The number of independent parameters that can be freely adjusted in a trading model. Each additional degree of freedom gives the model more flexibility to fit historical data—and more surface area for curve-fitting. A model with 15 free parameters can fit almost any historical dataset regardless of whether it contains genuine signal.

No-BS Truth: Degrees of freedom are not a feature. They are a liability. Every degree of freedom you add to a strategy increases the probability that its backtested performance is a statistical artifact rather than a real edge.

The principle that, among competing models with equivalent explanatory power, the simplest model should be preferred. In trading, parsimony means that if two strategies produce similar out-of-sample performance, the one with fewer parameters and conditions is the superior choice—because it has achieved equivalent results with less overfitting risk.

No-BS Truth: Complex strategies that outperform simple ones in backtests are almost always exploiting the additional degrees of freedom to fit historical noise. Parsimony is not a constraint. It is a quality signal.

A trading advantage derived from persistent, fundamental characteristics of market structure rather than from historical pattern-mining. A structural edge exists because of how markets are built—index construction, institutional capital flows, volatility regimes, liquidity dynamics—not because it happened to appear in a specific historical dataset.

No-BS Truth: Structural edges are the only edges worth building strategies around. They persist across regime changes because they are derived from market mechanics, not market history. If you cannot explain why your edge exists from first principles, it is probably a backtest artifact.

The proportion of meaningful price information relative to random market fluctuation in a given dataset or indicator. Adding more indicators to a strategy does not increase the signal-to-noise ratio—it typically decreases it, because each additional indicator introduces its own noise component while adding marginal signal at best.

No-BS Truth: More indicators means more noise, not more signal. The goal of strategy design is to identify the smallest set of inputs that captures the most signal. Every indicator you add beyond that point is reducing the quality of your model.

The statistical adjustment applied to complex models‘ performance metrics to account for their additional degrees of freedom. In model selection frameworks like AIC (Akaike Information Criterion) and BIC (Bayesian Information Criterion), complex models are penalized for each additional parameter—because additional parameters are expected to improve in-sample fit by chance alone. Simpler models that achieve comparable fit receive no penalty.

No-BS Truth: Every backtesting framework that does not apply a complexity penalty is lying to you. A strategy with 15 parameters will always appear to outperform a strategy with 3 parameters on in-sample data—by construction, not by merit.

Simple strategies outperform complex ones in live trading for the same reason they outperform in academic forecasting research: they generalize better. Complex strategies fit historical data more precisely—including both the genuine signal and the random noise. When the noise does not repeat in future data (which it never does), the complex strategy loses the performance it extracted from fitting noise. Simple strategies capture less of the historical noise, so they lose less when the noise changes. This is not a theoretical argument. It is the consistent empirical finding of the M-Competition forecasting studies over four decades.

Degrees of freedom are the number of independent parameters in a trading strategy that can be adjusted during development. A strategy with a 50-day moving average and a 20-day moving average has at least two degrees of freedom—the lookback periods for each average. A strategy with 10 indicators, each with adjustable parameters, might have 20 or more degrees of freedom. Each degree of freedom allows the strategy to fit historical data more precisely—and increases the probability that the fit is capturing noise rather than signal.

Three practical tests: First, can you explain the strategy’s edge in two sentences without referencing specific parameter values? If not, it is too complex. Second, does removing any single condition dramatically change the backtested performance? If yes, that condition is exploiting noise, not signal. Third, does the strategy’s out-of-sample performance fall significantly below its in-sample performance? If yes, the strategy has more degrees of freedom than the data can support. A robust simple strategy’s in-sample and out-of-sample performance should be broadly comparable.

A structural market edge is an advantage derived from persistent characteristics of how markets are built and how capital flows through them—not from patterns that happened to appear in a historical dataset. Structural edges matter because they persist across regime changes. An edge based on Nasdaq 100 index construction and institutional capital flow dynamics exists for structural reasons that do not change when the market enters a new volatility regime. An edge based on a historical pattern has no persistence guarantee—the pattern was in the data by definition, but there is no reason for it to recur.

Ordertune’s strategy design is anchored to a structural thesis about Nasdaq 100 price aggregation dynamics—a thesis that can be stated in two sentences and that exists for identifiable first-principles reasons. The development process applies explicit complexity penalties: strategies are evaluated not on raw backtested performance but on performance per degree of freedom. Any strategy that requires more than a small number of free parameters to express its thesis is rejected before further development. The result is a set of strategies that are transparent enough to stress-test completely, simple enough to execute without ambiguity, and structurally grounded enough to persist across changing market conditions.