Where's the edge?
Thoughts on where money is made based on my own career and tales of others
Introduction
Where’s the edge man? That’s a common question. How can I find alpha? Many of us will spend years searching for the answer to this question, and I don’t claim to be a prophet myself on how it’s done. There’s no clear and absolute answer to this, not one bit, but there is useful answers to this - and many of them. There’s also many shit answers to this question that won’t get you very far, or if you do go far it’ll be in the wrong direction.
Index
Introduction
The Sad Reality (everyone runs the same)
The Data (Budget) Edge
Theory of Edge (why you get paid)
Being Around The Money
Inherited Alpha (dynamic of PMs etc)
Forcing Ideas (how to get inspiration)
Stealing Ideas
Developing Theories
Expert Alpha (becoming a fucking nerd)
OpEx Investment Alpha (infra heavy edge)
Right Place Right Time
Spotting Who Has Edge (hiring edge)
Accumulated Improvement (market making)
Iterating Faster
The Sad Reality
Equities. A developed market. Right? Yes, and no. In many ways, there are still opportunities to be found, and plenty of people make a killing in the market, but there are so many people who run the same strategies.
There are many many times more hedge funds than there are strategies and even strategies you may think are novel are not that novel. Only in rare cases of liquidity constraint and where the strategy is relatively HFT (i.e. alphas are deterministic so you can see who else is trading) you get scenarios where you are the only one trade.
We live in a world where we compete with others. The whole concept of novel ideas is a falsehood - most ideas you will have will be taken by others. The only question is of how many others and how good that idea is. A very good idea, that only a few others have discovered is an ideal alpha.
Alpha decay in many cases is just others finding out over time and as more data becomes available to reveal that alphas existence.
The Data (Budget) Edge
Many firms spend tens of millions of dollars on their data budget. QRT spends about 100 million dollars a year on their data. This is everything from your credit card purchases to satellite imagery of the factory delivery docks.
In fact, the most expensive datasets out there are not what you expect. It’s custom trades from execution desks at banks which have been “anonymized”. This of course has led to various issues with the SEC, and those sources have been retracted on various occasions because of the clear breach of fiduciary duty.
There are many other examples of grey market data sources being retracted and careful analysis of whether the data is legal is often required.
My previous articles on genetic algorithms for automated alpha discovery apply quite well to this sort of edge because you have data which isn’t going to be very noisy when it comes to correlating it to returns, but it’s a lot of work to hire researchers to sit and investigate all the alphas that can be made from the dataset.
I actually have friends who work as analysts at funds where what they do is take the trial for the dataset and perform a cost benefit analysis on whether to acquire the dataset. I know others whose role is to work towards this objective and develop the alphas for those datasets (before and after purchase - for estimating the upside / correlation to existing alphas of purchasing and also for then monetizing this investment). It’s a whole industry and one that requires deep pockets.
You can, of course, collect this data yourself, and many firms do this as their edge. This is something I’ll get into with the expert alpha section where people becomes extremely specialized in certain areas.
Theory of Edge - Why should YOU get paid
There’s a few reasons you make money in the market. The simplest of these is that you are willing to do work that others can’t because of some constraint. This is really where you’ll find most of the worthwhile alpha.
You can also be lucky or early or perhaps you are simply better than everyone else, but let’s be real these aren’t proper strategies. Luck can be written off from the start as an approach for fools. Hit the casino if you want that. Being early can be a strategy, but it requires you to actually be searching about - which most aren’t doing well enough. I’ll get back to this point. Finally, being better than everyone else. That sure sounds nice, but you aren’t.
You are not going to be better than everyone else in the market. Stop trying to be Djokovic because that’s a very hard game. The super competitive markets are usually dominated by teams of really great people - if you want to be Djokovic, then you need to have a firm that can finance your ambitions and an eye for talent. Otherwise, focus back on where others are constrained.
What counts as constrained? Here’s a couple examples:
Operational PITA
Capacity vs. Skill
Risk Premiums
Small Margins
Starting with operational pains in the ass, I’ll walk you through an example of a trade that I wanted to put on a couple years ago regarding the Turkish Lira. The trade was quite simple. You start in USDT or any high liquidity cryptocurrency, and send that to a Turkish exchange for crypto such as BTCTurk. Then you convert to Lira on that exchange, withdraw the Lira to an international bank account, convert to USD, send the USD back to Binance, convert back to your cryptocurrencies, and repeat. We estimated an unlevered return of 40% a year off this trade and even more if you can shave off time via “operational edges” (which typically amount to bribery to speed up transactions for some of the scummier players engaging in this trade across various countries - to clarify, we did not put this trade on). The edge was that transfers of Lira were incredibly hard, you needed to know Turkish to get the banks to even speak to you, prevent the government investigating you for money laundering (in an attempt to stop you transferring money). All of this was to exploit the 2% difference between local and international exchange rates. This gap expands into the tens of % when we look at Russian and Ukrainian currencies (which you are not even legally allowed to transfer and thus the traders doing this physically transport bills out of the country illegally).
Trading on shoddy foreign exchanges in general, even if it isn’t a geographical arbitrage of a currency generally fall under this category as well. If you are a local in South Korea, Brazil, Turkey, Nigeria, or India - and have a knack for arguing with government officials then perhaps this trade is for you. Not that I recommend or endorse it for obvious reasons.
Capacity vs. skill is one of the more valuable edges. What you are simply saying is that those who know don’t care and those who care don’t know. There is limited capacity in niche trades and this means that big firms (which have the skill to exploit them) pass on them. Institutional skillset + a retail approach to markets (which doesn’t require an institutional appetite in the tens of millions) is the key here. Imagine you are a professionally trained market maker who has worked at a top firm, you have the skillset to crush a shitcoin market as a maker - and it would probably pay a few million a year. In fact, that’s the entire premise of the small trader alpha series I write. How can I take uncompetitive niches that only really look like good money if I am working by myself or in a very small team. If you have a 10 person team at a prop trading firm then it is really hard to make a couple mil a year be worthwhile - or even pay the bills in some cases. If you are doing it by yourself, then that changes to being worth a lot of money. If you are smart and can implement your own strategies, you can dominate a niche easily.
Risk premiums are scenarios where you take a risk off someone’s hands for them, and get paid for it. As much as the academic literature likes to pretend that certain things are risk premiums and there’s an efficient market that means all premiums are created equal - they aren’t. Some premiums are a lot bigger than others for a lot less risk. The edge here is knowing when to trade and load up on risk, and when to back off.
You can mix edges as well. You can suck at market making and not make a dime during normal market conditions, but if you start market making during a meltdown, you usually make money regardless. This is partially a tech edge as you need to build the system to execute the alpha (market making), but it only makes money because of the meltdown. Tune properly and find the rare scenarios when the market breaks down. This is similar to the niche markets the big firms won’t market make, but instead it’s the times the big firms can’t market make because their risk is all full. Rotate around. When it’s altcoin season, add depth to altcoin books. When funding arbs are good, run that. When BTC is going crazy, quote wide and make money there. Have the tooling to run these strategies ready (which is the barrier which prevents everyone at the time), and keep the capital spare.
Small margins. This is where there is an alpha, but it’s so noisy that it will take decades to realize the edge. Momentum is a clear and profitable edge, but so many managers are benchmarked to the S&P500 and the edge takes decades to prove itself as superior so that edge sticks around and isn’t fully harvested. This isn’t an edge that will make you rich though - it’s a small margin afterall. So the trick here is to tell people about why it will work and they need to stick with it and then load up on AUM. This is the entire business of CTAs. That’s why you hear them on podcasts so damn often. Their strategy only barely beats the market and takes decades to prove itself so their business is in explaining to you to stick with it until that happens first and foremost.
Being Around The Money
A lot of people make their money by being around the money - there’s a lot of money here. I know entire firms that are built around one founder’s decision making skills as a discretionary trader and every single other person there works to provide execution, tooling, etc for him. Sure, you don’t make a 10mil bonus when your PNL cut hits, you don’t have one to begin with, but plenty of people do this and make 300k a year.
Look at Citadel. They have a core edge from being the biggest bully on the playground. There’s an economies of scale there. You don’t need to necessarily be the guy creating edge, you just optimize parameters or do a bit of machine learning on an existing feature set. These are people that can create edge in some ways (there’s some value provided by improving execution or ML, etc), but they’re not creating the primary source of edge, and are simply optimizing it.
Edge from new alphas is additive. Edge from better execution / ML is a multiplier. When you already have lots of edge from various alpha - it can still be valuable to have those multipliers. Even beyond this, plenty of people coast. They work at Citadel, do nothing but are charismatic enough to last a couple years. Get fired, move to a small shop, work there until they realize they didn’t come with the Citadel edge, then rinse and repeat at many shops. This is a lot more common than you think although usually at some point someone questions their lack of duration at any given firm or they eventually come up with (or more likely steal) some alpha for themselves.
For many developers, this is the primary plan - not the coasting part, but the being around the money part. Developers can be replaced. Great developers are hard to find, but they still need an alpha to implement. They can’t function on their own, there’s an inherent power imbalance. They still get paid a lot of money though, mostly in salary and bonuses, but not any mega upside unless they help found a firm. That is all to say that developers often implement the alpha of a PM and thus exist around the money. There are counter examples to this of course, some developers make latency edges, some steal alpha (and implement it easily themselves), some become great at finding alpha, some found shops as CTO - plenty of routes to put yourself in a position to *potentially* hit that 8 figure PNL split when the strategy moons one year.