AI Trade Bot won 815% GAIN while everyone else LOST — Transcript

Full transcript

1. 0:00AI just helped me create an unstoppable
2. 0:02strategy for trading. Look at this blue
3. 0:05line. This is where this bot failed. I'm
4. 0:06going to go deep and show you
5. 0:08everything. This blue line is where it
6. 0:09filled here. Cash cat, it's up um 290%
7. 0:13300%. Uh this is the analysis with
8. 0:16leverage here. 272% at 1x leverage. U 2x
9. 0:20leverage 543%. While if you did it the
10. 0:23other way where you're just like trying
11. 0:24to buy a dip, you would have been
12. 0:25liquidated on 2x leverage. You would
13. 0:27have been liquidated on 3x. uh you would
14. 0:29have been liquidated if you bought a 25%
15. 0:32dip. So this is anchored to
16. 0:34liquidations. This is when other people
17. 0:36have to give up their positions. I can
18. 0:38see everybody's position in the market
19. 0:40at all times. So this gives me an edge.
20. 0:42I'll explain here in this short video.
21. 0:44You can see returns up to 815%. It's
22. 0:48crazy. Go ahead and lock in. I'll show
23. 0:50you absolutely everything. My name is
24. 0:51Mundav. I've helped thousands of people
25. 0:53go from emotional hand trader that is
26. 0:56non-technical. I was non-technical as
27. 0:58well. Now I'm a fully automated trader
28. 1:00and I'll help you do the same here in
29. 1:02this short bit. So what I want you to
30. 1:03look into is okay, we had a scenario
31. 1:05here where essentially I have this bot.
32. 1:08Um, let me go ahead and show you that
33. 1:10bot. Actually, let me go ahead and show
34. 1:12you that bot. It is um,
35. 1:16okay, it's this bot that essentially is
36. 1:18sitting there and buying people's
37. 1:20positions when they get liquidated. It's
38. 1:22predicting when they'll get liquidated
39. 1:24and then buying at that placement.
40. 1:28Now, my question here for you is this.
41. 1:39Got a new key. I got I got a new key. I
42. 1:41got a new key. Was I even saying
43. 1:43anything there?
44. 1:45What's my whisper looking like?
45. 1:48Open Whisper Flow. Okay, perfect. Okay,
46. 1:52I do have my Whisper Flow. We got a new
47. 1:54API key that I'm dropping in the chat
48. 1:56here. in a second. Um, new key here. Who
49. 1:59this? Okay, let's get a unlock app.
50. 2:02Boom.
51. 2:08Let's go ahead here and let's actually
52. 2:10just get to the actual place. It is
53. 2:17stat or 26, I believe. Yes.
54. 2:22Okay, let's see if I got any fable left.
55. 2:24I think I'm out of fable for the week,
56. 2:25but okay if I am, okay, if I'm not okay,
57. 2:28I'mma keep going. Always. Yes. Thank
58. 2:30you.
59. 2:32So, let's go ahead and pull up the code
60. 2:34here. Open in cursor. I already gave you
61. 2:37all this code yesterday. Bought cash
62. 2:39lick hunt. It's on my YouTube right now
63. 2:41if you want to go look at it. It smoked
64. 2:43them.
65. 2:45It smoked them.
66. 2:48All right. So in this unique case, it
67. 2:50was sitting at the buy of 0.04
68. 2:54something, which ended up being the
69. 2:56bottom tick of the liquidation. It's
70. 2:58looking at people's positions and seeing
71. 3:00when they get liquidated and putting the
72. 3:01orders there. So when they do get
73. 3:03liquidated, what happened was there was
74. 3:04a flash crash and it filled perfectly.
75. 3:07Okay.
76. 3:09And then it ripped to like 20 cents. So
77. 3:12this was actually a 200 230% move. And
78. 3:16using 2x leverage, you can do the math.
79. 3:183x leverage. You can do the math.
80. 3:19Anyways, that's a huge, huge move. I
81. 3:22want you to run a study here.
82. 3:26I'm trying to figure out something that
83. 3:29new people can do instead of just
84. 3:32getting directional. Okay, I'm on year
85. 3:35six of algorithmic trading. Obviously,
86. 3:37my systems are a little more intricate
87. 3:39than most, but most people come to me
88. 3:41and say, "Hey, can AI just trade for
89. 3:43me?" And it's like, "No, AI can't just
90. 3:44decide for you."
91. 3:48But if you do have a bias in the market,
92. 3:50there's much better ways to get short or
93. 3:52long than just doing that. So I want you
94. 3:55to run a deep study here with three
95. 3:57different agents
96. 4:00answering the question of is it better
97. 4:04for that hand trader to just get biased
98. 4:06long or short or is it better to sit
99. 4:10there and wait for people to get
100. 4:11liquidated? the case that I showed you,
101. 4:13you can see moondev.com/doccks
102. 4:15to to run the current liquidations on
103. 4:17cash or other symbols.
104. 4:21My intuition says yes, it's better to
105. 4:25buy liquidations for smaller tokens, but
106. 4:27it may not be better for big tokens. But
107. 4:29I want you to run this study without
108. 4:31that bias. I want you to go ahead and
109. 4:34answer that question. Maybe do some
110. 4:36Monte Carlos, whatever you need to do to
111. 4:39answer that question. Go ahead and run
112. 4:40it at three agents. Let's cut up. Okay,
113. 4:43so you heard what I'm getting started
114. 4:45with. Let's see. Let's see, dude, cuz I
115. 4:47dropped it for free.
116. 4:50777 Nick, how are you, dude? It's good
117. 4:53to have you here. Welcome, welcome,
118. 4:56welcome.
119. 4:59Welcome, welcome, welcome, welcome,
120. 5:00welcome in. Welcome in.
121. 5:03I got on a bit early today. My bad.
122. 5:11It's hard for me to stick to a time when
123. 5:13I'm ready to go. I wake up and I'm ready
124. 5:15to go.
125. 5:18All right. So, she's cooking on that
126. 5:20here. Um, I'll dig into the bot first,
127. 5:22then set up three agent study. [snorts]
128. 5:25Let me read the bot and the API helper
129. 5:26so the agents get the real sources.
130. 5:28Okay. Thank you. That's exactly what I
131. 5:30want you to do. I think this is going to
132. 5:31be absolutely fire. Okay. In the
133. 5:33meantime, let's go ahead and check in on
134. 5:35a couple things here. We go to
135. 5:37mundave.comai.
136. 5:39We have that AI trading battle still
137. 5:41going. And I just want to take a quick
138. 5:44little update here to see who's winning.
139. 5:46So, you can see here that Anthropic Opus
140. 5:495 is in the lead. Let's go.
141. 5:53Let's go. Opus 5, baby. Somebody asked
142. 5:56me the other day who I'm rooting for,
143. 5:57and I think I'm rooting for Anthropic
144. 5:59cuz Open AI was in the lead before.
145. 6:02Damn. They're barely in the lead,
146. 6:03though. You can see 5.98% for Anthropic
147. 6:06Opus 5 plus 5.95% for OpenAI
148. 6:12Moonshot. Kimmy 3. Okay. Okay. Coming
149. 6:14back from the trenches.
150. 6:17Kimmy 3 is up 3.96%. Google is up 1.73%.
151. 6:22XAI is up.34%.
152. 6:24Okay. Deepseek is down near 10%. How do
153. 6:28you do that? I don't know. How do you do
154. 6:31that? I don't know.
155. 6:33I don't know, dude. Okay, so that is the
156. 6:35update to the AI. Next decision is in 1
157. 6:38hour. You can read the rules here.
158. 6:40Essentially, it's a bunch of AIs
159. 6:42competing against each other because
160. 6:43everybody wants to know the one
161. 6:44question. Mundave, can AI trade for me?
162. 6:47Well, I mean, technically they're up.
163. 6:53Technically, it's a little choppy, but
164. 6:55they're up.
165. 7:02Does it beat buy and hold is the
166. 7:04question.
167. 7:06I should have done that. I should have
168. 7:07had a buy and hold line in here. That'd
169. 7:09be fire.
170. 7:14We'll see. We'll see. Anyways, you can
171. 7:16read about the rules here. Six of the
172. 7:18world's top AI models trade real money
173. 7:20live on Hyperlid. Each starts with $100.
174. 7:22Same rules, same data, same schedule for
175. 7:24all.
176. 7:26We want one answer honestly forever.
177. 7:28which AI trades the best. We do not
178. 7:30think AI can just trade for you. This is
179. 7:32the running public proof and the
180. 7:34benchmark. Now, there's a ton of code
181. 7:36here. People always ask me for code for
182. 7:39trading bot code. Go ahead and snipe
183. 7:41some of this. This literally runs the
184. 7:44AI. So,
185. 7:47pretty fire if you ask me. All right.
186. 7:49So, she's cooking on something down
187. 7:51here. You can see that she's running the
188. 7:52study for me because this is a very
189. 7:54interesting study. All three agents are
190. 7:55now running Moath. Here's a setup while
191. 7:58they work. What your bot is actually
192. 8:00doing. Plain answer first. Pulls every
193. 8:03open long on Cash Cat, sorts them by
194. 8:05liquidation price,
195. 8:08and finds the exact price where 83% of
196. 8:10the reachable long book, gets forced
197. 8:12liquidated. It rests one buy order.5%
198. 8:16under that cluster, and rechecks every 5
199. 8:18minutes. When the cascade hits,
200. 8:21liquidated longs become forced market
201. 8:23sells and your bid is the liquidity that
202. 8:26catches them. That's why it nailed the
203. 8:28bottom ticket 04.
204. 8:31It wasn't predicting price. It was
205. 8:33standing where the force selling had to
206. 8:35land. Exit is mechanical as well.
207. 8:39Midpoint of the highest wick and highest
208. 8:41close over 14 days. The study now
209. 8:43running. Three agents. No bias baked in.
210. 8:46small caps agent cash plus 10 low open
211. 8:51interest hyperlquid per okay majors and
212. 8:55then stats agent. Okay, so we should see
213. 8:57this in a few minutes. I don't know how
214. 8:59long it's going to be, but we will be
215. 9:00here when it's done. I can tell you that
216. 9:02much. I want to go ahead and see what
217. 9:04shorty said about this yesterday. Done
218. 9:07and tested yesterday. Here's the full
219. 9:09picture. Your answer has been Your
220. 9:10question has been answered. Yes, the bot
221. 9:12was rebasing every 5 minutes. It
222. 9:14computed 83% of ritual book
223. 9:20against whoever survived. Wipe the top
224. 9:2240% and the new 83%
225. 9:25[snorts] at what used to be the 89.8%
226. 9:30level, a deeper price. So mid cascade,
227. 9:33the anchor retreats. The bot cancels
228. 9:35your bid and replaces it lower, running
229. 9:38away from the exact force selling it was
230. 9:40built to catch. Damn, the 1.5% dampening
231. 9:43or damping only filters noise. It can't
232. 9:47stop a cascade size shift. This isn't
233. 9:51the hysterious what is this word thing
234. 9:54from the multi- symbol talk
235. 9:57that was about which symbols get orders.
236. 10:01This is the denominator of the trigger
237. 10:03shrinking under you. The fix using your
238. 10:05hourly window idea. The two knobs in the
239. 10:09config block. Cascade freeze drop
240. 10:11percentage is 20. Cascade window freeze.
241. 10:14Your 1 to two hour parameter set to 60
242. 10:16if you want it to be twitchier. Each
243. 10:19loop the bot records the reachable book
244. 10:22size per symbol. It keeps a rolling
245. 10:25book. Holds the ladder. The freeze.
246. 10:27Okay. Verified with a unit. One boundary
247. 10:29worth knowing if the cascade wipes less
248. 10:31than 20% of the book. No freeze.
249. 10:35Small
250. 10:39nibbles at the top of the ladder are
251. 10:41normal.
252. 10:44All right. So, I want to go ahead and
253. 10:45implement this into both of them. I want
254. 10:47to do the I want to do a 4hour freeze
255. 10:49for now. So, essentially that would mean
256. 10:51that it's only recalculating the 83% or
257. 10:54whatever every 4 hours.
258. 10:57This will give it plenty of time to
259. 10:59cascade into uh into it. I don't like
260. 11:03that thing you did about the nibbling,
261. 11:06but go ahead and uh the 20%. I don't
262. 11:08understand that
263. 11:11cascade wipes less and no freeze anchor
264. 11:13will rebate. I don't understand that.
265. 11:15What's the point of having that? So,
266. 11:16let's just let's remove it. Okay.
267. 11:21All right. So, I got her working on
268. 11:23that. I got her working on that. Let's
269. 11:24get it, dude. Let's get it. Okay. We got
270. 11:27this going. This is a deep research
271. 11:28project I need to do here today and I
272. 11:30need to answer that today. So, I'm going
273. 11:32to answer that with you because I
274. 11:33believe code is a great equalizer. Damn,
275. 11:35it's AI is a great equalizer now, huh?
276. 11:37Cuz everybody can code now. We have AI.
277. 11:42So, let's go ahead and keep cooking
278. 11:43every single day. I don't care what this
279. 11:45fear and greed is. I'm here every single
280. 11:47day. We were here when the fear and
281. 11:48greed was eight. We're going to be here
282. 11:50when the fear and greed is 99. I don't
283. 11:52care. Let's just get it. Every single
284. 11:55day.
285. 11:57Every single day, dude.
286. 12:07every single day until you're an
287. 12:08overnight success. That's how the world
288. 12:10works. You're not going to be able to
289. 12:12tell the difference day by day. But if
290. 12:14you lock in on the same thing every
291. 12:16single day in one year from now, your
292. 12:19life will never be the same.
293. 12:23But if you keep changing up, you keep
294. 12:25going back to bad habits,
295. 12:28nothing's going to change.
296. 12:30So that's why I'm here. That's why I
297. 12:32have the fear and greed every day, no
298. 12:34matter what. I'm here with the hate fear
299. 12:36and greed. When everybody's scared, I'm
300. 12:39here with the 99 fear and greed where
301. 12:41everybody's excited. I don't care.
302. 12:48You've reached your fable five limit.
303. 12:55Run usage credits to continue or switch
304. 12:58model.
305. 13:01I ran out of Fable 5. I don't need her
306. 13:03anyways. I can do it with Opus one.
307. 13:11I could do it with Opus one. I don't
308. 13:12need that.
309. 13:15So, what's she going to do over here is
310. 13:16my question.
311. 13:19She's still cooking on it. Okay.
312. 13:24I got that. I got the agents everywhere.
313. 13:26Okay. Fiddle fing. She says filling
314. 13:29falling trying to get it. Philly fif
315. 13:34trying to get it. Step on the gas, fam.
316. 13:37Let's get it.
317. 13:42How to start this [ __ ] bro.
318. 13:46Just get started, baby boy. Get started.
319. 13:48Get started. Lock in here. I'll show you
320. 13:50ways. I'll show you the ways to get
321. 13:53started.
322. 13:54Okay. So, while she's fiddly fing, let's
323. 13:56go see if there's anything that I need
324. 13:58to see here. Let's go ahead and open up
325. 14:00moonab.com/news.
326. 14:03I didn't do this yesterday either.
327. 14:08So, I got the two AIs working for me in
328. 14:10the bottom. Let's go ahead and move one
329. 14:12of these over because
330. 14:15I need to see both of them at the same
331. 14:17time.
332. 14:20All right, Duba is done. It looks like
333. 14:27she done as well. No, not yet.
334. 14:31All right, let's go through this really
335. 14:32quickly. Uh Justin Michael Sailor
336. 14:34Strategy is now has a $4 billion
337. 14:36unrealized profit on his Bitcoin
338. 14:37investment. Good job, gang.
339. 14:40Breaking. American wages have fallen 43%
340. 14:42of national income and the lowest since
341. 14:44the Great Depression. For every $1 the
342. 14:45US economy produces, workers are getting
343. 14:47less of it than any point since 1929.
344. 14:52No, dude. The rest is going to corporate
345. 14:54profits, which are now at an all-time
346. 14:56high. This is why the economy keeps
347. 14:58growing, but Americans feel poorer. GDP
348. 15:00goes up, wages go up in dollar terms,
349. 15:02but workers are getting a smaller and
350. 15:04smaller slice of what they actually
351. 15:06produce. Asset prices keep rising for
352. 15:08the same reason.
353. 15:10Money that used to go to wedges is now
354. 15:12going to shareholders instead. And
355. 15:13shareholders put it into stocks, real
356. 15:15estate, and other assets. Damn. So,
357. 15:17America's getting absolutely cooked
358. 15:18right now. Unless you own assets, you're
359. 15:21absolutely smoke. [ __ ] I'm sorry, B.
360. 15:27Apple's liquid glass icons have have to
361. 15:30be designed by people so young who have
362. 15:33never used the objects they're modeled
363. 15:35after. I don't know what this is. Don't
364. 15:36care. The Bitcoin reformation. No
365. 15:39thanks. Hit 2 M on pump. 8 a.m. More to
366. 15:42go. Okay, so that's fire.
367. 15:47Warning. Something really bad happened.
368. 15:49Bitcoin has failed to break 80K and the
369. 15:52last bounce before the cope coke coke
370. 15:54coke coke coke. The plan now is simple.
371. 15:5680k to 65k to 57k to 42k. That's coke
372. 16:01coke. Who knows where prices are going
373. 16:03to go, but you don't know and I don't
374. 16:04know either. So, and watch out because
375. 16:06I'm not talking about just any drop. If
376. 16:09the scenario confirms, we'll be facing a
377. 16:12cleanup phase before the next big
378. 16:14market. Okay, cool. Um, Bitcoin's
379. 16:17four-year cycle just broke. So, the
380. 16:19bulls 1,00 days, bare 365 days, 10,00
381. 16:25bull 365 days. Or maybe the bare
382. 16:28market's not over. I don't know. I'm not
383. 16:29here to say. I'm just here to be here no
384. 16:31matter what. 74 greed is a little
385. 16:33suspicious. Sure. Sure, we were just at
386. 16:36eight.
387. 16:38Sure, but
388. 16:40you don't know which way price is going
389. 16:42to go. You never know. I'm now
390. 16:44officially on Pump Fun app and I'm
391. 16:45streaming trading on it. Let's see if I
392. 16:48can make money. Follow me at
393. 16:52Is this for real though?
394. 16:55He went to pump fun.
395. 17:00Martinrell about to go back to Hely,
396. 17:03bro. pump.f fun
397. 17:13about 24 hours ago. This actually him.
398. 17:17Nah, this ain't him.
399. 17:23Is this him for real?
400. 17:29It's not. It can't be him.
401. 17:31This is him.
402. 17:34So, he's up 841%
403. 17:37here.
404. 17:40Top trades here.
405. 17:48Holy hell. Holy hell. Martin Shreley.
406. 17:52He got to pump fun now.
407. 17:55This boy trying to go back to jail or
408. 17:57something. I don't know.
409. 18:00Scary stuff. B. You can see we're still
410. 18:02cooking both sides here on the AI. So,
411. 18:05I'm waiting for them while and while
412. 18:08doing that, I'm going to go ahead and
413. 18:09see if there's anything that I'm missing
414. 18:12cuz this gives me tons of ideas.
415. 18:15Nvidia AI factory compute is becoming an
416. 18:17investable asset class. What do you
417. 18:19mean? Nvidia, that is that is a sign of
418. 18:22a bubble. That is a sign of a bubble.
419. 18:24But if you're not in a bubble, you
420. 18:26better find one. Nvidia AI factory
421. 18:28compute is becoming an investable asset
422. 18:29class. Today we announced partnerships
423. 18:31with Apollo, Black Rockck, Blackstone,
424. 18:33Brookfield, Goldman Sachs, and KKR
425. 18:36to establish independent financing
426. 18:38platforms designed to mobilize over 500
427. 18:40billion of third party capital to
428. 18:41support the buildout of the AI
429. 18:42infrastructure over time. This is a
430. 18:44major milestone for Nvidia and the AI
431. 18:46industry. We moved from an era in which
432. 18:48companies bought chips and built data
433. 18:50centers projects by built data centers
434. 18:53project by project to one in which AI
435. 18:56factories can be financed as productive
436. 18:58infrastructure.
437. 19:00With repeatable platforms, long-term
438. 19:03institutional capital, and diverse
439. 19:04customer base that uses compute to
440. 19:06create revenue, AI has reached an
441. 19:09inflection point. It is moving from
442. 19:11research into production. AI is creating
443. 19:14real value and the infrastructure behind
444. 19:16it is becoming one of the world's most
445. 19:18productive assets. In AI, compute is
446. 19:20revenue, a new infrastructure asset.
447. 19:22Nvidia compute is not just a chip. It is
448. 19:25a complete AI factory platform including
449. 19:29accelerated computing networking
450. 19:31systems, software, AI frameworks, and a
451. 19:33global developer ecosystem. Nvidia DSX
452. 19:37AI factories can run the world's
453. 19:38broadest range of AI modalities and
454. 19:41algorithms.
455. 19:46All right, bro. That's that's awesome.
456. 19:47Good to hear. But you know, you're
457. 19:49writing too much. So, I'm just going to
458. 19:51go ahead and um I'mma ask Grock.
459. 19:55TLDDR, please. Grock, thank you.
460. 19:58TLDDR, please. Grock, thank you.
461. 20:03We announced partnership with Apollo.
462. 20:04blah blah AI factories can now be
463. 20:06financed as productive infrastructure
464. 20:07with repeatable platforms long-term
465. 20:09capital and diverse customers using
466. 20:12compute to create revenue. Nvidia
467. 20:13compute is a flexible
468. 20:16complete fungeible AI factory platform
469. 20:19improved over time by CUDA serving broad
470. 20:23workloads and retain residual value.
471. 20:26Independent financing platforms let
472. 20:28capital providers underwrite demand
473. 20:30utilization cash flow and residual value
474. 20:33with limited Nvidia residual value
475. 20:35support up to 25%. This builds an open
476. 20:38capital market for AI factories as the
477. 20:40infrastructure powering the intelligence
478. 20:42area. Okay. I mean cool.
479. 20:50All right. Let's see what she said about
480. 20:52this because this was the study that I
481. 20:53started with here today. So
482. 20:56actually this is the study I started
483. 20:58with but this is an update I made to one
484. 21:01of my bots. So done in both bots. Mundav
485. 21:03tested and working. What changed? The
486. 21:04old 20% detector is gone. Bye. You're
487. 21:07right to kill it. It was doing this.
488. 21:10Only hold the ladder still if the book
489. 21:12drops more than 20% 2 hours. Otherwise
490. 21:14keep reiming every 5 minutes. That left
491. 21:17small book changes still walking your
492. 21:19bid around which is the same bug in a
493. 21:22smaller costume. Facts.
494. 21:24Now it's one plain rule. The anchor
495. 21:26locks for four hours once drawn between
496. 21:29reames. The ladder does not move. No
497. 21:33matter what the buck the book do.
498. 21:38No detector, no threshold, nothing to
499. 21:40tune. The cascade falls into a bid that
500. 21:43is standing still and the trigger reims
501. 21:46on the clock instead of one on carnage.
502. 21:48I like that new knob here. Who this?
503. 21:55It's in the config block. Okay.
504. 21:58I also deleted the anchor move PCT
505. 22:01from both. It's dead weight. The lock
506. 22:04handles the big moves.
507. 22:06And the reconcile.35
508. 22:0835%.
509. 22:10Price tolerance already stops churn when
510. 22:13a re aim only shifts things a hair. One
511. 22:17bot bug I caught while doing this. In
512. 22:19the single bot, the sanity rails ran
513. 22:22before the lock did. Okay. So, a cascade
514. 22:25that dragged the fresh anchor under 02
515. 22:27would have made it cancel everything and
516. 22:29stand out. Proof it works. Raman both
517. 22:32dry. Okay. Trigger lock. Beautiful. The
518. 22:34trigger says 04. Okay. The overnight
519. 22:37drift shows your worry was real. Not
520. 22:40theoretical. Since yesterday, cash
521. 22:42anchor moved from 06 to 056. Yes. and
522. 22:46Fartcoin from 05 to 0.047
523. 22:50purely from the book changing under it
524. 22:52[snorts] under the old code.
525. 22:57Okay, so this is actually super fire. So
526. 22:59I'm going to get you all the code here
527. 23:01because um I did make an update to on my
528. 23:05Twitter and YouTube. I posted this code
529. 23:08already yesterday, but um now it's
530. 23:10updated with a pretty pretty pretty
531. 23:15important update here. So, since this is
532. 23:18a pretty important update, let me go
533. 23:20ahead and just scroll through all the
534. 23:21code for you so you can see it all
535. 23:23because if you can see it all, you can
536. 23:25be it all. Okay, I'll have to go fast
537. 23:27because you know time is money. B, let's
538. 23:29get it. So, let me go ahead from the
539. 23:32stop the t stop the top here. Mundas
540. 23:35cash liquidation hunter. Okay, you can
541. 23:37use this for any symbol and this is just
542. 23:39a much better way to take a bias in my
543. 23:42opinion, but the problem is that my
544. 23:45opinion doesn't really matter. So, I'm
545. 23:46running a study over here on the left
546. 23:48side in the middle. But in the meantime,
547. 23:50this is a very interesting liquidation
548. 23:52bot. While most people will just get
549. 23:54long or short, I tend to take over
550. 23:57people's liquidations because that's
551. 23:58what Hyperlid does. That's how they've
552. 24:01made, I don't know, $500 million or
553. 24:03whatever. So, you can see this is the
554. 24:04lick ladder here. The variable is this
555. 24:06is if they get wiped out essentially I
556. 24:09can see everybody's positions. So at the
557. 24:11end of the day if you're able to see
558. 24:13everybody's positions well if you want
559. 24:15to be long or short well it's better to
560. 24:17take over their positions when they get
561. 24:18liquidated. They're a force seller.
562. 24:20They're probably smart right?
563. 24:23They're probably smart people. So
564. 24:26I'm not the sharpest tool in the shed.
565. 24:28So I'd rather take over somebody else's
566. 24:30positions than just try to decide.
567. 24:33You can see here that this builds the
568. 24:35rungs essentially. It looks at where
569. 24:39people will get liquidated and then this
570. 24:41is the exit here. So once in position
571. 24:44and this exact piece of code is
572. 24:48something that caught a 200% move on
573. 24:51Cash Cat. That's crazy. I don't think
574. 24:53that's going to happen again, but maybe
575. 24:55maybe. And if it do, uh, we'll be ready.
576. 24:59We'll be ready. So you can see if it's
577. 25:02in position, it's going to go ahead and
578. 25:04sell in the supply zone. So it would
579. 25:05have already been sold,
580. 25:08but I want to get you all the updated
581. 25:09code here because I think if I spend my
582. 25:13time chasing Jim Simons every single
583. 25:15day, you know, I'm going to get plenty
584. 25:16of edge out of this. And it doesn't
585. 25:19hurt. I mean, it probably hurts a little
586. 25:20bit to share it with you, but I think it
587. 25:23also I also benefit quite a bit because
588. 25:25I'm around a bunch of people that are
589. 25:27interested in algorithmic trading and
590. 25:29you guys have great ideas. So, people
591. 25:31always ask me, why do I share this?
592. 25:33Nobody else shares it. That's why. And
593. 25:35if I share it, I get a ton of good ideas
594. 25:37as well. So, thank you. That's all of
595. 25:38the code for that liquidation bot update
596. 25:42here. And now you have it. Now you can
597. 25:45do anything. Obviously, don't just run
598. 25:48it bare, but I'm still waiting on this.
599. 25:51And this is going to be very interesting
600. 25:52to me. So, let's go ahead and um hurry
601. 25:56up and wait here. Let's see what the
602. 25:57main says.
603. 26:02Waiting for one background to finish.
604. 26:06Is it going to finish?
605. 26:17model.
606. 26:26Okay, it's still working, which is good
607. 26:27to see.
608. 26:32So, we're just waiting on this.
609. 26:35How's it going?
610. 26:41Porsche signs a $ 1.5 trillion dollar AI
611. 26:45deal.
612. 26:48Billion, [laughter] I was going to say
613. 26:50trillion, dude. They signed a trillion
614. 26:52dollar deal. Damn.
615. 26:56Why is the US artificially creating a
616. 26:58crypto bull market at the moment? The
617. 27:00simple, the reason is simple. Two words,
618. 27:02debt transformation. The US bond yields
619. 27:04rising. It shows no one is buying US
620. 27:06debt anymore. The solution from Trump's
621. 27:08think tank is short-term. The government
622. 27:11buys long-term. The crypto circle buys.
623. 27:14How does the crypto circle buy? By
624. 27:16purchasing the USD stable coins because
625. 27:18the reserve. Okay, that's um interesting
626. 27:21interesting idea.
627. 27:24This is top tier conspiracy. US stocks
628. 27:26are the first globally to be tokenized
629. 27:29on chain as global premium assets. The
630. 27:32724 trading of stocks
631. 27:36will inevitably bring a surge in global
632. 27:38trading volume which will drive onchain
633. 27:41prosperity. Onchain prosperity will
634. 27:43boost the continuous upward trend of
635. 27:45crypto's second brother Ethereum's
636. 27:46native token. Onchain and stocks will
637. 27:50form a spiral upward interaction
638. 27:53gradually increasing the total issuance
639. 27:56of stable coins continuously
640. 27:58strengthening the purchasing power of
641. 27:59the US debt. Interesting. Okay. So,
642. 28:03nobody's buying the bonds or the debt, I
643. 28:05guess. And the people that are are the
644. 28:08US stable coins. So, you've got like
645. 28:12Circle buying
646. 28:14bonds and whatnot, I think. I mean,
647. 28:16that's how I'm understanding it at
648. 28:18least.
649. 28:21Everything just changed. 6 days ago,
650. 28:22Bitcoin was trading at 62,000. Today,
651. 28:25BTC is over 80,000. Soul went from 70 to
652. 28:2875. yada yada. Okay, mint of the day. No
653. 28:31way. 32 for finishing 37.
654. 28:36I'm out.
655. 28:40Coinbase tokenized stocks are live on
656. 28:42base. Day one liquidity is launching on
657. 28:44Aer Drrome. Okay. Available to persons
658. 28:47in eligible jurisdictions outside the
659. 28:49US. BTC just ran out of liquidity above.
660. 28:53The biggest liquidation clusters are now
661. 28:55below. There is only one obvious target
662. 28:58left. You know what that means?
663. 29:03Uh-oh.
664. 29:07Bitcoin is going to 100K in 3 months.
665. 29:09BTC just reclaimed the 200E moving
666. 29:11average this week. The most important
667. 29:13level for a trend reversal. When BTC
668. 29:16reclaimed the 200E MA last bare market,
669. 29:18it pumped 48%.
670. 29:21In the next 90 days and started a new
671. 29:23bull run. If history repeats, the bottom
672. 29:25is in and the biggest bull run starts
673. 29:27now. This is super interesting to see.
674. 29:30Everybody has a different opinion.
675. 29:31That's why trading works. Some people
676. 29:32are long, some people are short. Justin,
677. 29:34100 million worth of longs liquidated
678. 29:36from the crypto market in the past hour
679. 29:39as Bitcoin falls under 76,000.
680. 29:43The GTA 6 leaker just burned all the
681. 29:45cryptos he had over $1 million just to
682. 29:47prove a point. It's not about the money.
683. 29:49Rockstar is officially
684. 29:56Samsung down 8% despite announcing the
685. 29:58biggest ever shareholder payout of 80
686. 30:00billion. Damn. The company has committed
687. 30:0221 billion in dividends and 10 billion
688. 30:04in the buyback so far with the rest only
689. 30:07being decided in January 2027. The
690. 30:09market is not willing to wait longer.
691. 30:12Bearish whales have built a huge sell
692. 30:14wall at 80,000 for BTC. Bullish whales
693. 30:17are stacking bids at 73 to 74,000, but
694. 30:20they're outnumbered.
695. 30:24Breaking Bessant to unleash 1 trillion
696. 30:28war chest to rescue the US bond market.
697. 30:30Okay, that is 250x the Treasury's new 4
698. 30:34billion for a single long-term bond
699. 30:36buyback operation. According to CNBC,
700. 30:38the Treasury could tap its nearly 1
701. 30:40trillion Treasury general account to
702. 30:42fund expanded bond buybacks. The move
703. 30:45could lift bond prices and push
704. 30:47long-term yields lower without requiring
705. 30:49the Fed to intervene. This comes after
706. 30:52Treasury doubled buybacks from 2 billion
707. 30:54to at least 4 billion per operation. But
708. 30:58the relief quickly faded and the 30 year
709. 31:00year the 30-year yield returned to 5.25%
710. 31:04showing that far more support may be
711. 31:06needed. Drawing down Treasury cash could
712. 31:09also inject major liquidity into the
713. 31:11markets, potentially benefiting stocks
714. 31:13and crypto. The full one trillion has
715. 31:15not been committed, but it represents
716. 31:17the Treasury's potential firepower. The
717. 31:21bond market just learned how big
718. 31:22Besson's bazooka could be.
719. 31:28Besson could unleash one trillion. Okay.
720. 31:32If you hold a memecoin, you just get
721. 31:34stock tokens air dropped to you. So,
722. 31:36this is what Vlad, the founder of Robin
723. 31:38Hood, said on a podcast that dropped
724. 31:41yesterday essentially, and it made Cash
725. 31:43Cat go absolutely berserk.
726. 31:49All kinds of
727. 31:51unique pools that we didn't even think
728. 31:53would exist and and all kinds of
729. 31:55protocols where, for example, if you
730. 31:57hold a memecoin, you just get uh stock
731. 32:00tokens uh air dropped to you, right? So
732. 32:04they've like connected meme coins, core
733. 32:07cryptos, and the stock tokens and
734. 32:09composed them into into unique products
735. 32:11that yeah, we we wouldn't have
736. 32:14anticipated creating, we probably
737. 32:16wouldn't have thought of to make
738. 32:17ourselves. So it's it's um Robin Hood
739. 32:20chain is this combination of developer
740. 32:24platform with uh financial primitives uh
741. 32:28like high yield and stock tokens the
742. 32:31chain but that
743. 32:33>> okay I mean he's bullish
744. 32:37Michael strategy sold $5 billion
745. 32:41of BTC
746. 32:43clog codebd pays me blah blah blah Apple
747. 32:47pay tour is crazy. Walmart is finally
748. 32:50rolling out Apple Pay. Thank God. Thank
749. 32:53God. How long is it going to take to get
750. 32:54to Puerto Rico is the question. October
751. 32:578th. All right. I'll be at Walmart then.
752. 32:59Until then, I'm not going. This post was
753. 33:02suspended. Okay. Okay. Finally, we are
754. 33:04finished here. So, this is what we
755. 33:06started with the study of whether buying
756. 33:09liquidations is better or not. And this
757. 33:11is the truth based off what she says.
758. 33:13All three agents are finished. Here's
759. 33:14what came out.
760. 33:16What your bot does, it reads every long
761. 33:18position on Cash Cat, sorts them by
762. 33:20liquidation price and finds the exact
763. 33:22price where 83% of the reachable long
764. 33:24books gets force liquidated. It rests
765. 33:27one bid at.5% under the price and
766. 33:30redderes it every 5 minutes as the book
767. 33:32shifts. It isn't predicting price. It's
768. 33:35standing at the spot where forced
769. 33:36selling has to land and runs out. Exit
770. 33:39is the midpoint of the highest wick and
771. 33:42highest close over 30 days. So, there is
772. 33:44an update that we just made here. It's
773. 33:46not every 5 minutes. Now, I I feel like
774. 33:47that's pretty much a bug. Not a bug, but
775. 33:49it's just like, are you really going to
776. 33:51liquidate 83% of the book in 5 minutes?
777. 33:54Well, it did.
778. 33:56It actually did happen. The answer, and
779. 33:58it wasn't what uh any of the three
780. 34:00agents first reported. All three agents
781. 34:02said proxy tested a proxy rest a bit at
782. 34:05x% below the recent price. That proxy
783. 34:08graded out weeks to negative everywhere.
784. 34:10But that proxy is a is dip buying and
785. 34:12it's not what your bot does. When I
786. 34:15checked the actual 822 crash, the
787. 34:18difference was the entire result.
788. 34:21So entry here, 18% trailing close, 25%
789. 34:24trailing close. Your lick anchor here.
790. 34:27You can see here the bid would would
791. 34:28have been 10% if you just had a uh 18%
792. 34:32dip by
793. 34:34and you would have been liquidated on 2x
794. 34:36leverage, 3x leverage. You would be up
795. 34:3860% on
796. 34:401x levers. No leverage. 25% below
797. 34:44trailing close would have been put the
798. 34:46bid at 9 cents and would have been up
799. 34:4975% or if you're using 2x leverage, it
800. 34:52would have been liquidated. 3x leverage
801. 34:54would have been liquidated. Now, this is
802. 34:56what I had built in. Your lick anchor
803. 34:59would have put the bid at 04 cents. 04
804. 35:03cents. Let's go ahead and look at it.
805. 35:05app.hyperlquid. exactly where that
806. 35:08position would have filled. Okay, cash
807. 35:10cap at 0.04 cents with the bot I
808. 35:13literally gave you today. Okay, it would
809. 35:16have filled right here.
810. 35:21It would have filled right there. So,
811. 35:22you can go ahead and do the math on that
812. 35:24move
813. 35:27just like free on YouTube. That's insane
814. 35:29to me. But whatever. Let's let's keep
815. 35:31getting I'm going to keep dropping
816. 35:32everything I know and hopefully you send
817. 35:34me back an idea or at least some
818. 35:36motivation here and there. Thank you. So
819. 35:38you can see the lick anchor here would
820. 35:39have um with no leverage 272%. With 2x
821. 35:43leverage 543%.
822. 35:45With 815 with 3x leverage it would have
823. 35:49been 815%.
824. 35:51Same crash same exit at.1693
825. 35:55same 2.3day hold. The dip buyer was
826. 35:57right about direction and still lost the
827. 36:00account. Your bid sat at half the depth.
828. 36:04So, it's never more than 17.5%
829. 36:06underwater and 3x leverage is around
830. 36:0830%. So, the edge isn't buy
831. 36:10liquidations. It's that the lick book
832. 36:13tells you where forced selling runs out.
833. 36:17A percentage bid parks you in front of
834. 36:20the cascade. A cluster bid parks you
835. 36:23underneath it. Your intuitions were half
836. 36:26correct. Small tokens, yes, that's where
837. 36:28the cascades are violent enough to make
838. 36:31a real discount and the book is small
839. 36:32enough to map. But size,
840. 36:35but size, it is
841. 36:39but size of that is a lottery ticket
842. 36:41because cash Monte Carlo came back with
843. 36:43a medium of zero and a mean of 5,194
844. 36:48on a starting of 1,000.
845. 36:51Fat right tail ugly middle big tokens no
846. 36:54confirmed twice independently forward
847. 36:55returns after real long liquidations on
848. 36:58BTC and ETH were worse than random only
849. 37:012.5% of the biggest down wicks even
850. 37:05matched a recorded long lick event
851. 37:08flush buying
852. 37:10flush buying majors at 3% trigger lost
853. 37:14money in a window where BTC ETH and soul
854. 37:17rose 16 to 25% two things I threw out
855. 37:20the small cap App agent concluded
856. 37:22shorting wins.354%
857. 37:25a day verse 079%.
858. 37:28I checked the window. Eight out of 11
859. 37:30coins fell. Median
860. 37:34-10%.
861. 37:35That's a bet on the tier declining
862. 37:37measured in a window where the tier
863. 37:40declined. Cash cat rose 92% in the same
864. 37:43window. Discarded. The Monte Carlos
865. 37:45agents flush numbers are biased. Low
866. 37:48construction block bootstrapping 24-hour
867. 37:51chunks separates a crash from the bounce
868. 37:54that follows it while keeping the drift
869. 37:57which deletes exactly the structure the
870. 37:59strategy trades. Its directional numbers
871. 38:02are fine. Buy and hold has no timing to
872. 38:04destroy. So that half stands for the
873. 38:07person asking can AI just trade for me.
874. 38:09Liquidation hunting beats a directional
875. 38:11bet for three reasons none of which are
876. 38:14a prediction. You name your price in
877. 38:17advance instead of taking whatever the
878. 38:19market gives at the moment that you feel
879. 38:22confident, which is usually the worst
880. 38:24price of the day. Your worst case is no
881. 38:26fill rather than wrong. And you're the
882. 38:29maker at 015% instead of the taker at
883. 38:32045. The honest catch fills are rare.
884. 38:35roughly once per coin per two to three
885. 38:38weeks. So, capital sits idle
886. 38:42and per day expectancy looks
887. 38:44unimpressive
888. 38:46even at 80% win rate.
889. 38:49Two things you should know. Your logs
890. 38:51don't contain the fill yada yada wasn't
891. 38:54running. The dual log had six flat and
892. 38:58the risk the bot runs leverage three at
893. 39:0095% equity
894. 39:04on 82. that survived because the anchor
895. 39:06happened to resolve deep. Every flush
896. 39:09fill in the sample deeper than 8% below
897. 39:12reference would have liquidated at both
898. 39:142x and 3x despite winning at 1x. I'd
899. 39:17gate leverage on how far the anchor sits
900. 39:20below the mark rather than leaving it at
901. 39:23fixed three. The full writeup is here
902. 39:25with each agent's
903. 39:27um three subfolders. Okay, that's
904. 39:29actually super informative. So, I think
905. 39:32I've got everything I need to know about
906. 39:33this. And um my conclusion, okay, so my
907. 39:37conclusion from this is that this may
908. 39:40work on small coins, but not on on big
909. 39:43coins. Like Cash Cat is eliquid
910. 39:45essentially. So, uh if I were to run
911. 39:48this, it would have to be on small, not
912. 39:49BTC, not ETH, but the liquidation edge
913. 39:53when it comes to capitulation is really
914. 39:55on smaller coins with low leverage, like
915. 39:582x leverage max. You know, I know this
916. 40:01printed on 3x leverage, but I don't want
917. 40:02to get greedy. Go ahead and um let me
918. 40:05know if that thinking is correct based
919. 40:06off of your ideas. Thank you, Phil.
920. 40:09What's up, bro? Much love. 777.
921. 40:14We're over here in the Zoom. If you do
922. 40:15want to jump into the Zoom, we have the
923. 40:17Zoom every single day. I'm putting the
924. 40:19API keys in the chat. There's a bunch of
925. 40:21good stuff I give you when you do come
926. 40:23to the chat. You can see here all the
927. 40:26things that I give you. Um, I explain
928. 40:30them here.
929. 40:34You get uh daily Zoom calls with me
930. 40:37from, you know, 12:00 to 3:00 p.m.
931. 40:40Eastern. AI trading bot master class. I
932. 40:42give you replays of every call, so if
933. 40:44you can't make it, don't worry. Uh, you
934. 40:45get the Mundave API key. I just posted
935. 40:47that in the chat. The Mundave AI is
936. 40:49controlled by the API key as well. The
937. 40:51Mundave app, uh, that's what I'm using
938. 40:53as I have all these AIs working for me
939. 40:55all day. the AI training road map and
940. 40:57then you can ask me anything live. So
941. 41:00the docs go ahead and explain what the
942. 41:02AI the API does and also has the AI in
943. 41:05here. You can see the docs are pretty
944. 41:07thick. There's a lot of cool custom
945. 41:08data. This is you know what I use so you
946. 41:12get access to it if you're in the crew.
947. 41:14Uh you also get the road map so you can
948. 41:16get caught up to speed with me what I'm
949. 41:18doing. Also um just how to get started
950. 41:21with automated trading. If you want to
951. 41:22build a trading bot, you want to use AI
952. 41:23for trading, this will walk you through
953. 41:25it. Um, this is all available inside of
954. 41:28the community. Um, but I want to finish
955. 41:31this off here with answering this. And
956. 41:34while I'm waiting for this, I'm going to
957. 41:36go keep going through this news and
958. 41:37we'll see what's going on. Breaking
959. 41:38news? Nope. Uh, should I take a 1
960. 41:42million profit? Nope. Uh, 3 Inc. Club?
961. 41:45Nope.
962. 41:46NFTTS? Nope. Do you see it? Nope.
963. 41:49[snorts] Over 30 trillion. President
964. 41:51Trump, you should be buying right now.
965. 41:52Over 30 trillion could soon be unleashed
966. 41:54in the crypto market. That would be
967. 41:55insane. Justin said, "Bitcoin records
968. 41:58largest weekly gain since March 2023."
969. 42:00Wow. That was since 2023. Tomorrow will
970. 42:03be the worst day for of 2026 for the
971. 42:06markets. You must read this before
972. 42:07August 24th. Jeez, dude.
973. 42:11Copage.
974. 42:12Your tier call is correct. Your leverage
975. 42:14call is right in direction but aimed at
976. 42:17the wrong variable. And there's a bigger
977. 42:19risk in the config that you didn't name
978. 42:22where you're correct. Small coins. Yes,
979. 42:25Bitcoins, no. That holds up hard on BTC
980. 42:27and ETH. Ford returns after real long
981. 42:30liquidation cascades were worse than
982. 42:32random. And only 2.5% of the biggest
983. 42:36down wicks even matched a recorded lick
984. 42:38event. The book is too deep and
985. 42:40forcelling
986. 42:42to move price isn't a real discount. On
987. 42:45small coins, the whole reachable book is
988. 42:47mappable and the cascade actually
989. 42:49overshoots. Two independent agents
990. 42:52landed there separately.
991. 42:54where the reasoning needs adjusting. I
992. 42:57think you're picturing 2x is buying you
993. 42:59a safety cushion. 3x dies at 30%. 2x
994. 43:02[snorts] dies at or survives 45%. So 2x
995. 43:05gives you
996. 43:0715 more points of room. I tested that
997. 43:10across 571 real fills. On small cats,
998. 43:13that cushion is empty. Postfill draw
999. 43:15downs are bi modal. Either the thing
1000. 43:20surely or barely dips. 78% of the fills
1001. 43:23stay under 10% underwater or it
1002. 43:25collapses 60%. Of 67 small cap fills,
1003. 43:28zero landed in the 30 to 45% band where
1004. 43:322x lives and 3x dies. Every fill that
1005. 43:35killed 3x also killed 2x. Interesting.
1006. 43:38On majors, the band does contain 2.8% of
1007. 43:42the fills. So 2x helps there, but not
1008. 43:44trade but you're not trading majors. So
1009. 43:472x doesn't survive crashes that 3x
1010. 43:50wouldn't. Interesting. Okay. The
1011. 43:52variable that actually matters most is
1012. 43:54percent of equity. You're putting 95% of
1013. 43:56the account behind one fill. I
1014. 43:58compounded the real fills over 50
1015. 44:00trades, 20,000 runs.
1016. 44:03Zero chance of ruin here at 1x leverage.
1017. 44:07That makes sense. 43% chance of ruin
1018. 44:11here at Okay. So, it's this. The thing
1019. 44:15that matters is the size per fill.
1020. 44:19Dropping it from 95 to 50% sizing cuts
1021. 44:23ruin from roughly a third to under 1%.
1022. 44:27That's a far bigger lever than anything
1023. 44:29between 2x and 3x. Note that the 2x
1024. 44:32scores worse than 3x at 95% sizing,
1025. 44:37which is the empty gap effect showing up
1026. 44:40in the compounding. I would not act on
1027. 44:44that ordering. The sample is too thin to
1028. 44:47rank 2x against 3x. But it does tell you
1029. 44:50the 2x versus the 3x choice isn't where
1030. 44:53your risk lives. The part I'd push you
1031. 44:56further on. 1x wins outright. And the
1032. 44:58reason is the whole thesis of your
1033. 45:00strategy. At 1x you cannot be
1034. 45:02liquidated. The 9% of fills that creator
1035. 45:0660% become a draw down you sit through
1036. 45:09instead of an account that ends. Your
1037. 45:12edge is the price comes back after
1038. 45:14capitulation. Leverage is the one thing
1039. 45:17that guarantees you aren't there when it
1040. 45:19does. On 822, your anchor sat 17.5%
1041. 45:23above the low, which survived 3x with 12
1042. 45:26points to spare. If the trigger has
1043. 45:28resolved a bit shallower, the same
1044. 45:30cascade would have taken the account
1045. 45:32before the bounce that happened. And the
1046. 45:35thing that actually made 822 work was
1047. 45:38bid depth, not leverage. The lick
1048. 45:41anchored bid was 17.5% underwater. A
1049. 45:44mechanical 25% below price bid was 61%
1050. 45:48underwater on the identical move. That's
1051. 45:50a 3.5x safety improvement. No leverage
1052. 45:53setting can buy you. So revive small
1053. 45:56coins only. Keep it low leverage. Keep
1054. 45:58the instinct, but go with 1 to 2x rather
1055. 46:01than treating 2x as a safe ceiling.
1056. 46:05and understand you're buying
1057. 46:06survivability not cushion
1058. 46:10the cut then cut the sizing because 95%
1059. 46:14of the account on a single fill is where
1060. 46:16the risk of ruin happens caveat worth
1061. 46:19holding this is a 67 small cap fills
1062. 46:23around 188 days okay so I think this is
1063. 46:26interesting now I am using 95% of the
1064. 46:29account but controlling risk with the
1065. 46:31size of the account it's not my whole
1066. 46:32book of course that I'm putting into
1067. 46:35that account. So, how does that differ?
1068. 46:37Because to me, yeah, how does that
1069. 46:40differ? Because if you get liquidated,
1070. 46:41you get liquid. If you're using
1071. 46:42isolated, unless you're infirm that
1072. 46:44inferring that I use cross margin here,
1073. 46:46cuz cross margin would obviously bring
1074. 46:48down the liquidation point.
1075. 46:51So, this guy is
1076. 46:54fear-mongering, so I'm not even going to
1077. 46:56read it. It's too much slot for me. BTC
1078. 46:58is probably from 63K to 79K in just a
1079. 47:00few days, and 5 billion has been
1080. 47:02liquidated. About 90% of the people are
1081. 47:04opening shorts and very few dare to FOMO
1082. 47:06into the market. But fear and greed
1083. 47:08index is 77
1084. 47:1074. Now what I see is that market makers
1085. 47:13have manipulated this index and will no
1086. 47:15longer monitor it. Damn.
1087. 47:19While everyone is distracted by the
1088. 47:20manipulated oil price. Diesel is trading
1089. 47:23at 183 per barrel. WTI 87, diesel 183.
1090. 47:27People say oil is the lifeblood of our
1091. 47:29economy, but it's actually diesel.
1092. 47:31diesel powers, most farming operations,
1093. 47:34most mining operations, 80% of freight
1094. 47:37by weight. We're going to feel the
1095. 47:40inflationary pressure of $183 diesel,
1096. 47:43while crude is struggling to stay above
1097. 47:4580. Brace for inflation.
1098. 47:48Apparently, our Reddit and Marvel had
1099. 47:50two major catalysts recently, the 12.2
1100. 47:53billion Google TPU ecosystem agreement
1101. 47:55and Spider-Man brand new day. Okay.
1102. 48:01All right, so it looks like I went
1103. 48:02through all the news here. I'm all
1104. 48:03caught up to speed. Cash Cat is surging
1105. 48:05still.
1106. 48:07Let's read this and then I think I'm
1107. 48:09going to put this to rest. You're right
1108. 48:11on both counts and my sizing critique
1109. 48:12was aimed at the wrong level. Funding a
1110. 48:14sub account with a slice of the book is
1111. 48:17position sizing, right? Isolated margin
1112. 48:19confirmed in the code at the code point.
1113. 48:22So, a liquidation is capped at the sub
1114. 48:25account and can't reach the rest. But it
1115. 48:27changes which metric decides the
1116. 48:29leverage question and flips the answer
1117. 48:31in direction that doesn't help 2x two
1118. 48:34different games. If you restake a fixed
1119. 48:36slice of sweep uh and sweet profits back
1120. 48:40to the book, you're making repeated
1121. 48:42bounded bets. Okay. Arithmetic expected
1122. 48:46value rules and leverage raises it. So
1123. 48:49leverage 1x
1124. 48:51can't get liquidated.
1125. 48:53Leverage 2x 3.6x
1126. 48:57liquidated.
1127. 48:59If a sub account compounds, geometric
1128. 49:01growth rules and leverage eats it. 0%
1129. 49:05dead at 1x. I mean, no leverage is
1130. 49:07always the answer. It's always the
1131. 49:09answer. I'm just over here getting
1132. 49:11greedy.
1133. 49:12This was a 238% move. It's always the
1134. 49:15answer, dude. It's always the answer.
1135. 49:18No. So 3x has the highest mean in both
1136. 49:20models and a terrible median. So you can
1137. 49:22see the mean
1138. 49:25is 62,000 but the median is very low.
1139. 49:30That gap is the whole story. A fat right
1140. 49:32tail you'll probably never personally
1141. 49:34experience. The catch you should check.
1142. 49:36Your mental model is the first game. The
1143. 49:38bot is coded for the second percentage
1144. 49:41of equity is 95% sizes off whatever
1145. 49:45equity is sitting in the sub account
1146. 49:48right now and wins automatically relever
1147. 49:51into the next fill unless you manually
1148. 49:53sweep profits out. It compounds and
1149. 49:55compounding is where 2x and 3x get
1150. 49:57punished. to actually be in the bounded
1151. 50:00bet regime, you'd cap notional at a
1152. 50:02fixed dollar amount rather than a
1153. 50:05percentage of current equity. On 2x
1154. 50:08specifically,
1155. 50:09it's the one choice the data doesn't
1156. 50:11support. 2x comes last or near last in
1157. 50:15both games. In the bounded version, it
1158. 50:18takes the liquidation risk of leverage
1159. 50:21and returns less than 1x. In the
1160. 50:24compounding version, it has the worst
1161. 50:26median of the three. That's the empty 30
1162. 50:29to 40% draw down.
1163. 50:32Again, on small caps, the fills that
1164. 50:35kill the 3x also kill the 2x. So, you
1165. 50:38pay for leverage without getting
1166. 50:39protection. The coherent picks are 1x if
1167. 50:42you're compounding or 3x if you're
1168. 50:44genuinely restaking and a fixed slice
1169. 50:47and sweeping. 2x, don't get greedy, is
1170. 50:50the middle ground. is a dead middle
1171. 50:53where I could be wrong. The whole 2x
1172. 50:55conclusion rests on zero of 67 small cap
1173. 51:00fills landing in a 30 45% band with a
1174. 51:04sample that size. Something that truly
1175. 51:07happens 3% of the time would show up
1176. 51:10once or twice. So seeing zero is weak
1177. 51:12evidence rather than proof. If small cap
1178. 51:16flushes do sometimes stop in that range,
1179. 51:182x earns its keep.
1180. 51:21The tier call and the sweep or compound
1181. 51:24distinction are solid.
1182. 51:27The 2x versus 3x ranking is the part I'
1183. 51:30want another few hundred fills before
1184. 51:33trusting. Okay, I understand it now. So,
1185. 51:35you must sweep. You must sweep the
1186. 51:37account. If you compound, you're cooked.
1187. 51:40That's it. I've already shown you all
1188. 51:42the code. Now you know everything I
1189. 51:44know.