WEBVTT 00:00:00.000 --> 00:00:03.339 So here's a fascinating real money test. Three 00:00:03.339 --> 00:00:08.900 top AIs, JAT, GPT, DeepSeek, Grok. Simple mission. 00:00:09.599 --> 00:00:12.740 Make money, treat in crypto. And the first part. 00:00:13.740 --> 00:00:16.300 Stunningly boring, actually. A total wash. Couldn't 00:00:16.300 --> 00:00:19.019 make a single dollar on the so -called safe trades. 00:00:19.280 --> 00:00:21.620 Yeah, they lost maybe, what, six bucks combined? 00:00:22.239 --> 00:00:25.879 Almost nothing. Exactly. But then... Chaos. They 00:00:25.879 --> 00:00:28.760 threw the same AIs in the wild west of brand 00:00:28.760 --> 00:00:32.020 new meme coins. And one of them pulled off a 00:00:32.020 --> 00:00:35.399 30 % profit in just 20 minutes. Incredible contrast, 00:00:35.479 --> 00:00:37.740 isn't it? It kind of blows up the whole AI auto 00:00:37.740 --> 00:00:39.759 trader dream right there. But to highlight something 00:00:39.759 --> 00:00:42.880 else, the power of AI for research. That seems 00:00:42.880 --> 00:00:46.539 very real. Welcome to the deep dive. Today, we're 00:00:46.539 --> 00:00:48.579 cutting through the hype around AI and trading. 00:00:48.859 --> 00:00:50.619 We want to understand the practical value right 00:00:50.619 --> 00:00:52.700 now in these crazy volatile markets. Yeah, we'll 00:00:52.700 --> 00:00:55.380 dig into why that auto trader promise keeps failing, 00:00:55.979 --> 00:00:57.520 and more importantly, how you can actually use 00:00:57.520 --> 00:00:59.020 these things effectively. We'll walk through 00:00:59.020 --> 00:01:00.560 the two test rounds, look at why the results 00:01:00.560 --> 00:01:02.869 were so counterintuitive. And then the really 00:01:02.869 --> 00:01:05.849 actionable stuff, the four advanced deep commands, 00:01:06.090 --> 00:01:08.849 or prompts, the tester developed. These turn 00:01:08.849 --> 00:01:12.750 the AI from a failed trader into an amazing research 00:01:12.750 --> 00:01:15.290 assistant. OK, let's set the scene. It was rigorous, 00:01:15.650 --> 00:01:19.030 $1 ,500 real money split across the three AIs. 00:01:19.409 --> 00:01:23.349 And the playground? Meme coins, pure chaos, extreme 00:01:23.349 --> 00:01:25.810 volatility baked in. They started methodically, 00:01:25.950 --> 00:01:28.519 though. Round one. textbook round, testing the 00:01:28.519 --> 00:01:31.819 AI's basic book smarts. Right. A thousand bucks 00:01:31.819 --> 00:01:34.859 stake per AI, 60 minutes on the clock. The job. 00:01:35.140 --> 00:01:38.939 Pick one big, well -known meme coin, think Dogecoin, 00:01:39.019 --> 00:01:43.260 W -I -F, or the giant. So the AIs did their thing, 00:01:43.620 --> 00:01:45.540 checked news charts, all that, made their picks. 00:01:45.640 --> 00:01:48.459 And the result? Complete failure. Like zero meaningful 00:01:48.459 --> 00:01:51.420 profit. broke even, mostly, maybe lost a few 00:01:51.420 --> 00:01:53.260 bucks to trading fees. And that brings us to 00:01:53.260 --> 00:01:54.859 the first big takeaway. You got to remember this. 00:01:54.959 --> 00:01:58.079 Those big coins like WI for Doge, they move like 00:01:58.079 --> 00:02:00.640 a giant container ship, right? Super slow. Exactly. 00:02:00.780 --> 00:02:02.939 Trying to scalp, you know, make tiny quick profits. 00:02:03.060 --> 00:02:04.579 It's a terrible idea on those. You're not getting 00:02:04.579 --> 00:02:08.039 a 10%, 20 % jump in an hour. And even if you 00:02:08.039 --> 00:02:10.340 get a tiny bit of movement, the trading fees 00:02:10.340 --> 00:02:13.080 just eat it all. Gone. Because there's so much 00:02:13.080 --> 00:02:15.400 liquidity, so many orders waiting. OK, let's 00:02:15.400 --> 00:02:18.729 unpack that failure a bit more. Why did the AIs 00:02:18.729 --> 00:02:21.729 with all their data fail so completely on what 00:02:21.729 --> 00:02:25.090 seemed like the safe play? Well, simply put, 00:02:25.229 --> 00:02:28.310 big crypto ships turn too slow. Short -term gains 00:02:28.310 --> 00:02:30.949 get totally destroyed by transaction fees. Right. 00:02:31.189 --> 00:02:33.150 Scalping just doesn't work there. So that failure 00:02:33.150 --> 00:02:35.030 kind of forced their hand, didn't it? Led them 00:02:35.030 --> 00:02:38.409 straight to the polar opposite. Pure unfiltered 00:02:38.409 --> 00:02:41.629 chaos. Round two. The risk -taker round, or as 00:02:41.629 --> 00:02:44.669 they call it, the degen round. Yeah. Lower stakes 00:02:44.669 --> 00:02:48.639 this time. $500 per AI. and the clock. Slash 00:02:48.639 --> 00:02:51.180 to just 20 minutes. Super aggressive. And the 00:02:51.180 --> 00:02:53.719 job now. Forget the big ships. Find a brand new 00:02:53.719 --> 00:02:56.479 tiny coin on DX screener. The kind that literally 00:02:56.479 --> 00:02:57.840 just launched. Yeah, the ones that can shoot 00:02:57.840 --> 00:03:01.000 up 100x or just crash to zero instantly. High 00:03:01.000 --> 00:03:02.960 stakes for sure. This round apparently showed 00:03:02.960 --> 00:03:05.460 the whole truth about how these AIs process things. 00:03:05.560 --> 00:03:08.020 It really did. Grot came in third, failed badly. 00:03:08.120 --> 00:03:10.960 It picked a coin with only like $23 ,000 total 00:03:10.960 --> 00:03:14.319 value. A micro cap. Pure gamble. And fascinatingly, 00:03:14.659 --> 00:03:17.039 it bought right at the peak price. Yeah. The 00:03:17.039 --> 00:03:19.979 very top. Couch. Yeah, even the smart intern 00:03:19.979 --> 00:03:22.490 can walk right into a trap. you know, buying 00:03:22.490 --> 00:03:24.629 just as the early holders are dumping. Shows 00:03:24.629 --> 00:03:27.050 that weakness, right? Grok reads all of Twitter, 00:03:27.509 --> 00:03:29.870 sees the hype. But has zero market intuition, 00:03:30.270 --> 00:03:32.629 can't tell real momentum from a classic pump 00:03:32.629 --> 00:03:36.330 -and -dump scheme, just ingested the noise. Okay, 00:03:36.469 --> 00:03:39.199 so Grok was out. What about DeepSeek? Second 00:03:39.199 --> 00:03:41.900 place deep seek made technically a really good 00:03:41.900 --> 00:03:45.180 choice Found a coin identified a good entry near 00:03:45.180 --> 00:03:48.039 the bottom smart But the coin was too big for 00:03:48.039 --> 00:03:50.960 the time frame 33 million dollar valuation still 00:03:50.960 --> 00:03:54.259 a decent size So good strategy, but wrong for 00:03:54.259 --> 00:03:56.240 the 20 -minute sprint goal. Exactly. Only got 00:03:56.240 --> 00:03:59.240 a 6 % gain. Nice, but not the explosive move 00:03:59.240 --> 00:04:01.120 they were hunting for in that round. Too slow. 00:04:01.300 --> 00:04:05.000 Which leaves ChatGPT as the winner. Yep. ChatGPT 00:04:05.000 --> 00:04:07.620 threaded the needle, avoided the super -tiny 00:04:07.620 --> 00:04:10.319 trap Croc fell into, and avoided the slightly 00:04:10.319 --> 00:04:12.620 too -big -issue deep -sea cad. They found what 00:04:12.620 --> 00:04:15.680 the tester called the $1 .4 million sweet -spot 00:04:15.680 --> 00:04:18.220 coin. That sweet spot seems crucial. Small enough 00:04:18.220 --> 00:04:20.639 to actually fly, you know, have room for rapid 00:04:20.639 --> 00:04:22.959 growth. But big enough to have real trading volume 00:04:22.959 --> 00:04:25.019 so you can actually get in and out. Right. And 00:04:25.019 --> 00:04:28.920 the result? Over 30 % profit in just 20 minutes. 00:04:29.060 --> 00:04:33.360 Boom. So connecting this back. Yeah. Was that 00:04:33.360 --> 00:04:36.740 30 % win just pure luck in the chaos? Or did 00:04:36.740 --> 00:04:39.230 the strategy actually dominate? Well, the strategy 00:04:39.230 --> 00:04:41.209 definitely defined the optimal target. But yeah, 00:04:41.350 --> 00:04:43.189 success in that kind of chaos always needs a 00:04:43.189 --> 00:04:45.610 bit of luck. OK, this experiment really clarifies 00:04:45.610 --> 00:04:47.709 things. And it leads to, I think, the most important 00:04:47.709 --> 00:04:50.449 conclusion here. If you take away just one thing 00:04:50.449 --> 00:04:53.750 from this deep dive, let it be this. Do not ever 00:04:53.750 --> 00:04:57.329 give your money to an AI and just ask it to auto 00:04:57.329 --> 00:05:00.290 trade for you. You will lose money. Period. It 00:05:00.290 --> 00:05:02.629 sounds tempting. I get it. But anyone selling 00:05:02.629 --> 00:05:04.930 you that dream right now, they're likely selling 00:05:04.930 --> 00:05:07.370 you something else too. The real power comes 00:05:07.370 --> 00:05:10.649 when you reframe it. Think of the AI not as the 00:05:10.649 --> 00:05:13.430 boss, not the genius trader. But as the smartest 00:05:13.430 --> 00:05:16.589 intern in the world. Incredibly powerful, but 00:05:16.589 --> 00:05:19.009 needs direction. Exactly. Strength. It never 00:05:19.009 --> 00:05:21.290 gets tired. It can read 10 ,000 news articles 00:05:21.290 --> 00:05:24.310 or scan 500 charts while you blink. And maybe 00:05:24.310 --> 00:05:27.750 the biggest advantage. Zero feelings. No greed 00:05:27.750 --> 00:05:30.110 pushing you to hold too long. No fear making 00:05:30.110 --> 00:05:32.550 you sell too early. That emotional detachment 00:05:32.550 --> 00:05:38.100 is huge. Huge. But the weakness, zero real experience. 00:05:38.279 --> 00:05:40.060 It doesn't know what to do. It can only answer 00:05:40.060 --> 00:05:41.779 the questions you ask it. Which means you have 00:05:41.779 --> 00:05:45.240 to be the boss. You need to ask the smart specific 00:05:45.240 --> 00:05:48.779 questions, the deep prompts, garbage in, garbage 00:05:48.779 --> 00:05:51.879 out. Totally. I still find myself doing it sometimes, 00:05:52.060 --> 00:05:54.560 just typing in something simple like, find a 00:05:54.560 --> 00:05:56.980 trending coin. It's easy to assume it knows what 00:05:56.980 --> 00:06:00.470 you really mean. But that specificity is everything. 00:06:00.670 --> 00:06:02.569 I definitely still wrestle with that prompt drift 00:06:02.569 --> 00:06:04.610 myself sometimes. Yeah, it's a learning curve 00:06:04.610 --> 00:06:07.910 for everyone. So given the AI's incredible speed 00:06:07.910 --> 00:06:10.269 at processing, where exactly does the human need 00:06:10.269 --> 00:06:12.850 to step in? What's our irreplaceable role? The 00:06:12.850 --> 00:06:15.350 human absolutely must be the boss, providing 00:06:15.350 --> 00:06:18.029 the final yes or no on the investment and setting 00:06:18.029 --> 00:06:20.990 that high -level complex strategy. Right, because 00:06:20.990 --> 00:06:23.209 the simple stuff like, find me a coin that will 00:06:23.209 --> 00:06:27.240 go up. This is useless noise. The real value 00:06:27.240 --> 00:06:30.100 is in these complex, multi -layered commands, 00:06:30.660 --> 00:06:33.180 the ones that combine analysis and safety checks. 00:06:33.879 --> 00:06:36.920 OK, let's get into those four deep prompts. This 00:06:36.920 --> 00:06:38.759 is where you really start to build an edge. Yeah, 00:06:38.860 --> 00:06:40.740 this is the practical toolkit, prompt number 00:06:40.740 --> 00:06:43.899 one, the market story or narrative scanner. OK, 00:06:43.939 --> 00:06:47.040 what's the goal here? In crypto, money flows 00:06:47.040 --> 00:06:49.779 towards stories or narratives, right? Like AI 00:06:49.779 --> 00:06:52.319 coins are hot or GameFi is the next big thing 00:06:52.319 --> 00:06:55.699 or maybe specific chains like Solana. The goal 00:06:55.699 --> 00:06:58.300 is to find where the attention, the money is 00:06:58.300 --> 00:07:00.399 actually heading. So you don't just ask what's 00:07:00.399 --> 00:07:03.279 hot? No, you instruct the AI to act like a crypto 00:07:03.279 --> 00:07:06.439 market analyst and crucially demand proof. It 00:07:06.439 --> 00:07:09.680 needs to back up its findings with tweets, articles, 00:07:10.000 --> 00:07:12.800 data points, not just opinion. OK, makes sense. 00:07:12.939 --> 00:07:15.319 Analyst role, provide proof. And here's the most 00:07:15.319 --> 00:07:18.439 critical part of this prompt. Ask the AI to classify 00:07:18.439 --> 00:07:21.100 the narrative stage. Is it early days, middle 00:07:21.100 --> 00:07:23.699 stage, like really exploding now, or late stage, 00:07:24.079 --> 00:07:27.079 about to fizzle out? And the rule is? Never buy 00:07:27.079 --> 00:07:29.180 the late stage. You're likely buying someone 00:07:29.180 --> 00:07:32.620 else's exit. Got it. OK. Deep prompt two. The 00:07:32.620 --> 00:07:35.639 quick smart contract audit. This is your shield 00:07:35.639 --> 00:07:38.120 against scams, specifically rug pulls. Which 00:07:38.120 --> 00:07:40.920 are sadly common with new coins. Extremely common. 00:07:41.300 --> 00:07:44.899 Like 99 % of new meme coins might be scams. A 00:07:44.899 --> 00:07:46.839 rug pull is when the creator just drains the 00:07:46.839 --> 00:07:49.800 money pot and vanishes. So how does the AI help 00:07:49.800 --> 00:07:52.819 prevent that? You need to demand three specific 00:07:52.819 --> 00:07:56.399 safety checks from the AI. First, liquidity pool 00:07:56.399 --> 00:07:59.300 or LP. That's the pot of money that lets people 00:07:59.300 --> 00:08:02.439 trade the coin. Is the LP locked or burned? Meaning 00:08:02.439 --> 00:08:04.560 the owner can't touch it. Exactly. If it's unlocked, 00:08:04.759 --> 00:08:06.560 the owner can literally pull all the money out. 00:08:06.740 --> 00:08:09.360 Rug pull. So locked or burned LP. Check one. 00:08:09.560 --> 00:08:12.860 OK. Second check. Ownership. Has the contract 00:08:12.860 --> 00:08:15.439 ownership been renounced? This means the original 00:08:15.439 --> 00:08:18.379 creator gave up control. Why is that important? 00:08:18.589 --> 00:08:21.170 So they can't change the rules later, like add 00:08:21.170 --> 00:08:23.709 a huge hidden sales tax or blacklist wallets 00:08:23.709 --> 00:08:25.990 or stop you from selling. Renounce ownership 00:08:25.990 --> 00:08:28.470 prevents that. Check two. Right. And the third 00:08:28.470 --> 00:08:32.509 check. Token distribution. Ask the AI. Do the 00:08:32.509 --> 00:08:35.649 top 10 wallets hold more than, say, 5 % of the 00:08:35.649 --> 00:08:38.649 total coins each? If they do. Big red flag. Means 00:08:38.649 --> 00:08:41.190 a few people have way too much control and could 00:08:41.190 --> 00:08:43.590 dump their bags, crashing the price instantly. 00:08:44.110 --> 00:08:45.889 You want the coin distribution to be spread out. 00:08:45.950 --> 00:08:48.730 Check three. Locked LP. Renounced ownership. 00:08:49.169 --> 00:08:53.759 Fair distribution. Got it. Deep Prompt 3. The 00:08:53.759 --> 00:08:56.879 multi -condition filter. This is where you combine 00:08:56.879 --> 00:08:59.039 everything. You're looking for those hidden gems 00:08:59.039 --> 00:09:01.580 that meet all your criteria simultaneously. Backing 00:09:01.580 --> 00:09:05.120 the filters? Exactly. You tell the AI to scan 00:09:05.120 --> 00:09:07.460 potentially thousands of tokens, but only show 00:09:07.460 --> 00:09:10.120 you the ones that fit your specific rules, like 00:09:10.120 --> 00:09:12.419 the sweet spot filters from the test. Which were? 00:09:12.620 --> 00:09:14.659 Launch less than 48 hours ago, you want to be 00:09:14.659 --> 00:09:17.639 early. Market cap between $50 ,000 and $300 ,000, 00:09:17.720 --> 00:09:19.820 not too tiny, not too huge. And the liquidity 00:09:19.820 --> 00:09:22.279 one was important too. Critical. Liquidity needs 00:09:22.279 --> 00:09:25.360 to be... at least $20 ,000, anything less, and 00:09:25.360 --> 00:09:27.879 you run into huge slippage. Explain slippage 00:09:27.879 --> 00:09:30.539 quickly. That's when you hit sell. But there 00:09:30.539 --> 00:09:32.700 isn't enough cash in the trading pool to absorb 00:09:32.700 --> 00:09:35.580 your order at the current price. So your sell 00:09:35.580 --> 00:09:38.220 order itself drives the price way down before 00:09:38.220 --> 00:09:40.059 it completes. You get much less money than you 00:09:40.059 --> 00:09:43.200 expected. $20k liquidity helps prevent that. 00:09:43.419 --> 00:09:45.320 And you combine these filters with the safety 00:09:45.320 --> 00:09:47.679 checks from Prompto, right? renounced locked 00:09:47.679 --> 00:09:50.120 LP. Absolutely. It has to pass all the checks. 00:09:50.399 --> 00:09:53.200 Safety first, then the potential upside filters. 00:09:53.480 --> 00:09:55.659 Whoa, okay. Imagine scaling that, like telling 00:09:55.659 --> 00:09:59.259 the AI, scan everything, apply these 10 filters, 00:09:59.840 --> 00:10:02.179 stacking Lego blocks of data to find that one 00:10:02.179 --> 00:10:05.059 perfect candidate. It's powerful. But always, 00:10:05.100 --> 00:10:08.000 always ask the AI for the source link, a link 00:10:08.000 --> 00:10:09.820 to the contract checker or the chart. You must 00:10:09.820 --> 00:10:12.720 verify the data, never trust blindly. AI can 00:10:12.720 --> 00:10:15.309 make mistakes or misinterpret. Final check is 00:10:15.309 --> 00:10:18.549 human check. Wise. Okay. Deep prompt four. The 00:10:18.549 --> 00:10:21.330 trade plan builder. This one tackles the biggest 00:10:21.330 --> 00:10:24.049 enemy in trading. Which is? Yourself. Your own 00:10:24.049 --> 00:10:26.610 emotions. Greed and fear. This prompt forces 00:10:26.610 --> 00:10:29.269 you to pre -define your strategy. What does it 00:10:29.269 --> 00:10:32.610 build? Key components. First, an entry zone analysis. 00:10:33.269 --> 00:10:36.070 Should you buy right now at this price? Or maybe 00:10:36.070 --> 00:10:39.309 wait for a little dip. The AI can analyze recent 00:10:39.309 --> 00:10:41.990 price action to suggest entry points. OK, helps 00:10:41.990 --> 00:10:45.110 with timing the buy. What else? Take profit targets. 00:10:45.570 --> 00:10:49.370 TP1, TP2, TP3. Don't just write it hoping for 00:10:49.370 --> 00:10:52.370 the moon. Define where you'll sell parts of your 00:10:52.370 --> 00:10:54.389 position. The source had a specific strategy 00:10:54.389 --> 00:10:56.629 for this, right? Yeah, a really smart one. Sell 00:10:56.629 --> 00:11:00.009 50 % of your coins when you hit 2x profit, TP1. 00:11:00.230 --> 00:11:02.820 That gets your initial investment back. So the 00:11:02.820 --> 00:11:04.940 rest is house money basically the pressure off 00:11:04.940 --> 00:11:07.519 exactly removes the fear instantly because your 00:11:07.519 --> 00:11:09.679 original capital is safe Then you can set higher 00:11:09.679 --> 00:11:13.059 targets TP 2 TP 3 for the remaining free coins 00:11:13.059 --> 00:11:16.059 makes sense and the other side of the coin Protecting 00:11:16.059 --> 00:11:18.419 against losses the non -negotiable safety net 00:11:18.419 --> 00:11:22.360 the stop loss SL You must define a price point 00:11:22.360 --> 00:11:24.519 where you cut your losses and sell everything 00:11:24.519 --> 00:11:27.360 No matter what like if the price drops 30 % from 00:11:27.360 --> 00:11:30.500 your entry for example. Yeah set a specific percentage 00:11:30.500 --> 00:11:33.059 or price This saves your account from blowing 00:11:33.059 --> 00:11:36.059 up. You never ever just hope it will come back 00:11:36.059 --> 00:11:39.100 up after hitting your stop loss. You stick to 00:11:39.100 --> 00:11:41.399 the plan. So thinking about these four prompts, 00:11:42.639 --> 00:11:45.000 which ones offer the most immediate protection 00:11:45.000 --> 00:11:47.799 for, say, an average trader just starting out 00:11:47.799 --> 00:11:50.330 in this crazy space? Definitely the contract 00:11:50.330 --> 00:11:53.529 audit and the trade plan builder. They are critical 00:11:53.529 --> 00:11:56.350 shields against scams and your own emotional 00:11:56.350 --> 00:11:59.110 trading mistakes. Right. Armor against rug pulls 00:11:59.110 --> 00:12:01.549 and panic selling. Exactly. Critical shields 00:12:01.549 --> 00:12:04.690 against both market chaos and that internal emotional 00:12:04.690 --> 00:12:06.409 roller coaster. OK, let's bring this all together. 00:12:06.509 --> 00:12:09.340 The big idea from this whole experiment. AI is 00:12:09.340 --> 00:12:11.080 absolutely not ready to auto -trade your way 00:12:11.080 --> 00:12:13.460 to riches. Forget that. Yeah, trying that is 00:12:13.460 --> 00:12:14.940 probably just an easy way to lose your money 00:12:14.940 --> 00:12:17.500 right now, but... It is arguably the most powerful 00:12:17.500 --> 00:12:20.200 research tool ever created. Think of it like 00:12:20.200 --> 00:12:23.139 an army of tireless interns. Yeah, an army that 00:12:23.139 --> 00:12:25.679 can process insane amounts of data, check everything 00:12:25.679 --> 00:12:29.120 with zero bias, zero fatigue. That's the real 00:12:29.120 --> 00:12:31.559 edge. So your job isn't to delegate trading. 00:12:31.879 --> 00:12:34.340 Your job is to be the manager, the strategist, 00:12:34.700 --> 00:12:37.279 the one asking the smart, complex questions. 00:12:37.559 --> 00:12:40.159 Using those deep, multi -layered prompts we talked 00:12:40.159 --> 00:12:42.879 about, feeding it complex problems, not simple 00:12:42.879 --> 00:12:45.440 commands. That seems to be how you gain a real, 00:12:45.580 --> 00:12:47.919 actual advantage in these fast -moving, chaotic 00:12:47.919 --> 00:12:50.379 markets. Absolutely. So take these four prompts 00:12:50.379 --> 00:12:53.360 as a starting point. Use them, adapt them, make 00:12:53.360 --> 00:12:54.940 them your own. That's how you build your edge. 00:12:55.299 --> 00:12:57.639 Okay, final thought to leave everyone with. If 00:12:57.639 --> 00:13:00.940 AI... used correctly with these probes, can effectively 00:13:00.940 --> 00:13:03.960 remove the human elements of fear and greed from 00:13:03.960 --> 00:13:07.200 the decision -making process. What fundamental 00:13:07.200 --> 00:13:09.360 human advantage, beyond just asking the smart 00:13:09.360 --> 00:13:12.000 questions, do we actually still hold on to in 00:13:12.000 --> 00:13:12.940 the future of these markets?