Nick Baltas on Trend Following in 2026: Signals, Structure & Strategy | Systematic Investor | Ep.282

Episode Overview

• This episode explores the structural mechanics of trend following, defining it as a method of capitalizing on the market's slow digestion of information rather than just technical analysis.
• The hosts analyze the massive performance dispersion seen in 2023, attributing it to specific market environments—specifically "V-shaped" reversals—and the impact of universe size (concentrated vs. diversified portfolios).
• A significant portion of the discussion challenges "recency bias," warning investors against assuming that the recent outperformance of small, concentrated portfolios is a permanent shift rather than a cyclical anomaly.
• The conversation advances into quantitative strategy design, debating the merits of linear vs. non-linear position sizing (using "S-curves" to cap risk) and how machine learning validates these modern approaches.

Key Concepts

• Trend Following as Information Digestion Momentum is not a magical anomaly but a structural inefficiency. Markets do not price new information instantly; they "digest" it over time. This delay creates a price path (a trend). Therefore, a trend following strategy is essentially a mechanism for visualizing and trading the delay in consensus formation.

• Narrative vs. Price Action There is a critical distinction between "logical" narratives and actual market behavior. The example of Japan raising rates logically suggests the Yen should rise, yet prices often remain flat. Trend following succeeds by ignoring the narrative ("what should happen") and focusing exclusively on price ("what is happening"), acting as a filter that only trades when the market shows actual sensitivity to news.

• Dispersion Drivers: Speed and Universe Performance differences between trend followers are rarely random; they are driven by structural choices. In 2023, "Speed" was a differentiator because slow systems couldn't react to sharp "V-shaped" reversals fast enough. "Universe" was a factor because returns were concentrated in only a few sectors (like Metals), causing broad, diversified portfolios to drag compared to concentrated ones.

• The "Small Universe" Anomaly and Recency Bias Recent data (last 3-5 years) suggests trading a smaller, concentrated number of markets outperforms broad diversification. However, looking back 25 years reveals this is likely a cyclical anomaly. Historically, small universes were often the worst performers. Investors are warned against designing strategies based on this short-term window, as mean reversion suggests broad diversification remains safer long-term.

• Crisis Alpha: First vs. Second Responders Investors often misunderstand "Crisis Alpha." Strategies are either "first responders" (reacting immediately to volatility) or "second responders" (reacting to sustained distress). Trend following is a second responder; it requires time for a crisis to manifest as a directional trend. It may miss the initial crash but is designed to capture the prolonged dislocation that follows.

• Non-Linear Time Series Momentum (The "S-Curve") Traditional models scale position sizes linearly (the stronger the trend, the bigger the bet). New research and machine learning models suggest a "non-linear" approach is superior. By using a sigmoid (S-shaped) function, strategies can flatten or cap exposure when trends become extreme. This recognizes that at extreme levels, the probability of mean reversion increases, making it dangerous to continue increasing bet sizes.

Quotes

• At 0:04:06 - "What is momentum and trend following at the end of the day? It's just perhaps an expression of slow digestion of information." - Defining the fundamental economic reason why trend following works.
• At 0:07:02 - "How important it is not to have any kind of prediction in your model... [Japan] increased its interest rates by a lot. Clearly, that must mean that the Yen goes up... [but] the Yen has just done nothing." - Illustrating why relying on 'logical' narratives fails compared to following price.
• At 0:14:38 - "The universe and the speed are probably the two most important pillars of dispersion... at least for the last year." - Identifying why different trend followers had such different returns in 2023.
• At 0:18:41 - "The V-shapes are the ones to have caused this dispersion, rather than... 'oh actually you should have been longer term.' No, that's not what's going on here." - Explaining that market structure (sharp reversals), not just trend length, dictated winners and losers.
• At 0:24:50 - "We need to be very cautious of making too many conclusions based on people saying... 'look at this... we can outperform the CTA index.' ...If we were to expand this over a 25-year period, I'm not so sure that that would have been the same conclusion." - Cautioning against optimizing strategy design based on short-term data windows.
• At 0:28:40 - "It's not that a specific universe is better than another... it is more about that the V-shapes are the ones to have caused this dispersion." - Clarifying that performance differences often stem from specific market shocks rather than the inherent superiority of a strategy.
• At 0:31:50 - "The minute we depart from it [buying on positive signals/selling on negative], it is no longer just a trend strategy. It's something of a mixture." - Defining the boundary of trend following versus mean reversion strategies.
• At 0:36:50 - "Investors follow... an increasing concave utility function... which basically means that we enjoy gains less than the pain that we incur when a similar magnitude of a loss is coming." - Connecting behavioral economics to why avoiding large drawdowns is psychologically vital.
• At 0:39:27 - "Instead of us forcing them [non-linearities] by some sort of a parametric approach... we let the data speak, and maybe we use a neural network that allows us to uncover those relationships." - Discussing the modern approach of using machine learning to find optimal trading rules.

Takeaways

• Prioritize Price Over Narrative: Ignore "logical" economic outcomes (e.g., rate hikes must raise currency value) if the price action doesn't confirm it; markets often lack sensitivity to major news events.
• Resist Recency Bias in Strategy Selection: Do not switch to "small universe" or concentrated strategies simply because they won in the last 3 years; historical data over 25 years suggests broad diversification is statistically safer.
• Implement Non-Linear Position Sizing: Consider capping your trade size ("flattening the curve") when trends reach extreme levels to avoid being over-exposed when a trend inevitably reverses or mean-reverts.
• Evaluate Strategies by "Response Time": When seeking crisis protection, determine if you need a "first responder" (instant volatility protection) or a "second responder" (long-term trend capture), as trend following typically fills the latter role.
• Beware of V-Shaped Environments: Recognize that periods of sharp market reversals (V-shapes) are structurally difficult for slower trend systems; underperformance in these periods is a feature of the model, not necessarily a failure.
• Use Diversification as a Survival Tool: Maintain exposure to all asset classes even when specific sectors (like bonds) underperform for long periods, as profit drivers rotate cyclically over decades.