Consider the following passages from two different book reviews. First is an AIMR Book Review from Martin S. Fridson:
According to Brett D. Fromson (“Wired into Wall Street,” Washington Post, December 1, 1991:H1): “Steinhardt Partners also wields considerable clout because of the fees it pays to brokers for executing its trades” — $30 million in 1990. Thus, it’s not unusual for a brokerage to give him first word of a new “buy” or “sell” recommendation, which enables him to move before the crowd…Steinhardt’s access to information has engendered jealousy on Wall Street, and some competitors nodded with disapproval when he told a Wall Street Journal reporter in 1986 that he relied on “fancy information” when trading. That quote caused the SEC to question Steinhardt about what he meant, and the agency apparently was satisfied with his explanation.
Most people think the “game” is how Steinhardt describes it. People think it is rigged. But is it rigged when you can see the price action every day? Trend followers and other types of systematic traders use price cues for decisions, not “fancy news.” Beyond that, was Steinhardt lucky?
From a review unrelated to Steinhardt comes this AIMR Book Review from Mark S. Rzepczynski. Consider:
Seemingly random events at the extremes, or just those events that are unanticipated, are often the key determinants of performance — good or bad — but we often place too little emphasis on these uncertain events. When we win, we believe our victory is skill, not the luck of the draw, and when we lose, underestimate randomness and confuse noise for meaning. To stay in the game, traders have to minimize their maximum loss through expecting the unknown and random market behavior. We are fooled by randomness because we suffer from behavioral biases that make us our own worst investment enemies. We go to great lengths to downplay the rare events and place great confidence in our own abilities, which allows us the illusion of control over market randomness. Some investors eliminate outliers for the very reason they are needed: They are low-probability events that can occur. Those rare events are often explained away with the arguments “this time is different.” Some investors undervalue the randomness associated with survivorship. The fact that a trader (or a firm) has survived for years does not automatically make that trader a master of the universe — perhaps just lucky.
The Practical Answer to the Luck vs. Skill Question
Rzepczynski’s review identifies the asymmetry at the heart of the luck/skill attribution problem: we attribute wins to skill and losses to luck. This is the self-serving bias applied to trading performance and it prevents the honest evaluation that would allow genuine skill assessment. Every successful trader can construct a narrative about why their approach produces returns. The narrative feels like skill. Whether it actually is skill requires a much larger sample and a more rigorous standard of evidence than most traders or their investors apply.
The Steinhardt passage presents the other face of the problem: information-based trading that may not be skill in the generalizable sense. Access to brokerage recommendations before they are published is not systematic edge. It is structural access that most participants do not have. Whether that access constitutes skill or a form of first-mover advantage that regulatory authorities should examine is the question the SEC was apparently asking. Price-based systematic trading sidesteps this entirely. The price is public. Every participant sees it simultaneously. The systematic trader’s edge, if it exists, comes from having better rules for responding to the public price, not from having access to non-public information.
The survivorship bias point in Rzepczynski’s passage is the most important one for evaluating any trading track record. The traders who survived ten years include both genuinely skilled practitioners and lucky ones whose positive variance happened to persist long enough to build an impressive track record. Distinguishing between them requires examining whether the performance persists across different market environments, whether it was produced using consistent methodology, and whether it is consistent with the returns that the stated approach should produce given its characteristics. A manager who produced great returns in a single favorable regime is not demonstrating the same kind of skill as one who produced consistent returns across trending, choppy, high-volatility, and low-volatility environments over decades.
Trend following’s documented multi-decade performance across multiple managers addresses this precisely. The correlation data showing that independent systematic managers captured similar returns during the same market periods, the audited performance spanning multiple market regimes, and the consistency of the approach with a well-documented structural market characteristic — that prices trend — provides the statistical framework for skill rather than luck attribution that individual track records cannot.
Frequently Asked Questions
How do you distinguish skill from luck in a trading track record?
By examining performance across multiple market environments rather than a single favorable regime, by comparing the performance to what the stated approach should theoretically produce, by looking at consistency across multiple independent practitioners using similar approaches, and by testing whether the approach produces returns based on a structural market characteristic that has theoretical justification. A manager who outperformed during one favorable period and has one year of data has provided almost no evidence of skill. A manager with 20 years of audited returns across diverse market conditions has provided substantial evidence.
What is survivorship bias and how does it affect the skill/luck question?
Survivorship bias is the tendency to evaluate the performance of active participants without accounting for those who failed and stopped participating. The traders who are available to evaluate are those who survived. The sample of survivors includes both genuinely skilled practitioners and lucky ones whose positive variance happened to persist long enough to build a track record. The survivors’ average performance looks better than the full population’s performance because the bad outcomes are no longer in the sample. A manager who has survived for years has passed a selection filter, but that filter does not distinguish between skill and luck.
Why does systematic price-based trading avoid the “fancy information” problem?
Because price is public information simultaneously available to all market participants. A systematic approach that uses price breakouts as entry signals and price reversals as exit signals does not depend on having access to information before other participants. The edge, if it exists, comes from the quality of the rules for responding to public price data rather than from information advantage. This makes the approach replicable, testable, and free from the regulatory concerns that non-public information-based trading raises.
Trend Following Systems
Want to learn more and start trading trend following systems? Start here.
