Cross-Asset Return Forecasting
Horizon-aware predictive models spanning equities, fixed income, commodities, and FX, with explicit treatment of cross-sectional and time-series structure.
Quantitative Investment Research
Machine learning applied to the
structure of financial markets.
KHH Analytics builds systematic models that identify persistent, statistically grounded signals across global markets — engineered to pursue materially superior risk-adjusted returns rather than headline volatility.
-
Signal Research
Predictive structure extracted from price, fundamental, macro, and alternative data under strict out-of-sample discipline.
-
Systematic Execution
Models translated into positions by a cost- and capacity-aware optimizer. No discretionary overrides, no narrative drift.
-
Risk-First Construction
Every allocation is sized against drawdown, correlation, and regime exposure before it is sized against expected return.
01 — The Thesis
Markets are efficient in aggregate and inefficient at the margins.
The strong form of the efficient market hypothesis has never survived contact with real data. What survives is something more useful and considerably harder to exploit: markets are mostly efficient, most of the time, for most participants — and the residual is faint, non-stationary, and buried in noise.
That residual is a machine learning problem before it is an investment problem. Conditional relationships that shift with regime, nonlinear interactions across hundreds of features, and signals with signal-to-noise ratios far below anything a human analyst could detect by inspection are precisely the domain in which modern statistical learning outperforms intuition.
The difficulty is not building models that fit. It is building models that generalize — surviving transaction costs, capacity constraints, structural breaks, and the relentless multiple-comparisons problem that renders most quantitative research irreproducible. Our entire process is organized around that single distinction.
02 — Methodology
A research pipeline built to reject its own ideas.
Most candidate signals die here. That is the point.
-
01
Data Engineering
Price, fundamental, macroeconomic, and alternative datasets are ingested into a point-in-time research environment free of survivorship and look-ahead bias. Every model sees only what was genuinely knowable at the moment of decision.
-
02
Signal Research
Gradient-boosted ensembles, deep sequence architectures, and probabilistic methods are trained under walk-forward validation. Candidates must clear out-of-sample thresholds, regime segmentation, and multiple-comparison correction before they are considered at all.
-
03
Portfolio Construction
Forecasts become positions through a constrained optimizer that prices in transaction costs, market impact, capacity, factor exposure, and the full correlation structure of the book — optimizing the portfolio, never the individual trade.
-
04
Monitoring & Decay Control
Live models are continuously measured against their research-period distributions. Performance decay, feature drift, and regime dislocation trigger systematic de-risking well before they become losses worth discussing.
03 — Capabilities
What we build.
Research and infrastructure across the full modeling lifecycle.
Alternative Data Integration
Evaluation, cleaning, and orthogonalization of non-traditional datasets — measuring genuine incremental signal rather than repackaged beta.
Regime Detection
Unsupervised and state-space methods that identify shifts in market character early enough to adjust exposure rather than explain losses.
Volatility & Tail Risk Modeling
Conditional volatility forecasting and tail-dependence estimation that inform position sizing directly, not as an afterthought to allocation.
Factor & Attribution Analytics
Decomposition of returns into compensated risk premia versus genuine alpha, so that performance is understood rather than merely observed.
Custom Model Development
Bespoke research engagements for allocators and asset managers, delivered with full methodological transparency and reproducible code.
04 — Philosophy
Return is a claim. Risk-adjusted return is an argument.
Raw return is trivially easy to manufacture: add leverage, concentrate exposure, sell volatility, and wait. Each of these will outperform a broad index for a period of time, and each will eventually surrender that outperformance in a single adverse move.
We evaluate every strategy on what it earns per unit of risk taken — Sharpe and Sortino ratios, maximum drawdown, time to recovery, tail-loss behavior, and the stability of those measures across independent out-of-sample periods and market regimes. A model that beats the market with twice the volatility has not beaten anything.
This discipline is also what makes returns compoundable. Capital that avoids severe drawdowns compounds from a higher base, and the arithmetic of that advantage widens with every year it is sustained. Risk control is not a constraint on performance. Over any horizon that matters, it is the performance.
05 — Contact
Discuss your mandate.
We work with allocators, family offices, and asset managers seeking systematic, evidence-based exposure. Reach out to begin a conversation.
contact@khhanalytics.com