Still a work in progress

Tian Xuan Zhou (Oliver)
Bachelors Student, New European College – Munich

This won’t be like the usual post, instead this will be my thoughts on the potential of openclaw and hermes agent. Will be back to the usual frequency upload.

Currently still a work in progress. I caught wind of openclaw around mid March, but didn’t start my experimenting with it until a month later, this was also when I discovered hermes agent. Both allow the user to create AI agents that run on a large language model as its reasoning engine. Sounds very useful, but only if you know how to train it to produce tasks and outputs that you desire, otherwise this is no better than running the LLM by itself, thinking of it like a personalised and picked out response compared to a generalised and averaged response.

Still spending a lot of time in recent weeks experimenting with openclaw and hermes agent, still currently trying things out to see how I can potentially utilize these two agents. So far, the main idea of mine is to train an agent that has a similar thought process as mine, at least to pick up points of interest that I would focus on, but also manage tasks that would save me a lot of time like scanning for names through another broker for stocks of interest pre market that I am not aware of or compile research or analytics that I might find important. This would save me a lot of time and possibly find a lot more ideas and opportunities in the future.

As for the trading itself, I don’t really think that these would have much application, other than feeding it patterns and data, but people have much faster and larger models for that. Think of it like trading with LLMs, if LLMs outputs are all averages of the internet, all LLM outputs will be the same, but all being on the same side of a trade will make nobody money. This is also the main reason why I think that AI won’t take the jobs of traders and portfolio managers. Maybe I am being naive, but at the moment I wouldn’t use LLMs for trading, maybe it’s useful adding another layer for the “course sellers” to sell their amazing courses.

As for the portfolio itself, it is outperforming big indexes, now logged back to back 11% gains in the months of April and May, however the composition, I think it’s a bit of a mess at the moment, far from satisfactory. If I am to describe it, I would say that there is too much activity, for no particular reason. I would like to sort it out as soon as possible but the last few weeks have been quite stale, until the SpaceX IPO.

Speaking of which, a huge catalyst for aerospace in general, but 1.8 trillion in valuation? That sounds so steep. Redwire and all other space concept had a huge couple of weeks. My redwire caught upwards of 200% before I sold it off. This would need to stabilise before any further trades because the volatility right now in general concept stocks, quantum computing for example also riding a big part of the wave.

Amazon, starting to give back some of the gains, already offloaded half of my holdings. I’d expect activity in AI to pick up again but currently being overshadowed somewhat by the SpaceX IPO. The broader AI and tech investments however are much price insensitive, despite the recent economic worries (mostly worried about inflation) and geopolitical tension. There are projections upwards of 1 trillion in AI Capex spending through 2027 despite all the cost increases that are being felt.

Zscaler, top and bottom line beat, however management guided down on earnings call which triggered a premarket move downward of -24%, almost wiping out all my gains. The bottom line beat was significant in my opinion, +43.29%, therefore holding onto this name. The current concern is that Zscaler is threatened of being phased out despite being the market leader within the cyber security cloud space, especially with many cloud providers now offering their own system. However the strong earnings is still a tailwind that makes the name very attractive. This market leadership and current deals allows for further competition especially with high switching costs for enterprise focused cybersecurity providers.

Li Ning, one new idea this week. Steph Curry signing a 10 year sponsorship deal, this is a huge catalyst for Li Ning considering the amount of basketball fans within China, definitely on my watchlist now after the recent underperformance the last 6 months or so.