Warren Pies & Fernando Vidal, 3Fourteen Research - I Think That The Next 40 Years Are Unlikely To Look Like The Last 40 Years | #315

Episode Summary

Executive Summary: Warren Pies and Fernando Vidal of 314 Research explain why they founded a research shop combining deep domain expertise in real assets with machine learning rigor. They outline how their models work, emphasizing explainable signals, out-of-sample validation, trend breadth, and hierarchical risk parity. They also share current views: oil is neutral, inflation fears are overstated, gold is pressured by rising rates, the dollar is near a key turning point, and Bitcoin is an important diversifier in portfolios.

Main Topics: Origin story of 314 Research (Priority: 5/5): The guests describe their path from Ned Davis Research to launching an independent firm that blends qualitative market expertise with quantitative and ML tools. The firm emerged from a longstanding working relationship and a belief that real assets deserve more focus in a new market regime. Model design and research discipline (Priority: 5/5): They stress rigorous model-building: clear questions, failure modes, cross-validation, out-of-sample testing, and avoiding overfit 'backtest beauty contests.' Explainability is central, so model components are shown rather than hidden in black boxes. Oil framework and market call (Priority: 5/5): Their oil model combines inventories, technicals, positioning, and physical-market factors. Despite bullish indicators like draws and strong technicals, the model is neutral because futures positioning is stretched and OPEC spare capacity is still massive. Inflation debate and Fed outlook (Priority: 5/5): They argue current inflation fears are mostly supply-side distortions rather than broad demand-driven overheating. Their reading of CPI dispersion, slack in the economy, and break-evens suggests the Fed can stay patient and need not tighten aggressively yet. Real asset allocation and portfolio construction (Priority: 4/5): Their asset-allocation model covers 17 assets and uses trend, correlation, and volatility, with hierarchical risk parity for robust allocation. They avoid fundamentals across heterogeneous assets and instead lean on trend breadth and risk-aware weighting. Bitcoin, gold, and the dollar views (Priority: 4/5): They see Bitcoin as an emerging uncorrelated asset worth a meaningful position; gold is treated as an 'infinite duration' asset hurt by rising nominal rates; and the dollar is at a transition point, with their model turning neutral but a longer-term bearish view on structural fiscal spending. Business model and client demand (Priority: 3/5): 314 Research serves institutions, RIAs, hedge funds, asset managers, and family offices through subscription research, daily model updates, weekly reports, and custom versions. Early interest has skewed more institutional than expected, with strong client focus on Bitcoin and inflation.

Key Arguments: A research process should be able to fail; the model should not be forced to produce a signal if the evidence is mixed. Price action is the one factor that consistently matters across heterogeneous assets, so trend analysis is central in every model. Oil is not an obvious supercycle yet because OPEC spare capacity remains a major overhang, even if inventories and price action look supportive. Today's inflation narrative is driven more by supply-chain bottlenecks and commodity-specific shortages than by economy-wide overheating. The Fed likely has room to remain accommodative because CPI dispersion is extreme and broad structural inflation is not yet confirmed by break-evens or slack measures. Gold is highly rate-sensitive and behaves like an infinite-duration asset, so rising nominal rates can suppress it even when inflation narratives are loud. Bitcoin can be valuable in portfolio construction because it is uncorrelated and can be scaled down by trend-based risk controls when drawdowns hit. Hierarchical risk parity is preferred over mean-variance optimization because it is less sensitive to unstable correlation estimates and produces more robust allocations.

Data Points: Oil model components: 4 - Inventories, technicals, positioning, and physical markets are the four inputs in the oil model. Real asset allocation model size: 17 assets - Their high-level allocation framework compares 17 heterogeneous assets. Yield optimizer inputs: 13 income-producing assets - They built an income-focused model around 13 yield-generating assets. Bitcoin position: 8% - The real asset allocation model has held roughly an 8% Bitcoin weight since inception. Model portfolio decline in 2018: 5.5% - Despite Bitcoin falling about 75-80% peak-to-trough, the model portfolio declined only 5.5%. Bitcoin decline in 2018: 75-80% - Used to illustrate how the model reduced exposure as trend weakened. 60/40 performance in 2018: roughly -3% - Referenced as a benchmark that slightly outperformed their model that year. Average equity weighting historically: 38% - Historical average equity exposure in the real asset allocation model. Equity weighting at time of interview: 43% - Current equity exposure after scaling back from earlier levels. Equity weighting at year start: ~50-51% - The model began the year with a larger equity allocation. Bond weighting: 23% - The model remained underweight bonds at the time of the discussion. Government spending share of GDP: up about 50% - They said U.S. federal spending rose from around 20% of GDP to about one-third. U.S. federal spending level: about one-third of GDP - Used to support the thesis of structurally higher government spending. OPEC spare capacity: almost 10 million barrels - Cited as the main reason oil could remain capped despite bullish near-term signals. Oil model stance: neutral - The model was neutral through much of the rally because bullish and bearish factors offset. Dollar model stance: buy to neutral - The dollar model recently moved from buy to neutral near long-term support. Dollar support level: 89.90 - A break below this level was highlighted as important for longer-term weakness. Median correlation of CPI components to CPI: negative - They said this was the first time since the 1950s that the median correlation of 200+ CPI components to CPI turned negative. CPI components examined: more than 200 - Used in their inflation dispersion analysis. Break-even threshold to worry on inflation: 3-4% - They said five-year break-evens would need to move to this range before the Fed should worry more seriously. Break-evens gap from concern level: about 100 bps - They said current break-evens are still roughly 100 basis points below their concern threshold. Gold timing vs rates: Aug. 6 vs Aug. 4 - Gold topped on August 6; nominal rates bottomed on August 4, underscoring rate sensitivity. Energy sector yield: highest yielding sector since April 2019 - They noted energy became the highest-yielding sector in the market and stayed there through present day. Energy yield premium to utilities: about 100 bps - Energy yield was said to be roughly 1% higher than utilities.

Pivotal Quotes: "We're not in this backtest beauty contest business in this company. That's for sure." — Warren Pies: Explaining 314 Research's emphasis on rigor, failure testing, and avoiding overfit models. "You want to build your conviction off fundamentals, but you want to manage risk off of price action and technicals." — Warren Pies: Summarizing the firm's philosophy on combining qualitative conviction with technical risk management. "Gold, in my view, is an infinite duration asset." — Warren Pies: Describing why gold is highly sensitive to rising interest rates and reopening dynamics.

Implications: Listeners should expect a disciplined, model-driven framework rather than narrative chasing. The episode suggests real assets, Bitcoin, and selective energy exposure may matter more in a new regime, while inflation and policy calls require evidence, not anecdotes.