Quick backstory: around a month and a half ago, I was partaking in my daily twitter doomscroll, and chanced upon a fascinating tweet from Steve Hou. I usually like Steve’s posts anyways, because he offers cool insights into the quant industry and is a pretty smart guy overall. But this tweet was particularly insightful - to the point that I bookmarked it to study in depth later.

It’s about the concept of “Alpha” and how even experienced practitioners can misuse this term. As someone who’s still learning about the field, it’s offered me a fresh perspective on a concept that is considered seemingly fundamental and basic. Around a month ago, I spent a few hours breaking each part of it down (within my own notes) and was rather happy with what I had learned. Then a thought struck me that I might as well share it publicly to potentially save someone else that time. So here I am!

Why post about this a month later you might ask? Well, because I forgot. Here’s the original tweet. I’d encourage reading through it first to get the author’s thoughts directly. If you end up skipping it, know that you are placing a shocking amount of confidence in my paraphrasing skills. Anyways, I’ll now try to break this down to the best of my ability - piece-wise, with the power of God and Claude on my side.

Takeaway 1: It Is Difficult To Concretely Define “Alpha”⌗

“Alpha” is an overused term without a concrete, rigorous definition. This is similar to other popular financial terms like “bull market,” “bear market,” or “correction,” which are widely used but lack precise, universally accepted definitions.

This gives us a nice introduction to the topic of discussion. We all know that “alpha” is what everyone in this industry chases…but what is it really? Everyone talks about “alpha” without ever specifically saying what it means. At this point, you might be tempted to scream “ur stupid, i know alpha. alpha is the trading edge we aim to find that no one else knows about!!!”

And you would be right. I’m dumb as a rock. Also, that quite possibly might be what alpha is. But I implore you to keep reading, you may find yourself surprised!

Side note: I always assumed that bull market, bear market and corrections all had well defined meanings. Guess not.

Takeaway 2: Is “Alpha” Really the Opportunity for Transient Risk-Less Arbitrage?⌗

If we define alpha strictly as transient genuinely risk-less arbitrage opportunities, it would be an extremely rare phenomenon. In reality, much of what we call alpha is actually just beta risk factors that are becoming better understood over time.

Following directly from the previous takeaway - is alpha really this risk-free edge that no one else knows about? If that’s the case, most of what people call “alpha”, is actually just betas! This is well illustrated in his analogy:

(A) Knowing about secret city street and alleyway short cuts to avoid traffic is alpha. However, it is likely small capacity and will quickly be arb’ed away as secret spreads. Interesting. What exactly makes this alpha?

  1. This knowledge provides excess returns (faster travel times) without taking on additional systematic risk. Like financial alpha, it offers outperformance that can’t be explained by common risk factors.

  2. It’s based on specialized local knowledge or insight that goes beyond the common risk factors reflected in general traffic patterns (such as rush hour congestion, weather conditions, or major events). This knowledge provides an advantage over typical strategies like using Google Maps/Waze suggestions, sticking to well-known routes, or adjusting departure times to avoid peak hours.

  3. As more people discover these shortcuts, the advantage diminishes, mirroring how alpha opportunities in financial markets tend to disappear once widely known.

(B) Is driving on the LA highway at 4 am to avoid traffic also alpha? Maybe, but it depends on whether you are better at consistently waking up at 3 am to go to work - which makes it feel more like a beta risk factor. So, why does this feel like beta?

  1. Just as beta in finance represents market-wide risk that affects all participants to varying degrees, the “3 AM factor” is a widespread option available to all but with known negative impacts (potential health issues from lack of sleep, social life disruptions, and increased stress, etc.)

  2. The “excess returns” (faster commute, higher productivity) from this strategy are essentially compensation for bearing this widely available but often undesirable risk, much like how higher beta stocks offer higher potential returns as compensation for their increased volatility.

Interesting! However, does that really make it beta? Not everyone can do it right? That leads us to the third takeaway.

BUT before I move on to that, I want to go on a short tangent on why exposure to lesser-Known risk factors can lead to excess returns. If you’re experienced this is probably already obvious to you; feel free to skip. When I read the tweet though, I needed this to be spelled out to me. So in case there’s other strugglers out there, this is for you.

TANGENT START

The basic reasoning goes like this - excess returns are more likely when risks are not fully understood or priced by the market, then once a risk becomes widely recognized, the potential for excess returns often diminishes quickly.

That’s the theory, but it is better explained through some historical examples:

Case 1: The Downfall of the Small Value Factor⌗

In the early 1990s, researchers identified that two groups of stocks tended to outperform the market over long periods:

  1. Small-cap stocks (companies with lower market capitalization)
  2. Value stocks (companies with low price relative to their book value)

When first discovered, the outperformance of small value stocks was seen by many as a market inefficiency or a source of alpha. Investors who tilted their portfolios towards these stocks appeared to be generating excess returns through skill or insight.

Further research revealed that this outperformance was actually compensation for taking on specific types of risk:

  1. Small companies are generally riskier and more vulnerable to economic shocks.
  2. Value companies often face financial distress or other challenges.

As this understanding developed, these effects were reconceptualized as systematic risk factors and incorporated into asset pricing models like the Fama-French Three-Factor Model. Once widely recognized, many investors started targeting these factors explicitly, leading to reduced excess returns.

Today, exposure to small value stocks is generally seen as a form of beta - a known risk factor that investors can choose to take on or avoid. It’s no longer considered alpha or a sign of special skill to invest in these stocks.

Moral of the story: Factors that were once considered “alpha” but became widely known turn into a recognized “beta” factor.

Case 2: The Low Volatility Anomaly⌗

Traditional finance theory (like the Capital Asset Pricing Model) suggests that higher risk (measured by volatility) should be rewarded with higher returns. However, researchers discovered that low-volatility stocks have historically outperformed high-volatility stocks over long periods, contrary to what theory predicted.

Initially, investors who tilted their portfolios towards low-volatility stocks and achieved higher risk-adjusted returns appeared to be generating alpha (outperformance due to skill). Turns out, this outperformance was actually due to exposure to a specific risk factor (low volatility) that wasn’t well understood or widely recognized at the time.

The low volatility factor was “lesser-known” in the sense that it wasn’t included in traditional asset pricing models and wasn’t widely appreciated by market participants.

By investing in low-volatility stocks, investors were exposing themselves to a specific type of risk - the risk that this anomaly might disappear or reverse, or that low-volatility stocks might underperform during certain market conditions. This exposure to a lesser-known risk factor led to excess returns, as the market hadn’t fully priced in the benefits of low volatility.

As this factor became more widely recognized and studied, it has transitioned from being seen as a source of alpha to being understood as a systematic risk factor (a form of beta).

Moral of the story: Researchers always ruin the fun.

TANGENT END

Takeaway 3: The Line Between Alpha and Beta Is Blurry⌗

Is it capturing alpha if you have to set up an elaborate and costly trading system, acquire a ton of quantitative education training and/or expensive human capital, and devote nearly all of your time to monitoring the system or markets every minute in order to capture outsized profits in a highly competitive arena where others are trying to do the same thing? Or is it alpha to try to find effortless “tricks” that only work bc few others know of them?

No paraphrasing necessary. I think this makes our conundrum clear - what really differentiates alpha from beta? Let’s discuss why each approach might or might not be alpha:

Exhibit A: A resource-intensive method involving complex systems, extensive specialized education and constant market monitoring to achieve outsized profits in a highly competitive environment.

Why it might be alpha:

  • Creates barriers to entry, sustaining advantage over time
  • Allows for processing of complex information others can’t handle
  • Can identify and exploit numerous small inefficiencies at scale
  • Demonstrates skill in building and managing sophisticated systems

Why it might not be alpha:

  • High costs may offset gains, reducing true outperformance
  • If widely adopted, it becomes a “known factor” rather than a unique edge
  • Relies more on resource advantage than pure skill or insight
  • May be replicable by others with similar resources

Exhibit B: A little-known strategy that provides us an advantage because it’s not widely recognized.

Why it might be alpha:

  • Represents pure insight and market inefficiency
  • Highly efficient in terms of return per unit of effort
  • Often based on unique understanding or perspective
  • Can provide significant advantage before being widely recognized

Why it might not be alpha:

  • May not lead to as many returns as you’d expect, since it’s usually short-lived and markets quickly adapt once discovered
  • Could be more luck than repeatable skill
  • Might be exposed to unknown risks not accounted for
  • Usually limited in capacity and scale

Some of these things may just make the edge a “worse alpha” rather than “not alpha”. Where do you draw the line? There appears to be no concrete answer. An unsatisfying ending, but we can never really have all the answers can we? To prevent closing out on that somber note, there’s one last question we need to answer-

Does It All Even Matter?⌗

From fdf in Peel, Pith and Seeds:

“Alpha” as colloquially used is underspecified. What most reasonable people can agree on is that it’s both rare and fleeting. I will offer three common perspectives:

  1. “Today’s risk factor is yesterday’s alpha factor”. In this setting, alpha is a risk premium. It has a legitimate return which is outsized because it’s esoteric and relatively unknown. Eventually that risk premium will become a part of the regular old market mechanics and cease to be as profitable, even if it persists as a factor. In this setting, mid-frequency CTAs that specialize in trend and carry may have alpha if they are able to consistently provide a positive market neutral return in excess of the momentum factor.
  2. Alpha refers only to true market inefficiencies which must be kept secret in order to work, and which are largely capacity constrained. It decays rapidly throughout the crowding lifecycle of its strategies. These proponents would argue Jane Street’s Indian options are emblematic of “true” alpha, and in fact you don’t really find it outside the intraday horizon.
  3. Alpha is superlative operational efficiency, wherein you gather together enough people who are just skilled enough and just diverse enough to get positive market neutral returns in (apparently) any regime. In this regard, Citadel’s and Millennium’s fundamental equities groups that operate on quarter+ time horizons have alpha.

In my view this debate is largely unimportant, in the sense of Wittgenstein. If you demonstrate legitimate skill at mixing together risk premia, keep at that. If you specialize in identifying inefficient parts of market microstructure you can harvest, do that. Find your flame and give it more kindling. Above all else, understand and attribute the risks you take, so you can be honest with yourself and your investors.

Suppose you and your friend are sitting on a beach and bet with one another about who could reach an island first. A group of gamblers comes around and decides to place wagers. You agree to terms, and at the last moment find a raft with a mediocre but serviceable sail. Your friend on the other hand swims. With a favorable tailwind at your back, you handily win.

To what do you attribute your success? Certainly not your swimming prowess. You won with your sail but you cannot control the winds, and one day that tailwind will become a headwind. You can cleave the debate about what alpha “is”, because this is what really matters. If you understand your risk you become the master of your fate: you know if winds favored your sail, or if you were actually a good swimmer.

I don’t think there’s much for me to add on here. I’m a LARP and that’s a bonafide expert. Even if the debate remains largely unresolved, the fact that of the matter is that you have an edge. It doesn’t matter what exactly that edge is - what’s important is that it works! A perfect conclusion, if I can say so myself.

See you in the next one :)