← Back to list
Colossus (Invest Like the Best / Business Breakdowns)Podcast1 Aug 2017Source: traffic.libsyn.comHost: Patrick O'Shaughnessy

Leigh Drogen - Sink or Swim--How to Combine Quant and Traditional Asset Management Techniques - Invest Like the Best, EP.48]

In plain words

This interview argues that traditional asset managers must rebuild their investment process from scratch, not just add quant tools on top. The key idea: portfolio managers should shift from being a 'quarterback' who makes all decisions to an 'offensive coordinator' who manages the process, letting analysts make stock picks and systems handle risk and portfolio construction. The best ideas are often controversial internally, not consensus. No specific stocks are mentioned, but the focus is on using 'force rankings' and tracking analyst predictions vs. market consensus to find opportunities, plus using quant models (like factor analysis to see if a stock's return is from the market or skill) to build better portfolios.

AI SummaryAI-generated · may contain errors · verify against the original

Leigh Drogen discussed on the program how traditional asset management firms are responding to the rise of quantitative investing. The core argument is that traditional managers must fundamentally change their mindset, integrating quantitative and traditional techniques into a hybrid structure rathe

~11 min full read · 8 sections
Deep Analysis

This Issue at a Glance

Leigh Drogen (Founder of Estimize, former momentum investor) argues in this issue that traditional asset management firms must "tear down and rebuild" rather than "layer on quant" in order to survive competition from pure quantitative funds. Core thesis: Traditional PMs should transition from being "quarterbacks" (deciding everything) to "offensive coordinators" (orchestrating the process), delegating subjective judgment down to the analyst level, while leaving portfolio construction, risk management, and timing decisions to systems and quantitative models.


1. Fatal Flaw of the Traditional Process: Political Gaming Replaces Measurability

Leigh Drogen argues that the traditional analyst-to-PM decision-making process in funds has a systemic flaw: a good analyst is not necessarily a good "politician," and PM intervention often stifles truly differentiated ideas.

  • In the traditional process, analysts cover 40-50 stocks, write 10-page research reports, and then "sell" them to the PM. The PM decides whether to adopt the idea, how to build the position, and when to exit. This process erroneously ties "analytical ability" to "political ability"—a good analyst may be overlooked due to poor salesmanship, while the PM simultaneously shoulders too many functions, including stock selection, timing, risk management, and position sizing.
  • The core issue is that consensus and alpha are inherently contradictory. Drogen cites research from First Round Capital and Union Square Ventures, noting that the best investments are often those that trigger "polarized" reactions in internal meetings—some think it's "great," while others think "you're crazy." Traditional PMs tend to adopt consensus ideas that "sound reasonable," which are precisely the ones least likely to generate alpha.
  • Consequence: The entire system cannot answer "when we did well, when we did poorly, who to listen to, and who not to listen to."

> "A good analyst isn't necessarily a good politician within a firm. But these two things have been unfortunately combined."


II. The Starting Point of Reconstruction: From "Core Beliefs" to "Structured Unstructured"

Drogen argues that the first step in reconstruction is to clarify "core beliefs"—the fund's investment philosophy, circle of competence, and time horizon—and then transform unstructured information into trackable structured data through "forced ranking."

  • Core beliefs are like a "constitution": The fund must clearly answer—which market cap range do we track? Which sectors? Do we believe in value or momentum? What is the investment horizon? Drogen points out that many funds cannot even clearly articulate their own investment universe and strategy ("we do everything"), leading to style drift and ineffective risk management.
  • Step two: Identify differentiated views. Analysts must periodically (three times per quarter) submit forward-looking forecasts for key metrics such as EPS, revenue, and EBITDA to a central database, along with target prices. PMs then review the largest deltas between "analyst forecasts vs. market consensus" through the system, rather than waiting for analysts to actively "pitch."
  • Structured unstructured: Drogen suggests that analysts score subjective concepts like "corporate governance" or "probability of being acquired" on a scale of 1 to 10, and track the correlation between these scores and stock price performance over the long term. Key principle: Simple systems are better than complex ones. Forced ranking (1 to X) is more effective than precise price targets, as it avoids the risk of "false precision."

> "Simple systems are better than complex ones. It's the beauty and the elegance of a force ranking, one through X, versus price targets."


3. Division of Quantitative Roles and the New Positioning of PMs

Drogen breaks down the four types of roles within a quantitative team and emphasizes that PMs should transition from "stock pickers" to "process coordinators"—their core value lies in integrating multi-source information and challenging analysts' assumptions, rather than replacing systematic decision-making.

  • Four types of quantitative roles:
  • Data Engineer (cheapest): Responsible for channeling data into analysts' forward-looking decision-making processes
  • Data Analyst (hardest to find): An interdisciplinary talent who understands both fundamental analysis and technology, helping analysts access and interpret data
  • Pure Quantitative Analyst: Studies factor models and validates the correlation between data and EPS/revenue
  • Quantitative Engineer (most expensive and scarce): Builds factor models and integrates third-party data
  • The PM's new role: PMs should not add alpha at the "which stock to pick" level, but rather at the "how to make the process run better" level. Specific responsibilities include:
  • Reviewing the stocks with the largest delta among analysts' top-ranked picks and challenging their assumptions ("Have you considered these three things?")
  • Assigning different confidence intervals and capital weights to different analysts based on historical performance
  • Adjusting exposure to specific factors based on macro judgment, on top of the beta-neutral portfolio suggested by the system
  • Measurability: Drogen suggests constructing an "analyst virtual portfolio" (equally weighted top 5 long/short recommendations from each analyst) and then measuring the deviation of the PM's actual portfolio from the virtual one—if the PM's deviation leads to worse returns, they should be held accountable.

> “PM shouldn't be trying to add alpha in the 'what stock should we pick' realm. They should be adding alpha in the 'how do I run a better process' realm.”


4. Portfolio Construction: Machine-Led, Human-Adjusted

Drogen argues that portfolio optimization and risk factor exposure management should primarily be handled by the system, while the PM's macro judgment can serve as a "tuning knob," but must be measured and held accountable.

  • The system should automatically neutralize beta to ensure that portfolio returns come primarily from alpha rather than beta. Drogen uses Microsoft and Tesla as examples: Microsoft's returns over the past six months have been almost entirely beta (replicable via a smart beta ETF), whereas Tesla's returns have been predominantly alpha (with low correlation to factors such as market, momentum, and value).
  • What the PM can do: Based on the system's foundation, adjust exposure to specific factors according to their assessment of market conditions (e.g., "the momentum factor has been excessively suppressed and is about to rebound"). However, the alpha generated by the PM through such adjustments must be measured annually, and underperformers are replaced.
  • Key exception: The PM may make "concentrated bets" on stocks that are strongly recommended by analysts and also receive a "green light" from the factor model—this is where the "human-machine combination" can potentially generate returns beyond pure quantitative approaches.

V. Lessons from War Theory: Measurement Over Outcome

Drogen directly maps lessons from military history onto investment management: decision-makers should not be promoted based on a "lucky victory," but rather on the soundness of their decision-making process.

  • A common phenomenon in the U.S. military is "performance chasing": a general is promoted after an extremely improbable victory (e.g., flipping heads five times in a row), yet may make fatal errors in the next decision. This is entirely analogous to portfolio managers in investing who are rewarded for excess returns driven by luck, despite lacking genuine competence.
  • Drogen cites the "Way of Life" app he uses as an example: after tracking the correlation between various variables and "happiness" for over two years, he found that the simplest variable — sleep quality — had the highest correlation. This underscores the power of "simple systems plus continuous measurement."

> "Measurement versus outcome is so important because you might flip heads five times in a row, but the next couple of times it'll be completely random."


Mentioned Positions

This section does not involve discussion of specific companies or investable positions.


Judgments Worth Remembering

1. Drogen: Traditional PMs must transition from "quarterback" to "offensive coordinator" — relinquishing full control over stock selection, timing, and risk management, and instead focusing on coordinating processes, challenging analyst assumptions, and adjusting factor exposures. Rationale: PMs juggle too many variables to excel in all areas; systems are far superior to humans in portfolio optimization and risk management.

2. Drogen: Consensus and alpha are inherently contradictory; the best investments are often those that provoke "polarizing" reactions — citing research from First Round Capital and Union Square Ventures: when everyone thinks something "makes sense," it is usually not a good investment. Rationale: Traditional PMs tend to adopt consensus ideas that "sound reasonable," precisely those least likely to generate alpha.

3. Drogen: Simple systems outperform complex ones; forced ranking is more effective than precise price targets — avoiding the risk of "false precision" and being easier for traditional analysts to accept. Rationale: The financial industry tends to believe "complex = better," but this is a marketing mindset, not an investment mindset; only a very few funds (e.g., Renaissance) are qualified to implement complex systems.

4. Drogen: Quant teams need four types of roles, with "data analyst" being the hardest to find — data engineers (cheapest), data analysts (interdisciplinary, hardest to find), pure quants, and quant engineers (most expensive and scarce). Rationale: Data analysts need to simultaneously understand fundamental drivers, technical skills, and the ability to communicate with the quant team; such talent is in short supply.

5. Drogen: Analysts should be highly specialized, and the best forecasters cover 10-50 companies per quarter — Estimize data shows that analysts covering fewer than 10 or more than 50 companies per quarter have lower accuracy. Rationale: Covering 10-50 indicates "serious engagement," while more than 50 suggests "over-diversification and lack of depth"; analysts should come from within the industry (e.g., tech analysts who have worked in Silicon Valley).

6. Drogen: A PM's alpha should be measured through a "virtual analyst portfolio" — constructing an equal-weight virtual portfolio of the analysts' top 5 long/short recommendations, then measuring the deviation of the PM's actual portfolio from the virtual one. Rationale: If the PM's deviation leads to worse returns, they should be held accountable; if analysts' forecasts are consistently inaccurate, they should also be replaced — the current system cannot distinguish who is creating value.

7. Drogen: The principle of "measurement over outcome" in war theory applies directly to investment management — decision-makers should not be promoted based on a single lucky victory (e.g., a general promoted after an improbable victory, or a PM achieving excess returns through luck). Rationale: The "performance-chasing" phenomenon in the U.S. military leads to unqualified generals being promoted; the same need to distinguish luck from skill applies in investing.

8. Drogen: Only 20-30% of traditional funds will successfully complete the "quant + traditional" integration transformation — most funds attempting to "layer on quant" will fail, while newly established funds building a hybrid process from scratch will succeed. Rationale: Estimize's mission is to serve as the "central dashboard" for these new funds, helping them run a structured investment process.