to those scenarios is the best
8:38approach. If I say, I want to, you know,
8:42teeter on the edge of a prop firm rule
8:44and maybe breach an account, and I'm
8:46okay with blowing them, or I'm okay with
8:48getting right up to the line of the
8:49rules, AI is more prone to saying, "I'm
8:52not going to let you do that. I'm not
8:54let you going to proceed with that
8:57process. I'm not going to run this test
8:58for you because that is a bad idea."
9:00It's going to interject its own opinion
9:02and its own judgment, which tends to
9:04lead to a lot of issues, really.
9:07And so, context that's relevant is
9:09crucial. We need to have relevant
9:11context, but we need to be careful about
9:13the context we're providing. If I was a
9:15malicious bad actor, I don't want to be
9:16telling it that I'm a malicious bad
9:18actor because it's not going to do what
9:20I want it to do. Now, that's not the
9:21case for this, but it's tend to sway
9:24that way. So much so that Fable 5.1 that
9:26was just released was reported to have
9:2960% fewer false positives, and that's
9:32because people know that people are
9:33experiencing all of these really
9:35negative outcomes. So, Anthropic is
9:38trying to undo some of these poor
9:40restraints that we're seeing.
9:42So, but for the large part, context is
9:45important.
9:47There's a couple tools that we need to
9:48plug in. So, we have our strategy here
9:50that's important context. We need that.
9:52We also need historical data. So,
9:55we have a couple resources that we can
9:57use. Number one resource is I have a
9:59TradingView MCP bridge connected, so my
10:02cloud code can read directly from
10:04TradingView. So, I'm going to say this
10:07is a backtest
10:09that is being
10:11performed
10:13on the TradingView or it's being
10:17performed on
10:19Nasdaq, right, Nick?
10:21>> Yeah.
10:22>> Cool. Nasdaq.
10:25I want you
10:28to use the TradingView
10:33MCP bridge
10:35for as much
10:38data as possible.
10:42Then use
10:43external
10:45resources.
10:46So, because Cloud Code can have access
10:48to the internet, we can go and search
10:50for more information.
10:52TradingView MCP bridge is great for
10:54looking at live data, but TradingView
10:55has a candlestick limit count. So, in
10:58the case of this backtest, it's going to
11:00be performed on the 5-minute time frame.
11:03This is being
11:05performed, and that's relevant to the
11:07strategy.
11:12Not the 45-second.
11:18And the issue that this leads us to is
11:20because it's being performed on the
11:225-minute time frame, TradingView has a
11:24candlestick count limit. So, we may only
11:26be able to go back a month or two before
11:28we run out of data that we can see on
11:30TradingView. And a backtest that only
11:32has 50 to 100 trades is unhelpful in
11:36giving us results that are actually
11:37important and actually can help us, you
11:39know, make a reasonable future
11:41predictions. We look for hundreds of
11:44trades per back test. A trade with or a
11:47back test with only 50 or 100
11:49isn't really reliable, especially, you
11:52know, if you're looking to get into prop
11:53firms.
11:54>> [clears throat and cough]
11:56>> So, two of the external resources that I
11:58wanted to use, I want you to look into
12:03data
12:05Bentos
12:07free resources
12:12for futures NQ data.
12:16And
12:17if
12:18you run out
12:22of candlesticks,
12:24I want
12:26you to switch
12:28to
12:30This is important.
12:32CFD
12:34data
12:35for NQ.
12:37And there is another resource that we
12:39can use
12:40for this.
12:42This other resource
12:46is going to give us more free
12:48data.
12:49And why did I say CFD?
12:52Futures data is protected by the Chicago
12:54Mercantile Exchange and it has a data
12:57delay. The live is delayed. Um
13:01and CFD data
13:03is more open access. So, I can get
13:05historical data that's CFD, which is the
13:08exact same asset with minor changes, but
13:12an overall general pattern. And that's
13:14what we're looking for in these back
13:15tests is an overall general pattern. So,
13:18we can get a whole bunch of futures data
13:19as much as we can, and then we're going
13:20to switch over to the CFD data, which is
13:23going to be the same asset, but with
13:25slight slight shifts in the way that the
13:28data is being trans
13:30um transmitted between companies.
13:34So, the other free resource that we can
13:35look at is London
13:38Strategic
13:40is the final
13:42data source
13:44for NQ futures then NQ CFD.
13:50Okay, great. So, this is going to give
13:52us a pretty good back test in itself.
13:55We're going to get some good information
13:56here.
13:57But this
13:59isn't very specific.
14:01And if you just want generic data, this
14:03will get you there. But generic data
14:05gets you generic results and generic
14:06results in the prop firm space and in
14:08the trading space are that you're going
14:09to lose. So, if you want to go in depth
14:11and if you want to make money in this
14:13game, you need to be more intentional
14:14with your actions. So, this is good.
14:17This is a good start. But what kind of
14:19data is going to lead to better results?
14:21And this is where kind of skill and just
14:23being in the game for a while makes a
14:25big difference.
14:27So, what kind of things do I know that I
14:29care about?
14:30I know that I care about New York
14:32session in particular and I know that I
14:33care about New York session at 9:00 a.m.
14:36and 10:00 a.m. and that those can make a
14:37difference. So, I'm going to start
14:39specifying deliverables that I care
14:41about that I think will make a
14:42difference in the results and then when
14:44we get the results, we can see if my
14:46theory was right. So, what does this
14:48look like?
14:50I want data results comparing all time
14:55frames
14:56specifically
14:58the New York
15:01open
15:06from 9:00 to 10:00 a.m. and 10:00 to
15:1111:00 a.m.
15:14Those are the time frames that I found
15:15in my experience lead to some different
15:17results.
15:19So, we're going to ask for that and then
15:20I'm going to say because this is AI, we
15:22can just throw on an extra sentence and
15:24what would have used to take an hours
15:25and hours of additional work, one single
15:27sentence is going to get me way more
15:29results.
15:30So, 9:00 a.m. to 10:00 a.m.
15:3310:00 a.m. to 11:00 a.m.
15:35Also, compare
15:37all other time frames.
15:40The strategy spec may
15:43limit the trading
15:46window
15:48to be only the open.
15:51We want to test
15:54around the clock.
15:57Why? Because we can. And if we can find
15:59something profitable, why wouldn't we do
16:00that if it takes one extra sentence of
16:02work? The amount of output you can get
16:04from simply having an idea is ridiculous
16:07and we should be using that to its
16:09greatest extent and we're going to see
16:10that even more as we kind of dip into
16:13additional deliverables.
16:15So, that's
16:17one thing that we can add. I also want
16:19to add
16:21seasonal trends. We can see trends that
16:24are historically proven
16:27that every year over the past 10 years,
16:29the market tends to pattern pattern
16:32tends to follow a certain pattern. So,
16:36I also
16:38want seasonal [snorts]
16:40trends.
16:41How does the strategy perform month
16:47to month
16:49on a year-to-year
16:51basis? Are there certain seasons
16:54throughout the year that the strategy is
16:56going to excel for some reason? And
16:59to go even deeper, wars exist,
17:01geopolitical tensions exist and that
17:03does move the market. So, we can dig
17:05deeper into this and say
17:08identify
17:09large historical events
17:13that
17:15cause
17:16volume increases in the market
17:21and
17:22attempt to find correlations
17:26between
17:28these events
17:30and strategy performance.
17:38Great. Let's add something else. So,
17:40right now we just have one strategy and
17:41we're testing the one and that's great.
17:44But, why don't we test five variations,
17:4710 variations, 50, 100, 500? Because we
17:50can absolutely do that.
17:53I want
17:54and this is where there's two different
17:56ways of prompting.
17:58We can ask for specific deliverables and
18:00then give it context or we can give it
18:02context and pain points and ask for it
18:04to give us deliverables. And then this
18:07area of the prompt, I'm going to ask for
18:09it to give us deliverables. So, rather
18:11than me coming up with 500 different
18:12strategies on my own, I'm going to ask
18:14for it to take creative freedom and make
18:17variations of the strategy to see how
18:20they perform. What if we add different
18:22filters like volume, ATR, VWAP, EMA,
18:25trend lines? What if we look for
18:27liquidity? What if we're only trading on
18:29high impact news events, high impact
18:31news days? We can ask for it to test all
18:33of these different theories and ideas in
18:35mind to find edges that we can never
18:37find on our own.
18:38So,
18:40I want you to take creative
18:43freedom to search
18:45for edges
18:48using
18:51variations
18:53of the strategy
18:55you create
18:58on your own.
19:00This includes
19:04high impact news event days
19:09adding different
19:11confluence
19:13filters like EMA,
19:17VWAP,
19:20volume,
19:24day of the week,
19:29and
19:30any other variations.
19:34Let's actually add time frames, too.
19:37Time
19:39frame
19:42variations
19:45and more.
19:46I want you to take
19:49creative
19:51freedom to
19:53develop
19:54500
19:56different
19:58variations and test all of them on the
20:02data and
20:05provide comparative
20:11results.
20:13This is great.
20:15Now, I want to get more into what I want
20:18the actual presentation to look like
20:19because obviously, you know, we're going
20:21to get our results and it's going to be
20:22a spreadsheet format, but how do I like
20:24to see things? I like P&L charts. So, I
20:27want all of these on a P&L.
20:38Now, I'm asking specifically for P&L
20:41charts because that information is going
20:43to tell me a story.
20:44Because if we get 5 years of data and we
20:48can see the way that the charts are
20:50moving for years 1 through 4 is
20:52stagnant, then in year 5 it takes off,
20:57that might give me some interesting
20:58information. Or, more importantly, if we
21:01see that years 1 through 3 we take off,
21:04but years 1 and 2 we just regress,
21:08that tells us the story that this was a
21:10profitable model,
21:12but in the past 2 years it's yielded
21:14negative expectancy. But the data, just
21:17the win rate and the profit factor, are
21:19going to tell us it's profitable because
21:21on average it has been. But for the past
21:232 years it's been losing. So do we
21:25really want to implement a losing
21:26strategy over the past 2 years just
21:28because 2 years ago it was working?
21:32Probably not. And so getting a P&L chart
21:34that illustrates those results is
21:35important for us to read that story.
21:41Let's see.
21:43So this is good. This is a lot of
21:44context. This is a lot of deliverables.
21:46But we also must speak into the AI's
21:48personality. And this is more AI
21:50tooling.
21:51So
21:52we have a deliverable. We want to
21:53backtest and we want results in a
21:54dashboard and in HTML and a PDF format.
21:57We have What kind of results do we want?
21:59And we want to talk directly to it
22:02in terms of our experience.
22:04So
22:06you are This sounds silly, but it
22:08genuinely makes a difference in your
22:10prompting. You want to clarify the roles
22:12that you and artificial intelligence
22:14play. So you are a
22:17quantitative
22:19backtesting
22:20expert.
22:23I am, we'll say for the
22:26example, lacking in experience and
22:32terminology.
22:37Do not ask
22:40for my input.
22:43Run the tests.
22:46And this is This next These next two
22:50words will literally make a difference
22:53in your prompts. And think deeply.
22:56Take your time.
22:59These literally will change the quality
23:01of your outputs because it's going to
23:03read that and it's going to listen. It's
23:05actually going to change your outputs.
23:08So, this is a pretty good basic prompt.
23:09It's going to get you some results. So,
23:11I'm going to go ahead and run it. And
23:12another thing, requested by my student
23:15Nick,
23:16keep the strategy
23:19rules rather private.
23:23Do not
23:26reveal strategy
23:29criteria in the results.
23:33So, I'm going to be using Fable 5.1 for
23:36this as it just came out and it's a
23:38higher quality model. And although it is
23:41overkill, I would say for something like
23:43this, it's not using as many tokens as
23:45Fable 5 was. And since I have a higher
23:47level subscription, I'm okay to go for
23:49it.
23:50Um
23:51And let's
23:53turn it up to Let's go ultra code and
23:56see what we get. So, this prompt's
23:57probably going to take
23:59a considerable amount of time and then
24:01we can see the results. Nick, is there
24:03anything else that you want to add to
24:04this prompt? Is there any other
24:05information that you would find relevant
24:06or important?
24:09>> Um
24:12No, I don't I don't think so. I think
24:13that's a really detailed
24:14prompt.
24:16>> Great. Let's see what we get.
24:21Okay, the prompt is running. It's
24:23probably going to take a really long
24:24time. So, I just spoke with Nick and I'm
24:26actually going to go to the gym and then
24:28I'll get back and we'll see the results
24:29as it's cooking up.
24:52All right, I'm back from the gym and
24:54Claude code has finished. Now, this test
24:56was so intense
24:58that it ended up shutting off my Mac
25:01Mini halfway through, which is weird
25:03because the request goes to a data
25:05center. So, what happened
25:07to shut it off is it built so much code
25:10and ran so much local information that
25:14was enough to
25:15blow up my Mac Mini just with Python.
25:17So, it really went in depth. Ultra code
25:20is pretty overkill for this, but the
25:22stats we have are pretty ridiculous. I'm
25:25really happy with these. So, we're going
25:27to kind of walk through and look at what
25:28we got.
25:32So, 2,000 trades, which honestly isn't
25:35as many as I would have expected.
25:37We have our net P&L. We have
25:40year-over-year. What is our expectancy?
25:42We can see in 2023, huge loss, but 2024
25:45and 2025, huge gains. 2026,
25:49nothing spectacular, really.
25:52Average P&L per trade by entry hour.
25:55This is what I was talking about with in
25:56terms of experience. We see huge
25:58differences here between 9:00 a.m.,
26:0010:00 a.m., and 11:00 a.m. But, you
26:03can't take these things at face value.
26:04You need to look deeper. So,
26:07there's 900 trades at 9:00 a.m.,
26:111,200
26:13trades at 10:00 a.m.,
26:15and only 23 trades at 11:00 a.m. So,
26:18just these are actually quite
26:19misleading. I don't trust
26:21this to be as successful as it is. I
26:25want to see
26:26these higher
26:28volume trades
26:30giving me more information.
26:32By weekday, very interesting again.
26:34Wednesday, Thursday, Friday are
26:36profitable. Monday and Tuesday are not
26:38profitable at all.
26:40We have our by calendar. We have
26:43strategy variations. We have
26:49our P&L comparison.
26:51So, this is our risk in terms of units.
26:55Just unreal amounts of information here.
26:59We have all of the different strategies,
27:02their win rates, their profit factors.
27:03We have a 1.21 profit factor, which is
27:06actually pretty good. How many trades is
27:08this?
27:091,300 trades. So, this would actually be
27:11something reliable that I would like go
27:13look to live test and see if it's worth
27:15implementing into the next stages of
27:17maybe a funded account or even further
27:18than that
27:20being on a live capital account because
27:24the net R, the win rate, the profit
27:26factor are all things that I think are
27:29favorable.
27:31And this is
27:33every single strategy. So,
27:37these are all the different things that
27:39Claude Code came up with
27:41to find profitable strategies. And you
27:43see at the bottom here,
27:45look at all of these negative strategies
27:47that ended up just being completely
27:49filtered out. Zero edge here. Zero edge.
27:51Zero edge. Zero edge. And as we get at
27:53the top, we have some of our highest
27:55performing strategies
27:56up here.
27:59Around the clock
28:02clearly illustrates we should not be
28:04using this strategy in any other session
28:07aside from New York. Everything is red
28:11except for New York session, a small
28:14window within here. And that's
28:15important.
28:19New York open windows seasonality. So,
28:23March 2020
28:25performed horribly. That's funny. That's
28:27literally the exact time of the
28:29shutdown. So, clearly
28:31huge market events like COVID-19, that
28:34was enough to completely shake that up.
28:37And then 2026 extremely well in May.
28:43Interesting. So, we had a really big
28:44bull run this year in about May, which
28:47is when I made a lot of my money with
28:48prop firms, and this would have made a
28:50lot of money as well.
28:52Month by month.
28:55Month of the year affect all the years.
28:57So, November, December are the highest
28:59profitability times. August through
29:01October, lowest profitability times.
29:05Or March and April, actually.
29:08Month by month.
29:10So, this is extremely in-depth, and it
29:12just keeps going.
29:13Outrageous amounts of data.
29:16You can read a lot from this. Like, this
29:19is truly an exceptional amount of data.
29:22We have the method that I got
29:23everything. So,
29:24this illustrates the capability that
29:27we're able to get from Cloud Code. Now,
29:29I'm going to share all of this with
29:30Nick. This This is all going to be your
29:31information to play with and look at and
29:33implement. And I'm going to do that with
29:35him
29:36because this is exceptional, clearly. I
29:39mean, we're seeing event of the day. We
29:41can see that the election has
29:44>> [laughter]
29:45>> If there's an election going on, a
29:47presidential election, there tends to be
29:49a 100% win rate with the strategy.
29:52>> [laughter]
29:53>> So, the the the degree of in-depth that
29:56we're able to get from just clearly
29:59illustrating what we want from AI is
30:01absolutely insane. But, the information
30:04we're reading on the screen only matters
30:06if it's going to be something that
30:08you're implementing. Other else or
30:10otherwise,
30:11this is just entertainment.
30:14If you read all of this and you take
30:16nothing away, if you learn nothing, if
30:17nothing gets implemented,
30:20this was a fun project to enjoy and be a
30:23hobby.
30:26This needs to be
30:28information you take and do something
30:30with.
30:32Take it to the next steps. The objective
30:35of something like this is to find out
30:37how to make money. And if you're not
30:38moving forward with this, there was no
30:40point in doing at all.
30:44And if you're someone who's going to
30:45move forward and take this information
30:47and keep marching, this becomes
30:49extremely valuable. Because I'm looking
30:51at this and I'm saying, "Hmm, okay. So,
30:53clearly some of these top strategies are
30:56worth doing live testing. So, maybe I'm
30:57going to take these top ones and throw
30:58them into live. Or maybe I'm going to
31:00take them and throw them into prop
31:01firms." And I'm also going to see,
31:03"Maybe I should stop trading for these
31:04next 3 months and only do it in
31:06November, December. Or maybe I should
31:08get rid of the first 2 days of the week
31:09and only trade the last three."
31:12There's so many things you can take away
31:13and learn from this.
31:15And that's what makes this valuable.
31:18So, I'm totally pleased with these
31:20results. This is incredible.
31:23And
31:24this is this is why we're This is why
31:26we're doing things with AI. Because we
31:27did 2,200 trades in an afternoon. I, in
31:31my entire life of backtesting, have done
31:33maybe around 1,000 manual trades. And it
31:35took me forever.
31:38Really?
31:43And it took me forever
31:45to do those trades.
31:48I was spending hours and hours,
31:51probably thousands of hours,
31:53backtesting, live testing, hindsight
31:54testing, doing everything. And we can do
31:56it in an afternoon and get way more
31:58information, way more helpful
31:59information to move forward.
32:10Yeah, Nick. What What do you think of
32:12this, Nick? What do you think of all
32:12this information?
32:15>> Oh, I think it's great. It's really
32:16detailed. I really like that. I've
32:18honestly gone into this much depth.
32:20Like, yeah. It's clear there's
32:22correlations between like geopolitical
32:25events, which I never thought there
32:26would be with strategies like this.
32:28>> Mhm.
32:28>> Which is really interesting to see.
32:30>> Yeah.
32:31>> when I'm I normally look at metrics, the
32:33more or less statistical metrics like um
32:35losing streaks, winning streaks, things
32:37like that, but it never crossed my mind
32:39to keep a record of this.
32:41>> Mhm.
32:42>> But I can definitely be able to manage
32:43this. This is really good. This is
32:44really
32:45helpful.
32:46>> Yeah, this can be really helpful for
32:49even avoiding certain days, too, like
32:51OPEC's.
32:53That's a day to be avoiding.
32:56And FOMC, that's a day to take advantage
32:59of.
33:01CPI may be worth avoiding.
33:05It's really really interesting here. And
33:09if there's a a presidential election,
33:11maybe check it out.
33:16Yeah, this is this is incredible.
33:19>> The more info the better. You can just
33:20narrow it down and see what works and
33:22what doesn't. You just practice. It
33:23doesn't It doesn't hurt to
33:25gain less knowledge. Like less
33:27knowledge, you know?
33:28>> Yeah, not at all. Not at all. We have
33:33the tariff crash and how that impacted
33:35things.
33:37That's crazy.
33:39This stuff is This stuff is wild, man.
33:41I'm really pleased with these results
33:42and uh hopefully this can be helpful to
33:45take next steps. So.
33:49Just a a thank you to Nick for sharing
33:52his strategy with me to uh
33:54do this back test on. And I'm I'm
33:56excited to see what Nick can pull
33:58through next. This will be good. So,
34:00thank you, Nick.
34:02>> All right, thanks, little one. This is
34:03amazing.
34:03>> I think I'm going to continue to do more
34:05of these in-depth free value tutorials.
34:07I think they're really helpful to you
34:09and kind of to understand the process in
34:11which I'm doing things. And clearly,
34:13like this is extremely helpful
34:15information, but it's only as helpful as
34:19the degree to which you implement. If
34:21you're interested in building with me
34:23one-on-one directly, I have a mentorship
34:25you can apply down below. Cohort 2 is
34:26closing soon, but um there are still
34:29spots open if you are interested.
34:31Otherwise, let me know in the comments
34:32if you want me to keep doing these kinds
34:34of videos and kind of dropping free
34:36value like this because uh
34:38this is pretty fun and man
34:41it's it's crazy stuff.