@stevehouf @itsandrewgao @__paleologo @quantymacro I think quants are best understood in the broader context of the market. The purpose of a market is to facilitate transactions and ascertain the correct price of an asset. Thus those who "make the market" provide liquidity. Buyers need not find sellers to transact immediately. @stevehouf @itsandrewgao @__paleologo @quantymacro To facilitate the liquidity, the makers must accept some risk, because they are offering to buy (sell) an asset when there isn't immediately another buyer (seller) to flip the asset. On the market making side, the quants involved here are preoccupied with understanding this risk

A persistent challenge in this industry is people who hold on to half-understood ideas. This is especially common with stakeholders who are generally intelligent but not quants, because they pick up important concepts but don't devote the time to rigorously understand them. Some examples are famous: "we don't want to do MVO for portfolio construction, because it's impossible to estimate the covariance matrix, and you'll end up with portfolios that don't make sense." They'll take these ideas and hold on to them, believing they've cracked the case.

@macrocephalopod "For each of the following questions, describe your answer as rigorously as you can.

Question 1: I tell you that I am regressing Y on X. Describe what I probably mean as rigorously and completely as you can, without sacrificing generality. @macrocephalopod "Question 2: I have a univariate linear model of X and Y, both n x 1 column vectors. Describe what happens when I try to estimate this model with an exact copy of X added as a second exogenous variable, and why, as rigorously and completely as you can."

As a reminder, this is what momentum has looked like YTD. Nothing meaningful happened today, yesterday, or in the past week. All the commentary you're reading about momentum is worse than wrong - it's not falsifiable. You can manage your portfolio successfully by ignoring it. - All the sell-side research you're getting about momentum is written by people who couldn't implement the momentum factor even if they were given an exact spec.

- The people at your firm talking about momentum are looking for a narrative and probably lack skepticism.

Very good thread. To extend this: signals that have positive correlation with future returns are still useful even if they don't surmount transaction costs on their own, because they lower volatility and costs when combined. I will show this via simulation.

https://twitter.com/macrocephalopod/status/1806436278067470524

Here's our setup, mathematically. We'll take normally distributed returns for convenience, n signals, t time
periods, c average cost (%), rho average corr between each signal and true return. This generates sample true returns and signals matching our target correlation.

Maiden Century is a good platform, but there are more clever ways to construct signals from typical sources of alternative data. Here is an example. Let's say you have some agnostic alt dataset that looks like this plot. How would you build an expected return signal from this?

https://twitter.com/0xFaust12/status/1801030853642055741

Further suppose that the dataset is ticker-tagged for 200 equities which respond to the dataset in roughly similar ways. Maybe the economic basis of the data is shared by all the equities, or maybe investor trading behavior on the data unites them.

@JaredKubin I like this. Some remarks.

1. First and foremost you need a team capable of high quality scientific research. They need to be past trivial mistakes on simple things. They need to deeply understand their models, estimation techniques, and priors. E.g. they can't just run sklearn. @JaredKubin 2. Assuming you have that team, the next thing that matters most in my opinion is operational stability. If your trading system and services are constantly breaking in production, if you don't have sufficient test coverage, if your code is spaghetti etc, that's a liability.

🧵A thread on how I interview and hire quants. Thought I'd take a break from technical content/paper reviews to discuss this a bit. Finding and developing talent is something I think about all the time. I think it's extraordinarily important - just as important as alpha and risk. I'll start provocatively: I think most quant hiring is broken. I've interviewed a couple hundred people over the past decade, I've sat for interviews with ~20 firms on the other side, and we as an industry do not approach hiring as critically as we approach trading.

This paper is unserious for any real practitioner, and I will show why with citations.🧵Along the way, hopefully this will be an example of how to quickly scan a new "strategy" or "alpha" paper for credibility.

For posterity, the paper is here: arxiv.org/pdf/2306.13661

https://twitter.com/iblanco_finance/status/1796550215056461982

Right off the bat, there is no mention of the specific universe under consideration in the paper. So the results are not only not reproducible, but it shows the authors haven't critically considered survivorship bias or why any given securities should be included.

A🧵on pricing data. I'm currently working on a series of posts studying different approaches to portfolio optimization. Since this is extracurricular, I won't use my firm's equipment or data (so: no bloomberg or factset). But "high quality" alternatives display trivial errors. I started off with IEX, and implement a full API client for a variety of endpoints. Let's look at the price and volume endpoint, and GOOGL in particular, since it's one of the most well known and liquid equities in the world. Surely it should be easy to maintain accurate data?

@__paleologo Okay, I read it. I'll summarize it and provide some commentary. In brief I think it's a useful and credible paper, with specific empirical results, and I'd explore it for further research. But it's not groundbreaking. @__paleologo So they start with the standard expected price impact we all know and love. Linear in vol, polynomial in participation rate. Just stage setting.

While I reflect on hiring, I want to discuss why so many firms prefer to hire PhDs to do quant research. 🧵

Quant research broadly falls into three categories: new signals, new risk models and more efficient techniques for optimization/execution. The reason quants are engaged in "research" is because they attempt to push the state of the art forward in a novel way. In so doing, hopefully they identify an approach that is uncommoditized and commercially valuable. But doing real research is extraordinarily difficult.