I Came, I Saw, I Placed a Bet

When I started reading trading books, the one concept that was the hardest for me to grasp was edge. I understood hedging; I had optimized enough theoretical portfolios by then. But edge? What is that, even?

I used to think edge was what you had when you knew economics, understood quantitative stuff like Black Scholes and what have you, had access to a lot of charts, understood all there was to know about technical setups, had access to great research, newsletters and good news sources, and had gained enough experience. I thought that if you had all these things, you would be unstoppable; that having these things was the only way to make money. You combined them to find the perfect trade. Anything else and you were likely to lose money because you would have only a partial view. Like Thanos, I went about collecting them, starting with a financial engineering degree to learn all the quant stuff. I didn’t go to quant school to be a quant, I went to quant school to be a good investor/trader.

In most games in life, there is one idea of what it means to be good at it. Say you are good at soccer, you know you are good when you have impressive stats for your position. Striker? most goals scored or shot conversion ratio. Goalie? Most goals saved or goals prevented ratio. Defender? Most successful tackles. It’s the same in many areas in life. In school you have grades, at work you have KPI and promotions and bonuses. My naive conception of edge led me to believe that if got all those things, you would always pick the winners and you would make a lot of money. I thought a loss meant you had failed. I didn’t know that it was even possible to do things right and still lose on a trade. Trading is different because success isn’t just about PnL but how you go about getting PnL.

Even when I started trading, and had read enough books to know that winning 55% of your trades was quite good, and that it mattered more that you made more when you won than what you lost, I still didn’t quite get what edge was. I started to think edge was a system. You had to know some fundamental analysis, and some technical analysis and then combine that with sentiment, psychology and risk management. At that point, I started to take what I now call academic bets. I would try and tick many boxes before I took a trade.

The problem I had is that I almost ended up never making a trade because of conflicting signals. When I went to do my fundamental analysis, and read all I could, no one seemed to agree on anything. Instead of getting clarity, I became more confused. When I looked at technicals, I got sucked in by the price action which always seemed to beckon me to do one thing, and then once I had done it, do the complete opposite. Sentiment wasn’t a hard one to figure out, but taking action based on it was damn near impossible. I struggled with the psychological aspects too, but my saving grace was risk management—I don’t do many dumb things when it comes to managing my risk. That was my edge, but I didn’t know it at the time.

The flaw in my second conception of edge is that traders are humans and are therefore irrational. Edge comes from the balance of our strengths and our weaknesses set against whatever it is we are trying to accomplish. What’s more, we all see things differently. I used to think rationality meant you instantly saw things as they are, but it’s not. To be rational, you have to take in the data as they are distorted by your own biases and subjective view, then you think your way through a lot of the irrational stuff ,and then you think about how you thought your way through the irrational stuff. And the goal isn’t to be perfectly rational; it’s to see clearly enough to make good decisions. This is what it means to do analysis in trading. Some academics call this bounded rationality, and it is the difference between trading and gambling. There is a threshold of how well enough you have to see things for you to make a good decision, and if you are below it you are simply gambling, regardless of how good your risk management is.

Some traders see clearly through macroeconomic analysis, so they analyze global macro. Some see clearly through charts, so they focus on price action. Others see clearly though sentiment so they look for signs of emotional highs and lows. Developing edge is about developing your own way of seeing things well enough to make good trading decisions.

The hard part of writing is coming up with good ideas. It’s the creativity that makes the writer, the rest can be taught. Creativity is about how we see things. Good writers and thinkers see more than others and differently than others. It is kind of like how Sherlock Holmes once told Watson “…you see yet you do not observe.” To see something you have to be aware of its existence. And there are things you can only see when you write. Whatever your analysis technique, writing will help you be better at it.

Edge is also about the ability to observe. We all see the same things in the market; same charts, same data, but we observe differently. This is why it’s beneficial to buy subscriptions to finance newsletters—not to blindly take the same trades as the author (although some do that), but to find out what the author has observed. You can then use this to see who else shares your view (i.e. is it contrarian or obvious), counter your own views, or find new places to look. What you want in a newsletter author is someone who has their own system nailed down, and it has a proven successful track record, then you want to know what they are thinking about the market.

Even though this is a quant newsletter, if there is one thing I don’t want it to be, is just another number crunching blog. Quants are notorious for overcomplicating things with their models, but complicated models don’t inherently have edge over simple ones. What quantitative models can offer is another view of the markets besides the fundamentals, technicals and sentiment stuff, but quantitative analysis is not inherently superior to rest.

In fact, the success of a quantitative analysis relies greatly on the quant. They say numbers don’t lie, but in finance they do. Think about the numbers we use. Economic data is based on estimates done by agencies like the BLS. First of all, there is the methodology used to come up with the figures; at a minimum, the methodology should give an unbiased estimate and the sample used should be representative of the population. Best case scenario the methodology gives a very good estimate of how things really are, but there is no way to ever know how good or bad it did in reality, ever. But even with a really good methodology, there are a lot of steps involved and a lot of work done before any final figure is presented to the public. Anything could go wrong and distort the results. The great thing about trading is that for most investors it is enough to take the published numbers as gospel.

These problems don’t exist for something like earnings because thanks to technology, it is possible to calculate to a high level of accuracy exactly how much a company spends and makes. In fact, the law requires that companies don’t misrepresent the data, so they kind of have to be accurate. The reality is of course different—companies cook the books all the time, or legally massage the numbers. Then you have some analyst somewhere doing DCF on those those books and placing bets in the market. Yikes.

My point is, in finance number’s lie all the time. Even if they don’t, there are people who either knowingly or unknowingly use the numbers to lie. Lie is a strong word; let me say they confabulate—they look at the numbers and create stories that explain them. So not only are the technical skills of a quant important, but also their character.

I also don’t think quantitative trading is in any way superior. The premise of quantitative trading is that machines have no emotions, so in theory, they should be better than you and me at making trades. They make buy and sell decisions on programmed logic, analyze the data objectively, size their bets appropriately, and exit at the right time. One problem: machines are written by humans. The best an algo can do for you is obey you perfectly, but you still have to write it and imbue it with all your biases. If you have a system or strategy that can be codified, then maybe bots can help you make the most of it, but AI can’t see for us. Maybe if we start adding AI to these algo they will start giving us useful feedback about the flaws in our trading strategies. *Beep. Boop.*: “You’re always late to go adopt a new narrative, do you want me to change my logic to enter trades at the first sign that the narrative has shifted.”

Another great thing about using AI to build trading bots will have to do with probing. You give AI some money and an initial premise, then you ask it to make very small bets in the market as it tries out different ideas. This process is useful even when you are an experienced trader and you know your edge. Sometimes you have conflicting ideas about what the market will do. Instead of sitting it out and missing the bus entirely, you can make some small bets on one idea and see how things develop, then either build on it, or take the other view depending on what happens next.

I’ll tell you one thing experienced traders have that newbies and bots don’t: the ability to diagnose why a trade didn’t work. It’s a different kind of sight to have in trading because it helps you calibrate quickly. That’s is why it’s a good idea to trade frequently to develop your edge. Any losses are just R&D. Developing your edge should be your first priority because then you can figure out how to make the most of it. That said, once you know your edge, you want to trade less frequently, that is, only when you can see a decent amount of money to be made. Warren Buffett famously said that if you had a punch card with only 20 punches to use in your lifetime for every financial decision you made, you would be very rich, and you probably wouldn’t use them all. Edge is you lens, but you want to spend more time looking before you leap.

It’s often said of good traders that they are robotic: they do the same things, the same way, day in, day out. Not robots, robotic; we humans don’t do the same thing the same way twice. But because of how unhuman it is to be so consistently repetitive, we call it robotic. Whatever your style of analyzing markets, you need to be robotic about it to help you calibrate and refine things.

Presumably, those who think trading bots are better think that not having emotions leads to better risk management. It’s why Wallstreet once tried to hire more psychopaths than neurotypicals. I don’t think this is true. Even risk management requires an analytical process that starts with the irrational and moves towards the rational. You need to think about your bet size carefully too, and there is no one right mathematically tractable answer. But that is a conversation for another day.