Full transcript
0:14You know about AI. You've seen everyone
0:16using it. I'm using it for my day
0:17trading and it is improving my win rate
0:20and is helping me make more money. And
0:22that's what this video is for. This is
0:25going to be a value dump of loads of
0:28information that I've picked up, not
0:30only through other people, but through
0:32my own experience. If you don't know who
0:34I am, my name is Luke. I've been trading
0:36for over three years. I've been using AI
0:38for over four years, and I have over
0:41five figures in day trading payouts.
0:45I'm going to go over myths, vocab,
0:47reframe, variation searching, pointed
0:48test, prop clustering, which AIS to use,
0:51and your order of operations to
0:53implement everything you've learned.
0:55This is going to be a value dump. So,
0:57grab your STEMIs, grab your notebook,
1:00and get ready to take some notes because
1:02I'm going to give you as much value as
1:03possible so you can learn, implement,
1:07and get paid. That's what I want for
1:09you.
1:11myths. High win rate equals a good
1:15strategy. I get DMs all the time of
1:18people saying, "I have this win rate. I
1:20have a 60% 70% 80% 90% win rate." That
1:24does not matter.
1:27It's common in the space to see someone
1:29come out and say, "Oh, I have this 80 or
1:3290% win rate." But a 90% win rate
1:35doesn't mean anything without
1:37risk-to-reward. Let me give you a very
1:40clear example.
1:41Someone might say, "I have a 90% win
1:45rate." But 80% of the time
1:50they go
1:52break even,
1:54so they make no money, but they call it
1:57a win because they didn't lose.
1:5910% of the time
2:02they win, and maybe they win
2:05a onetoone trade. and then 10% of the
2:08time
2:10they lose a one to one trade.
2:14So actually a 90% win rate strategy that
2:18goes break even 80% of the time wins 10%
2:21and loses 10% is not profitable at all.
2:25On the flip side, if I even have a 55%
2:31win rate and I win 55% of my trades, not
2:36break even, actually winning 55% of the
2:40time,
2:43a onetoone trade, and I lose
2:4845% of the time, a onetoone trade.
2:53This is more profitable.
2:55So just because you have a high win rate
2:58does not mean you have a good strategy.
3:00People online will be very foggy about
3:03what their win rate actually is because
3:05they will classify break even trades as
3:07wins because they simply didn't lose. So
3:10no, a high win rate does not mean you
3:13have a good strategy. Riskto-reward is
3:15essential because even if I change this
3:18to just one to two now this is better
3:24AI will find me a profitable strategy.
3:28I have in the past two weeks back tested
3:3220,000 different strategies and taken
3:35around
3:3717 no sorry 13 million trades. 13
3:42million trades.
3:44Yes, there are profitable strategies in
3:47there. Some of which I'm implementing
3:49starting this week. It can do that for
3:52you.
3:54More historical data [snorts] is better.
4:00No, it's not better.
4:04And we'll get more into that on when to
4:06test. But if I find a profitable
4:09strategy from 2020 and it's profitable
4:13for three years straight,
4:16but it made no money in 2024, 2025, 2026
4:20when the new regime hit, that means
4:22nothing. You can't make money with a
4:24strategy that's not making money
4:26anymore.
4:28So, no,
4:30more historical data is not always
4:32better. Why are you back testing back to
4:342010? The market doesn't move like that
4:36anymore.
4:39Profitable and back test means
4:40profitable and live. No,
4:45because again I can have a strategy
4:49that is profitable for a series of years
4:52in a row. Let's draw this out.
4:56Here's our P&L over time. Here's 2026
5:01and here is we'll say 2016,
5:05right?
5:07So from 2016 to 2026 we do this.
5:16So we made a lot of money up until about
5:192021
5:20and then we were break even or maybe
5:23even on a decline for the past five
5:25years. But if you look at simply the
5:29average of the decade,
5:36it looks profitable,
5:39but you don't know it's profitable
5:40anymore unless you're getting live
5:42results. So more and more I'm actually
5:45having a strong recency bias. Is it
5:48profitable within the last six months?
5:50Because if you've looked at the market
5:52for more than the month, you know that
5:53it's changing quick.
6:02Strategy is the hardest part.
6:14Not anymore.
6:16I would argue that it was
6:19when we had limited resources and
6:21information was guarded and it was
6:22protected and you had to buy people's
6:25courses to figure out what they were
6:26doing and learn like what's your secret
6:28sauce but you can find any strategy. Do
6:31you really think that I came up
6:33personally with 10,000 strategies or did
6:35I just go to Claude and say do some
6:37research for me?
6:40Strategy is no longer the hardest part
6:42anymore. We can find good strategies
6:46really easily. Tons of my students are
6:49killing it. Finding amazing strategies.
6:52It is a game of implementation and
6:54knowing the game. Because just because
6:56you can find a good strategy does not
6:59mean you can make money with it,
7:01particularly with prop firms. Prof firms
7:04are a different game than normal day
7:07trading with your live capital because
7:09most people don't do that. I traded with
7:12my life capital this summer. It was a
7:14very different game than it was with
7:15prop firms.
7:18Do I need to learn to code to use AI?
7:24Nope.
7:26That's the whole point of cloud code.
7:28That's the whole point of the barrier to
7:31entry being dropped down to zero because
7:33you no longer need to know coding
7:35languages. You no longer need to know
7:37Python or be this coding expert. The
7:39barrier to entry is now at zero. What
7:43you do need to know is how to clearly
7:45communicate your ideas in English so
7:47that AI can interpret what you mean. If
7:50you cannot clearly communicate what your
7:53strategy is or what you want or the
7:55right questions to ask, these are the
7:58skills of the decade. If you want to get
8:00good at using AI, this is prompt
8:02engineering. asking the right questions,
8:04analyzing the results you get and then
8:07moving further. That is the skill.
8:12You do not need to learn how to code
8:13anymore.
8:17Vocabulary.
8:19So, there's a lot of popular terms that
8:21get thrown around and a misunderstanding
8:24of these is going to leave you lost. And
8:26if you don't know what they mean,
8:29you're going to be in a lot of trouble.
8:31And some of these go into prop firms as
8:34well because we need to understand some
8:36of this vocabulary because profirms use
8:39it all the time.
8:41We start with a basic win rate.
8:45This is the most common is why you click
8:50this video. Win rate. How many trades
8:52are you winning? 90% means you're
8:54winning nine out of 10 trades. 50% means
8:56you're winning five out of 10 trades. It
8:59is just what percentage of trades you're
9:01winning
9:04percentage
9:07of trades
9:11one
9:13per
9:15sample.
9:19Per sample is important because if I
9:21test five trades and I won four and I
9:23have an 80% win rate. I mean you do
9:26technically speaking but your sample
9:28size was not enough. You need a bigger
9:31sample size to actually trust the data.
9:34Riskto-reward RR.
9:40So if I have a one to two riskto-reward
9:46that means I am going to risk
9:48$10 to make
9:51$20.
9:53If I have a two to one risk-to-reward,
9:58that means I'm going to risk
10:01$20
10:02to make $10. And both of these can work.
10:06There is such a strong bias towards, oh,
10:09you've got to be at least doing one to
10:10two, at least one to three, at least one
10:12to 1.5. Negative risk-to-rewards
10:16can be so profitable. I know traders who
10:19have made their entire brand off of
10:22negative risk-to-reward and they kill it
10:24in prop firms. They make so much money.
10:28This is just as good as this if you have
10:30a win rate that corresponds with them.
10:33Max loss limit. This is getting into
10:36prop firm terminology.
10:41So, when you see a prop firm say $25,000
10:44account or $50,000 funded account,
10:47that's margin. That doesn't matter. Your
10:51actual account size is what your max
10:53loss limit is. So, when you say a
10:55$25,000 funded account, you have a
10:57$1,000 max loss limit. It is a $1,000
11:01funded account because that is all the
11:03money that you have access to using
11:09not 25k funded. You have a 1k
11:14max
11:16loss
11:17limit
11:19because that is all that you have
11:21available.
11:23Those are flashy numbers that represent
11:25the margin which is rather irrelevant
11:28because it's simulated funds. Anyway,
11:31this is a big one and this is important
11:33[snorts] to know the difference. End of
11:36day draw down versus intraday draw down.
11:40And I'm actually going to make some
11:41space here to make this clear.
11:45So these are two different types. And a
11:48lot of times with certain profs, you'll
11:50see if you just check a box and switch
11:52it from end of day to intraday, you get
11:55a discount. And you get that discount
11:56because intraday sucks. And
11:58[clears throat] in my opinion, you
11:59should never do intraday. And here's
12:01why.
12:13So at the end of the day, if you made
12:15$100 on your funded account or on your
12:18evaluation, your max loss limit
12:21increases. So let's let's take the
12:24example of I have a $1,000 max loss
12:27limit. So max loss limit equals
12:33$1,000.
12:34Day one,
12:38I make plus
12:42$100 for the end of day trailing and the
12:47intraday.
12:52My max loss limit is going to go from
12:54a,000 to
12:58900.
13:02So, same thing, right? There's no
13:04difference yet. But what if on day two
13:11I lose $100.
13:15Okay. So, our balance I'll make a
13:18balance side to keep things clear as
13:20well.
13:21Our
13:23balance after day one is $100
13:28and on the next day it's $0.
13:32But in the middle of this trade,
13:37we actually went up $75
13:41and then we went all the way down to
13:43stop loss and the trade yielded negative
13:46$100.
13:48because we went up $75.
13:52Our max loss limit for intraday trailed
13:56with it. It kept going. It's calculated
13:59every second based off of your
14:01unrealized profit and loss. End of day
14:06is calculated
14:08end of day profit and loss. So for this
14:11number, it doesn't change because at the
14:13end of the day, you lost $100. So for
14:16your end of day max loss limit, it's
14:18going to be 900.
14:21But for your intraday,
14:24because you had an unrealized profit and
14:26loss of $175,
14:29at one point you didn't have to click
14:31buy or sell, but at some time during the
14:32day, you had $175. This has now trailed
14:35to
14:38$825.
14:41And now you have less room.
14:45And that gets a lot scarier the more you
14:47size up. So almost never could you ever
14:52go for intraday draw down especially on
14:55fun accounts when you're being more
14:58cautious to get payouts.
15:02Consistency rules.
15:04These have come about in the past one to
15:08two years before that. They weren't here
15:10and people made a lot of money because
15:12they weren't here.
15:33This is more common on fun accounts,
15:36less common on evaluations. You'll see a
15:38lot of single day or 50% day
15:42evaluations.
15:44So, a consistency rule is a rule put in
15:47place to stop you from hitting home
15:50runs.
15:52I always come back to the memecoin
15:54example for this because that's how
15:56people got rich in the memecoin era
15:58because there's no such thing as
16:00consistency. First off, people weren't
16:01doing funded accounts that people were
16:03trading with their own live money, but
16:05there was no consistency to stop them.
16:07So, their win rates would be 1% or less,
16:11but they would hit one home run for a
16:14one to 1,000 risk-to-reward and get rich
16:18off of a single trade.
16:20a consistency rule would stop you. So
16:26let's take an example for the evaluation
16:28stage.
16:34Let's say the eval has a 50% consistency
16:37rule.
16:4150% consistency means you cannot have a
16:43single day that is more than 50% of your
16:46profit target. So, if my profit target
16:49is $3,000,
16:54a single day cannot be higher than 1,500
16:59or 50% of this amount.
17:06One of my students ran into this problem
17:08just the other day before passing his
17:10account.
17:13So,
17:16my student
17:18on his Eval
17:21had an automated trade get placed and he
17:23made
17:25$1,600.
17:28Now, that doesn't mean they breached the
17:29account, but this target all of a sudden
17:32moved. So, that 50% consistency still
17:36applied. So for this to be half of the
17:40profit target, you double this value.
17:48So when he had a day where he made too
17:50much, he did too well, he was hit with
17:54this percentage and his profit target
17:56moved
17:58to $3,200 and he had to make more in
18:01order to pass that account.
18:06Now, on the funded stage, these
18:09consistency rules tend to drop lower.
18:12And I'll give you the reasoning behind
18:14why that happened
18:17because again, it wasn't like this. Most
18:19prop firms didn't have consistency
18:21rules. So, most are now 40% on the
18:25funded stage. And before that,
18:31you would have traders who would pass
18:33the eval get on the funded stage. they
18:35would have one day where they fullport
18:37the whole account and make $4,000 on a
18:40super small funded account. So, they
18:42made 4,000
18:45and the payout rules say you have to
18:48have three winning days. So, day one
18:53they made 4,000 and they only need two
18:55more winning days before they can
18:56withdraw a whole bunch of that money. So
18:59on days two
19:01and day three,
19:03they only make
19:07150 bucks
19:11to meet the bare minimum request 2,000
19:14to 2,500, however much, and the eval
19:17only costed them 50 bucks. So they take
19:20that 2,000, they go and recycle it and
19:22get 30 more evals and do the same thing
19:24again. But a 40% consistency rule means
19:27that your balance
19:30must consist of days
19:33that do not exist higher than 40% of
19:37your target. So now the case is if you
19:40have a $4,000 day on a prop firm,
19:47this can only be worth 40% of your total
19:50balance.
19:51So, in order to get paid now, you would
19:54have to do that again.
20:02But if I have my another $150 day,
20:06this is only 50% consistency. This
20:09doesn't even meet the 40% yet.
20:12My math is starting to get challenged
20:14here.
20:16But what you would really need
20:31Okay, we're almost there.
20:34Math on the spot, baby. Come on.
20:36Mechanical engineering degree be getting
20:38put to use.
20:40All right.
20:4340 45,000.
20:47I don't think this is even enough still.
20:49We're going to bump this up to 1500.
20:54Okay,
20:55so you understand hopefully by now. The
21:00total balance at this point is 10,000
21:04500.
21:08[laughter]
21:10Double check the math, someone please.
21:14[laughter]
21:16A consistency rule means your biggest
21:20day cannot be worth more
21:24than this percentage of your total
21:27balance.
21:29That is what that means. So, as a
21:32general rule, forget my poor math
21:34skills. If I have a total balance of
21:38$1,000
21:41and I need to meet a 40% consistency
21:43rule, no single day
21:50can be more than
21:55$400.
21:59It has to be $400 or less. You could do
22:03333 * 3. You could do 350 350
22:10300.
22:12But it cannot be worth more than 40%
22:15of your total balance. So you'll see on
22:18some prop firms
22:20you can do a straight defunded and you
22:23can pay a couple hundred and skip the
22:26eval stage. How awesome is that? You can
22:28get paid so fast. but they'll throw a
22:3020% consistency rule on there. And that
22:34makes it so hard to get paid. 20%
22:38consistency means you need five winning
22:40days of the exact same.
22:45And you can't lose a single day because
22:47if you lose a single day, you need to
22:49win again to make that up and win again.
22:52So, you would need six winning days of
22:54the same amount,
22:57which is an outrageous win rate. And the
22:59final vocab that's super important and
23:02gets into reframing
23:04is profit factor.
23:08This matters more than anything else
23:12because when we're looking to find our
23:15profitable strategies, we're looking at
23:18profit factor. and the profit factor
23:20combined with different win rate
23:22combinations and risk-to-reward
23:23combinations is how we decide how we
23:26approach our prop firms.
23:28So
23:30a 1.0
23:35profit factor
23:37means you're break even.
23:400.5
23:42means you're losing
23:450.5 of your risk per trade.
23:501.5
23:51means you are making 1.5 of your risk
23:55on average per trade. So
23:59if I had $1,000
24:03and I shed it through all of these,
24:08I would be left with $1,000,
24:12$500,
24:14and $1,500.
24:17This is essentially your multiplier
24:20for your starting balance.
24:23So, whatever amount of money you're
24:24starting with, you multiply by your
24:26profit factor and then that's the
24:28balance you're left with after your
24:30sample size. So, a lot of the strategies
24:33that I'm seeing right now and that I'm
24:34actually using are only like 1.2 or 1.4.
24:38I have this crazy guy in my community
24:40who somehow cracked a mechanical 1.9,
24:43which is awesome. If you have a profit
24:46factor over 1.5 or even at 1.5 and
24:49there's over a 100 trades in that back
24:51test sample and it's all within a recent
24:53amount of time, you're on to something
24:55good. Keep going.
25:08[snorts]
25:12So both of these
25:14Now we're on to reframing because you
25:17clicked on this video because you want
25:18to increase your win rate with AI and we
25:20can do that but that doesn't mean
25:22anything. What we really want to do is
25:25we want to increase our profit factor
25:27and we have three different levers that
25:29we can pull with AI to do that.
25:33So win rate up
25:38is not what we're looking for.
25:41It can be depending on what stage of an
25:44account you're at, whether you're on a
25:46holy
25:48whether you're on a funded account or an
25:49evaluation. But this doesn't mean very
25:52much.
25:54What we're really looking for is our
25:57profit factor
25:59to increase.
26:05This is the best summary of whether or
26:08not you have a successful strategy.
26:12If this is less than one, you do not. If
26:14this is greater than one, then you do.
26:17And we want this to increase. And we
26:18have a couple different levers we can
26:19pull using AI to figure that out.
26:24Those are
26:26number one,
26:28filters.
26:32So much of finding good trades in a
26:35sample size is getting rid of the bad
26:38ones. Let's filter out all of the crap
26:41and just find the best ones. And that
26:44can be a variety of things. Indicators,
26:47trend lines, VWOP, EMA, RSI, volume,
26:50like you name whatever you want. But we
26:52can apply filters
26:55to our strategy
26:57to determine
27:00what leads to a higher profit factor.
27:04The next thing that we can do
27:08is we can change our entry
27:12and our exit.
27:16At what point when our criteria for our
27:18strategy is met are we entering? Are we
27:22doing limit market time? What time
27:25frame?
27:27How far is our take profit? How far is
27:29our stop loss? Do we change point
27:30values? Should we do timebased?
27:35Can you imagine that? How revolutionary
27:37is that in the day trading space to
27:39imagine entering a trade and then just
27:42starting a timer and waiting till the
27:44timer's up to exit your trade? That is
27:47so unorthodox.
27:49But what if you just ask the question in
27:53her profit factor increases?
27:55Then that's a viable approach.
28:00Ask more questions to Claude.
28:06And the final
28:08is when
28:12this is volume based as well, but
28:15certain strategies are going to do
28:17better at certain times. If you have
28:19some consolidation strategy that
28:21requires really low volume and you're
28:23applying it to the New York Open, have
28:26you even thought of asking AI to apply
28:29it to Asia session open because there's
28:31not a lot of volume during that time?
28:34What about news events at 8:30 a.m. EST?
28:39Does that short window allow for massive
28:43profit factor strategies?
28:47The time of day does play a role. And it
28:50plays a role because the volume changes.
28:53Not because the market makers are waking
28:55up and having their morning coffee, but
28:57because you can literally see the volume
29:00throughout the day routinely change.
29:02Asia recession is low. London's a little
29:04bit higher. 8:30 a.m. news, we get
29:06massive spikes. New York open, 9:30 a.m.
29:09every day, we get a spike. We keep
29:11going. 2:00 p.m. on a Wednesday every
29:14month, we get a huge spike. And then
29:162:30 someone talks and we get random
29:18spikes. And then at 9:00 p.m. someone
29:20tweets on Twitter and then it blows the
29:22whole market up bigger than all of those
29:24events.
29:28These are your three levels levers to
29:30pull
29:32to increase your profit factor.
29:38Variation searching.
29:41So, when I say I tested 10,000
29:43strategies, which you're going to see in
29:45a video in a few days,
29:49I did not go and find 10,000 individual
29:53strategies.
29:56And you saw this on my conversation with
29:59Nick's client call posted a week ago.
30:04What we do is we have a core strategy.
30:08So, let's let's take an example.
30:11Right. We'll go with
30:14an acquaintance who's probably familiar
30:17to you.
30:18We'll go with this strategy.
30:27This strategy sucks. I'm just kidding.
30:30It's all right. But we want to see
30:35how we can increase our profit factor.
30:38And to do this, we can start testing
30:40variations. So, we can start start
30:42testing. We might take this on the one
30:44minute time frame.
30:46Why don't we test it on literally every
30:48time frame that exists? Because you
30:50might be taking it on the one minute
30:51every time. What if it's more profitable
30:53on the two-minut? Because it slices your
30:55quantity of trades in half, but every
30:57trade is more likely to win. That makes
31:00you more money. It's lower frequency.
31:03And that can be hard to stomach. a
31:04strategy that's lower frequency and a
31:07higher profit factor, but if it makes
31:09you more money and that's the goal,
31:11you're having to resist your gambling
31:13urges. That's one of the two secrets to
31:15profitability is separating yourself
31:16from the desire to make money. If you
31:19can separate yourself from that desire
31:21and simply follow the math, you're going
31:24to make more.
31:26So,
31:28call this time frame
31:32expansion.
31:36What if we also test seasonality
31:39expansion? There are routines year after
31:42year of how the market moves. I believe
31:44you can look this up, but I believe it
31:46looks something like this. From start to
31:48end in the past decade, the average year
31:51to year looks like this. So like
31:53January,
31:55December, you can fact check me looks
31:58something like this.
32:05You can verify whether or not that's
32:06true. But seasonality plays a role.
32:12So we can test variations
32:15month
32:17to month or even week day to week day.
32:22You'd be amazed
32:25how big of a role
32:28the day of the week can play.
32:35Then we can test our filters.
32:38ATR, VWOP, EMA,
32:42trend line, whatever you want. Add it on
32:44there. Ask the question.
32:47What else can we test? We can test news
32:50events.
32:53There are routine news events like
32:56unemployment claims at 8:30 a.m. once a
32:58month that will shake the market. You
33:00can you can bet on it that this is maybe
33:04it's not going to make a massive move,
33:06but it's going to inject a lot of volume
33:08and volume affect strategy.
33:11So for just these five,
33:15we can take five different news events,
33:18five different filters. We have five
33:20weekdays, we have 12 months, we can test
33:23out 20 different time frames. So we've
33:25just taken one strategy and we can now
33:27back test and create 100 plus different
33:30strategies around the same core idea.
33:33So, when I'm saying I'm testing 10,000
33:35strategies, I took 50 core ideas, made
33:39200 variations of each one, ran them
33:41through to see the most profitable ones.
33:44This is what variation searching is, and
33:47then you'll get a list of all these
33:48strategies, and then the ones with the
33:49highest profit factor, that's where your
33:51biggest profitability is going to lie.
34:07When to test
34:18I had a guy the other day message me on
34:20Instagram and say he was a little
34:23disappointed because he was struggling
34:26to find candlestick data from 2010.
34:42If you have a profitable strategy
34:44in 2010
34:47that does nothing over the next 16
34:51years, that is completely unhelpful. I
34:55drew the example in the beginning. I'll
34:56draw it again. This
35:01is technically profitable.
35:06But if you started any time after 2011,
35:12let's just say we started here
35:16and we started using this strategy,
35:20we have lost money year over year over
35:23year over year over year.
35:40but we want a a higher trade frequency
35:42and that's fine. So, this is where it
35:44starts to get really dependent on what
35:46kind of trading you're doing. And now
35:48I'm going to assume that a lot of people
35:49in here are doing intraday trading
35:51because that's the best approach for
35:53prop firms.
35:55And you do want a higher sample size. So
35:57if you're only taking 20 trades a year,
35:59yeah, you're probably gonna want to go
36:00back and look at the last five years
36:02because in my personal opinion, in my
36:04experience, I want at least 100 trades
36:07per sample size before I trust the data
36:09that I'm reading. I need a higher
36:11frequency. And we'll see that how that
36:13affects prop clustering.
36:17So when to test,
36:19I tested on my multi-million back test,
36:22seven years of data. And then what I did
36:24is I looked at the most profitable
36:26strategies and I looked at a chart like
36:28this. So we have our baseline of
36:32profitability here.
36:36And then I started looking
36:41year-over-year. What's the performance?
36:43And I saw interesting important data. I
36:46would see things like this.
37:00This gives me an important story.
37:03This tells me that in 6 54 2023 it was
37:09very profitable. 2024 very profitable.
37:122025 it was not very profitable. Now
37:15what could that be from? Maybe we zoom
37:16in more and we see that the tariffs that
37:18happened in 2025 in April
37:21totally bombed this strategy, but we see
37:24that a little bit of a smoother market
37:25in 2026 led it to do very well.
37:31All of a sudden, we can rely on this a
37:33lot more. So, the two things to keep in
37:36mind when you're looking at your amount
37:38of historical data to test are trade
37:40frequency
37:43and a recency bias.
37:45Because
37:47with a lot of confidence, I can tell you
37:50that in another two years when we have a
37:53different presidential regime, the
37:55market will not be moving in 2029 like
37:58it is in 2026.
38:01That is a reliable claim to make. And
38:04the same is true today. A strategy
38:07profitable in 2022
38:10is unlikely to be profitable today.
38:13And it's that same drawing again. The
38:15only thing we need to be sure of is
38:17getting a high frequency because if you
38:18take 10 trades a year
38:21or if you only test the past week and
38:24you got 50 trades,
38:27maybe let's test a couple weeks and see
38:29how we're doing. I'm taking an approach
38:32this coming week
38:34on some of my own strategy that I'm
38:36testing. And I have all these different
38:38charts, right?
38:41and I'll take you kind of through my
38:42decision- making process on these charts
38:44to really get this point across. Now,
38:47all of these strategies are profitable,
38:48tested with AI, they all look great, but
38:51I need to make decisions.
38:54So we'll take three or
39:00so every strategy is ending in profit
39:05and I'm going to be using them this
39:08coming week in automated executions with
39:10Claude. I'm not watching the charts
39:12anymore.
39:15But one of the strategies did this
39:19over the past six months.
39:23Okay, it's profitable. It got there.
39:26Stats are great. But in the past two
39:28months, we've been kind of rough. And if
39:30I'm only looking to squeeze a lot out of
39:32it in the next week, maybe I'm not a big
39:35fan.
39:37Another strategy is something like this.
39:40Okay, you're really reliable. I love the
39:43way you look. Like, that's that's sweet.
39:45That's awesome. Six months. This is
39:47beautiful. I love it.
39:51Then we have this
39:57Okay. Interesting.
40:00So, we're not making a lot over the past
40:01four months, but in the past couple
40:04weeks of market conditions, it's been on
40:07fire.
40:08So
40:10assuming that things continue on their
40:13current path, applying what did well
40:16over the past three weeks to the next
40:18week
40:19in theory
40:21should yield better results.
40:24And I'm placing a bet on that this
40:27coming week because I'm looking for only
40:30this next range.
40:35I'm looking to see what can make the
40:37most money
40:39right here.
40:44And so if this continues its slope,
40:47we're going to crush it. If this
40:49continues, we'll probably do pretty
40:51well.
40:53If this continues, I don't know if we'll
40:55make anything.
40:57Recency bias is important because the
41:00market changes fast and it's changing
41:03faster and faster over time. We can see
41:05it. It's not a secret. It's not some
41:08mystery. We It's literally in front of
41:10our faces. We can see it changing. Zoom
41:12out on the weekly chart on the S&P 500
41:17and you'll find that from 2010
41:22to 2016 or to 2026
41:28that the S&P 500 looks like this.
41:36Things are [snorts] changing.
41:39That's why recency bias on the trades
41:41you're taking is more important than
41:43ever.
41:52But it has to be a high trade frequency.
41:55You cannot back test 30 trades and say,
41:59"Oh, my data set's great.
42:03Unless you're looking at a two-year time
42:05frame and you're looking to swing trades
42:07over a month on stocks, that is
42:10unreliable for day trading.
42:15Prop clustering. Prop clustering. This
42:18is for more of those prop farmers, the
42:21people who spend a whole bunch on
42:23property valuations, blow a whole bunch,
42:25get a bunch of payouts really fast.
42:27That's what I'm going to do this next
42:28week.
42:30And the question is, should you copy
42:33trade them?
42:37So, let's say I have 10 accounts
42:45and let's let's do an analogy, right?
42:50I'm going to flip a coin
42:54four times.
42:56Okay?
43:03My engineering statistics is going to
43:05come helpful here. So, I'm going to flip
43:09it four times. If I get one, two, three
43:13heads and one tails,
43:16how shocked would you be?
43:20Probably not that surprised.
43:24It's such a small sample. I mean, like
43:26that can happen. like that's not that's
43:28not extremely unlikely to happen. Now,
43:32what if I flipped it a 100 times
43:35and then instead of
43:3950/50, I got
43:4375 heads and 25 tails. How crazy would
43:47that be? Okay, now
43:50that's a little alarming. That says to
43:52me something's wrong with this coin.
43:56This isn't playing out the way I would
43:58like. This isn't playing out the way I
44:00would prefer. This is not probabilistic.
44:04It's leading to me leading me to believe
44:06something is wrong.
44:11So, should we copy trade our
44:13evaluations? Should we copy trade our
44:15funded accounts? Now, this depends on
44:18your approach. Again, a lot of this is
44:19dependent. That's why I have a
44:21one-on-one mentorship because specific
44:24conditions require specific approaches.
44:28It's not always a one-sizefitit all, but
44:30this is a good generic rule and applies
44:33to my case and
44:35a degree of other people's cases.
44:40So, we'll say we'll actually make this
44:4212 accounts.
44:4512 accounts. I'm going to divide them
44:46into three groups of four.
44:50and I'm going to copy trade four at a
44:52time.
44:54So, back to the coin flip example. If I
44:57have a 50% win rate with a 1.5
44:59risk-to-reward, and I'm trying to pass
45:01these in single days, how crazy would it
45:03be for me to lose one trade, lose a
45:07second trade,
45:09and then win the third?
45:13Because although I have a 50% win rate,
45:16which we know is true, this should be
45:1850%.
45:20But it is not insane to see numbers like
45:24this if we flip a coin three times.
45:28That is bound to happen. And if we do
45:31that here, then all of a sudden our 50%
45:34just turned into 33% because we were
45:36copy trading. That's not unlikely.
45:40The more samples that we give it, the
45:42more flips that we give it, the closer
45:44we are going to get
45:47to 50%. More flips,
45:53the closer we should in theory
45:54statistically get to our actual number.
45:58And so we if we have a backtested 50%
46:01pass rate,
46:03we want to flip it as many times as we
46:06can because the more flips we do, the
46:09closer we get to this number. So instead
46:10of doing 4, four, four, we do something
46:13like one, one, two, three, four, five,
46:176, 7, 8, 9, 10, 11, 12.
46:22Right? So maybe we get that still same
46:25playout. loss, loss, win, right?
46:31That could happen. And maybe we get
46:34another loss and another loss and then
46:37we get a win and a win and a win and a
46:40win and then another loss
46:44and then two more wins or a win and a
46:47loss.
46:48All of a sudden, since we kept flipping
46:51the coin, we got a lot closer to our
46:5350%. because we were flipping it more.
46:58That's six losses and six wins simply
47:00because we kept going and we increased
47:02our sample size. So when you're copy
47:03trading evaluations and you're trying to
47:05pass as many as you can, you want to in
47:08a high frequency environment, not be
47:11copy trading them.
47:13You want to give it more opportunity to
47:15reach that 50%. Doing this in a high
47:19frequency setup means you're gambling
47:22more and you're placing more at stake
47:25for no reason other than you're
47:26impatient.
47:34And then the same logic applies to
47:36funded accounts.
47:41This is again more applicable to high
47:43frequency trading. If you're only
47:45trading one time a week, you better have
47:47a high win rate or a really high profit
47:49factor and you're risking less on Eboss.
47:54Which AI?
47:56So, right now,
48:00today is September 19th or 20th, I
48:03think.
48:04Some of our top tier winners are Fable
48:105.1
48:12and Astra
48:166. This is Claude.
48:22This is chat
48:24GPT.
48:26Now, you can go and you can do your
48:28research and you can experiment with
48:29these.
48:31I'm going to talk based off of my
48:33experience, what I'm doing, and how I'm
48:34using these.
48:37These are awesome.
48:40These will get you everything you need
48:42and more. In fact, you're probably not
48:44even getting the most out of them.
48:46They're probably overkill.
48:49These are your CEOs,
48:52and you're going to eat up your usage
48:54when you're using these.
48:56You do not want to tell these models to
48:59write you a cat poem that is a waste of
49:01tokens. You want to tell a model like
49:06this is still overkill, but it's to make
49:08a point. Really, you'd want to tell a
49:11model like Haiku, which I think is like
49:14in its fourth generation,
49:16which is one of Claude's models, that's
49:18going to write a good cap home because
49:19it's low tokens, it's low usage, and
49:22writing a cap home is really easy to do.
49:25Now, making judgments based off of
49:27complicated strategies and market
49:29regimes. Let's use our highest model
49:31because why not?
49:34But if you're conservative on your
49:36usage,
49:38this is your these are your CEOs
49:42and they're your CEOs because they're
49:43really good at judgment and decision-
49:45making.
49:48And then I'm not too familiar with
49:50Codeex because I don't use it
49:51frequently, but Opus
49:564.8. Opus 5 is out, but Opus 4.8 is
50:00better. And I'll explain why in a
50:01moment.
50:07Opus 4.8 is your code monkey.
50:10And it's your code monkey
50:13because it's going to do a really good
50:14job at agentic coding for you and
50:17getting the point across and completing
50:19the task at hand without eating up all
50:22of your tokens. and it's going to do
50:24almost an identical job as some of these
50:26better models.
50:28And I say Opus 4.8 and not Opus 5
50:31because of ethical reasons. Opus 5 has
50:34been clearly shown and in my experience
50:37as well to give far more increased false
50:42positive rejections. Meaning you are
50:44saying hey Opus 5 enter this trade for
50:47me and it's saying no. and you're
50:49saying, "Oopus 5, build this algorithm
50:51that's going to automatically execute or
50:52run this back test and then it's feeding
50:54back to you. This is a poor choice and
50:56I'm not going to do that for you."
50:59Whereas Opus 4.8 is nearly it's
51:03marginally lower in quality of code, but
51:06it's not denying your requests. So,
51:08you're going to get a nearly
51:10unrecognizable difference in your
51:12result, but it's going to finish the job
51:14that you want. So, Opus 4.8 is going to
51:16be your code monkey. So really my
51:17pipeline is I'm asking for prompts from
51:21Fable 5.1 and feeding them to my code
51:24monkey that does all of it for me with
51:26less usage and still really good results
51:28that are not denying my requests.
51:34Same with Astra 6. These your decision
51:36makers. They have incredible depth and
51:40thinking, highly agentic.
51:44We don't need to use them all the time.
51:48In fact, in some cases, it's better not
51:49to use them because we can get our
51:51results faster and then loop our results
51:54quicker.
51:58So, don't give Fable 5.1 a cat poem.
52:16The final one.
52:39Order of operations.
52:45So that was a lot of information
52:48and a lot of the reason I have my
52:50mentorship is not because you are
52:53incapable of doing these things. It does
52:56take work. It does take investment. It
52:58does take time. If you join my community
53:03and you never show up, you'll never get
53:05results. If you join my community and
53:08you're asking me questions and you're
53:10seeking guidance and you're following my
53:13system, you will make progress and get
53:16closer to getting paid. That is what I
53:18want to happen.
53:22If you are interested in personal
53:24guidance from me, I work individually
53:26personally with every single one of my
53:28students. It's not a drop off of all
53:31right, go figure it out. It's no, where
53:34are you at? Tell me where you're at and
53:35let's keep going.
53:37But in this free sauce video,
53:43your order of operations should look
53:45[snorts] something like this.
53:48You should start with strategy.
53:53You should then look into back testing.
53:57You should then identify
54:01prop firms you want to rules properforms
54:03you want to use and rules you want to
54:06follow and you should adapt your
54:09strategy to those rules. You should then
54:12live test until you have the competence
54:14and then you should risk the money to
54:16get paid.
54:18That is the order that you should
54:20follow.
54:30I love this stuff. This stuff is
54:32awesome.
54:34It's a Saturday
54:36and I just came back from the gym. I was
54:38going to take a shower and I said, "You
54:40know what? No, I want to make this. I
54:41want to make this video. I want to
54:43provide value because I just love it. I
54:46love this stuff. I think about this all
54:49day long.
54:51I take calls all day long. It was Friday
54:55night and I was calling people. 8:00
54:59a.m. that same day I was calling people.
55:02I'm so invested in the success of my
55:05students because I love them. I I really
55:08want to see them win and it is the most
55:10rewarding feeling ever. One of my
55:12students passed an account the other day
55:14with fully automatic executions and I
55:17love that. But what I loved even more
55:18was he was working hard for it.
55:21And I love to see those who work hard
55:23and give it their all succeed. It
55:25[clears throat] fires me up. I love it
55:27so much. And that applies to every
55:29realm. The gym, trading, even school,
55:32maybe school. If you're working hard,
55:35you will reach what you want to reach.
55:39And I love working with the dogs in
55:41there in my community. I have one guy
55:43who a few days ago we were on a call and
55:47I literally told him to stop working.
55:50[laughter] I said, "Dude, literally
55:53close Claude. You've built everything
55:54you need. It's time to chill out and let
55:57the systems run." Because the point of
55:58AI automation is to let it be automated.
56:01If every time you build an automation,
56:02you look for the next thing to be
56:04automated, what are you doing it for?
56:07I don't watch the choice anymore. I'm
56:09enjoying the fruits of my labor. And
56:12that's the point of all this. Let's use
56:13the new technology to our advantage.
56:20And it requires work.
56:23If you think it doesn't, you're wrong.
56:28If you want to work one-on-one with me,
56:30I've got a mentorship. It's Ocean Front
56:31AI. I love working with my students. We
56:34talk all the time.
56:36And let me know what you think of the
56:37video. Let me know what questions you
56:39have, what things I can answer in the
56:40future. And uh I hope to see you soon.