With a 2 minute time window, why not write about some trading stuff.
What is HR abitrage? It’s a public market investment strategy that invests into equities and derivatives based on HR information, considering that people don’t care about HR. Easy. right? Why do I think about this? Because I realized how strong HR fuckups you identify on Glassdoor and nonsensical hires – found via deltas on linkedIn – predict non-performance in value delivery in a product sale. Something I studied at my prior VC job and kind of like find continuously interesting while asking myself what the issues are with scaling an organization with access to minimal viable talent – that doesn’t ruin these scores.
What are the key insights?
- Market efficiency is limited to what people do.
There are a lot of smart people working on the public markets. We have probably more trading strategies than we have crops of sand at the Atlantic ocean. But the big focus is stiill driven by quants, not hackers. So quants focus on weird stuff like low latency factoring hardware design, low latency low level code on hardware, getting closer to network entry points – understanding network topology and latency in high frequency trading. good for them But terrible ROI. Then there are the guys that rely on the public infrastructure on investor information. The guys that read of automatically evaluate using OCR and NLP and probit and VAR models what analysts are saying about companies and they spot either misjudgements by classifying uncertainty on analyst recommendations or deviation of behaviour driven trends from research consensus. Blablabla. Good for them. Still a shitload of work.
Then there are all the sentiment minders that try to pick up semi structured information from twitter and other sources where insiders leak confidential information without authorities and regulators picking up on it. But the scaling of mining billions of tweets, blockposts and that jazz is still a shitshow for ROI discussions, let alone how hard it is to capitalize on this information in real time and how many pseudo smart guys you have to employ to make it work.
There is a simpler hack. It’s acknowledging that humans suck and that company performance and good old Grahamian value investing is partially based on HR quality. If you ignore the CFO fraud and number magic and cancel it out as noise.
Well, there is glassdoor, kununu, there is linkedIn, Monster, public job boards, etc. All of them have fairly structured content messaging systems and regular approaches in presenting and formatting data in their CMS to publish it. So it’s easy to write scrapers and ML based mechanisms to update scrapers if they change HTML tags. You just have to solve a basic morphism between updated releases, but the visual presentation being the same, that is simpler than running OCR on screenshots of visual that always stay the same due to UX requirements.
Once you tap all those resources, you get a lot of rich HR data. Who moves from A to B. What profiles do they have. What is their career model, what level of bullshit do they sell online to createtheir profile, and how do all those individuals create teams that have political issues that try to fit into a organizational development plan which is created by idiotic HR guys. Lots of weaknesses to spot. Simple to find drivers of nonperformance for individuals.
Hard, but still easy on an ROI perspective.
The next step is that the market doesn’t provide you with derivative and instruments to trade on these opportunities. Buy and hold on a single company by company basis consumes too much capital and the investment horizon is too long. The “mean reversion” life cycle on arbitrage opportunities from HR based investment srategies is too long.
So you have to come up with indexing strategies. Which universe of companies provides the highest upside and the right measures on sluggishness in the share price compares to the overall incapability of the HR pool to provide a medium investment horizon with decent returns? That is the tricky part. But even the portfolio construction exercise can be tackled by crawling idiotic stuff like activity feeds from stocktwits – these guys only become active if they think something is wrong or news are coming around an investment – and the like. So if you solve the volatility delta over time for a selected universe and have general estimtes on HR based mid-term valuation shifts, you can construct tradable portfolios strategies that allow you to capitalize on an opportunity within a few weeks.
The transaction costs and portfolio purchase still kills the deal. So you have to add a way to add some “turbo” or non-linear participation features, so you look at reconstructing portfolio strategies around particular greeks and dynamics that you can capture with higher return on stock price volatility based on capital deployed.
All in all, nobody looks at HR strategy on my knowledge and it probably takes a ridiculously small amount to scale the process.
This raises the question: Why is no fintech tackling the structuring of opportunities in this fashion using A.I.?
It’s all about finding uncommon investment thesis – in this case HR data -, hacking out ways for information access using proxies (for HR data, job boards, for volatility prediction stocktwits and other services), building massive scale sensitivity analysis models to automate portfolio construction and then building a model to identify the secondary market securities that offer a capitalization strategy that offers highest ROI on capital deployed.
Isn’t it obvious? Or where is is the logic flaw here?
And yeah, the stuff we do at Acellere is still more interesting than this stuff. That’s why I dont really care. And probably because the only way to test this is to run a 20 billion fund wito 0.2% management fee to cover R&D and nobody gives u 20 billion to test it out unless you were in a hedge fun. Which you only were if your brain is retarted and you like excel spreadsheets, which kind of leads you to a point where you dont think about this stuff. Talking about market inefficiency in HR. Thank god there is blockchain.