PnL Enthusiast @TurtleNeckCap

Doesn’t it make sense to be long non-SPCX space tickers into the Aug 4 unlock? Surely insiders short the non-SPCX names to hedge exposure and will cover some into the unlock. Some are down ~50-60% since SPCX IPO

Added some more $ORCL and AI adjacent names. Read this quote and tell me you think they won't backstop the buildout. Scott Bessent verbatim: "We probably used to have about 50 or 60 percent of the compute power in the world, we'll probably soon be at 80, and that is an incredible national advantage ... I'm running a lot of the AI effort for the administration and what we are trying to calibrate is there's an innovation race between us and the Chinese that we just cannot lose. if the Chinese were ahead of us in this, I can tell you, this mischief and the havoc that they would wreak and the leverage that they would have over us, and the leverage that they would have over us, I don't even want to think about it. It's unconscionable. But we are ahead and we want to stay ahead"

Podcast Alpha @PodcastAlphaX

3 days ago

$NVDA's demand story just got a Treasury-level backstop. Scott Bessent @SecScottBessent says the US is heading to 80% of the world's AI compute. The Treasury Secretary frames US-China AI as a race for compute share: from 50-60% today to a projected 80%, with the US about a year

4 19 142 19K 70

@Investor_NICK_ @CapitalObserver This is where you buy more brother. And if it doesn’t regain the previous support slightly above you eject and try again if it regains it or retests $100 area. I may have just upped it to 30% of my portfolio from 10% this morning (currently down 2% total on the position)

Screened all of the recent momo names that were part of every leg of the unwind and >$500k ADV 141 names, guess how many had net insider buying >$100k since the unwind began? Zero. Insider selling >$100k? 38

keyboard_arrow_left Previous keyboard_arrow_right Next

The only thing I wish they’d do differently is really lean into being a competitor to Snowflake & Databricks. Seems they may be headed in that direction but I think it should be a priority. Listless I can work with as long as the underlying economics are there but yeah, probably won’t outperform a neocloud that’s executing well and can secure financing. Feels to me like it has less risk but maybe I’m deluding myself

6 months? $ORCL Makes no sense to me that people are bidding the hyperscalers but selling Oracle. Hated & moderately levered play on (somewhat) similar economics, well connected to Trump & politically in general. Ellison is a solid operator. Added today partially as a hedge to my book which is more inversely correlated with momentum than I’d like

Late reply but agree w/ the general logic. Think they’ll do fine long-term even if OAI did so poorly they couldn’t honor obligations, but if OAI valuation starts coming into question no doubt ORCL will trade in sympathy. With the 6 month timeframe GPT6 performance is likely to be the determining factor and I expect it to be significantly stronger than the new Chinese models. TBD if that can shift the narrative back to OAI and make ORCL work in that timeframe. Personally think it will but definitely something to consider

Probably can't provide any insight into HFC pricing that you haven't already found, imo it will be in line with their guide. Longer-term I think prices firm but HDSN seems like more a volume story than a pricing story currently (+20% YoY Q1 volume). Margin expansion obviously the kicker if HFC pricing does improve would be a 2nd large tailwind

Built a little homemade screener for lobbying spend/mcap, here are the results (2025 and Q1 2026). Will filter out the bios and post another top 25 later. Interesting to see $HDSN on there, only name I own from the list.

keyboard_arrow_left Previous keyboard_arrow_right Next

Adding more $ORCL here (5%->10%). Partially because I think it could do really well going forward and partially because it balances out the fact that my current portfolio is very negative momentum. If you're bullish on the 'hyperscalers' it seems hard not to be bullish on ORCL

Agree with this completely, doesn’t matter what the Fed does when we’re slated to run >6% GDP deficits for the foreseeable future. I’d actually go further and take an Sargent-Wallace-esque view (eg their paper “Some Unpleasant Monetarist Arithmetic”) that raising rates would reduce short-run inflation by moving it to the long-run tail and increase the risk of fiscal dominance over the long-run. That being said I mostly ignore macro, but, interesting to think about 😂

In my humble opinion there's still something going on with subagents; I was not running into usage limit issues prior to the 5.6 launch and I'd run four 5.5 xhigh agents to audit company financials. Post launch I was eating into my usage dramatically even running 8x luna medium agents for the same task. I've had Sol Ultra do some custom tweaks to limit the context fed to the subagents (fork_turns: "none" and only feeding it necessary context from main agent) and it's improved a bit, but given the cost comparison charts I've seen between GPT 5.5 xhigh and Luna medium it still feels like something is up. From the charts you'd think I'd be able to get away with 16x Luna medium subagents and it'd be less than 4x 5.5 GPT xhigh, but that has not been the case in my experience. Anyway you guys are doing great work, it's still an incredibly useful model I just get the sneaking suspicion something is "off" with subagents and potentially context. And if you do read this far: any recommendation on whether, for a repetitive task like analyzing 50 companies within an industry, it's better to spawn one agent per company and kill finished agents or keep the agent running and have it move to the next company?

I don't currently understand if there's a difference in functionality between Work and Codex, but directionally I love the change. I've been using Codex as an 'agent harness' to spin up Luna agents who flag companies in the equity universe where it feels there is a high probability of abnormal returns. It then passes the task off to Sol agents that get access to much more information and do a deeper dive on the flagged universe, and finally to a ranking agent that attempts to rank the remaining candidates by confidence level. So not being pigeonholed into a pure coding harness - although it still seemed to work alright - is a blessing. Keep up the great work! And if you read this far two random thoughts on where OpenAI can have a massive impact, although I'm sure you already know: online education (on demand tutor for classes/companies like Coursera) and basic financial planning and advisory. I've used your "in-program LLM" that utilizes local subscription to create a mini financial planner for a friend: they talk to it about their goals and risks and it comes up with a low-expense ETF portfolio tailored to their risk tolerance.

Complete portfolio rebalance yesterday and today. Sold down Coursera, doubled FLNC position and added ORCL yesterday. All highlighted names are fresh adds today (except for GPN which I sold yesterday and bought back today because I wanted more diversifiers). $EFOR is probably the sketchiest one but with the $1b buyback authorization and Q2 historically being their strongest quarter, I think it can work and work fairly quickly if it does. Looking to add names I can hold for >1 year to avoid another STCG catastrophe like this year.

I don’t even like real estate (pain in the ass) but you also have to factor in easy access to non-recourse leverage that you can’t get in equities. I agree with the general sentiment that it trails equities on a risk adjusted basis, but if you pick the right market you can make a killing with a low cash outlay

And here is the portfolio it made from my watchlist tickers. I don't think it does great on the portfolio construction side tbh you'd need to go with a more rules-based traditional framework. Also not sure I'd put BR in the 'payments' group

Slop post: For fun I set up GPT 5.5 + Opus 4.8 agents with a GPT 5.5 oracle and had them go through my watchlist. Models spit out 2 valuations for a company then the oracle judges them & determines final valuations. Here's the top 50 by weighted IRR according to the bots. Found it did a reasonable job but underweighted the impairment cases on stuff like WIX/INTU

Becoming more bullish on $VKTX. Wasn’t a fan back when it was the hot stock on Twitter but now it’s rarely mentioned. $3.5 bn seems cheap for a me-too drug that has a decent shot of being more efficacious than tirzepatide and is active orally. People who bought on borrowed conviction have likely been shaken out by now. I get the feeling it’ll begin to move higher over the coming year as the p3 readout gets closer

IMO $KWEB will always be a long-term underperformer simply because the companies it holds have extremely high foreign ownership and little domestic ownership. State wants the economics to accrue to companies w/ high domestic ownership. I'd rather own $KSTR, valuation be damned

Good day to own $QURE. I had planned on doubling my position this morning after hearing what the principal investigator said, but I guess that’s off the table now. Was slightly over 5% position and thankfully managed to make up all of the losses on payments (awful)