The Three Pillars of a Trading Model — Entry, Position Management, and Regime Filtering

What Makes a Trading Model, a Model?

A trading model is not just an entry signal. The word "model" implies a complete system — one with defined rules for when to enter, how to manage the trade once open, and what market conditions the whole thing is valid in. In darwintIQ, every model is built from exactly three components: an entry logic type, a position manager, and a regime filter. These three layers are evaluated independently, but they only produce meaningful results together.

Understanding what each component does — and what it doesn't do — is the first step to interpreting why one model outperforms another under the same market conditions.

Entry Logic — The Signal That Opens the Trade

Entry logic is the rule set that decides when a position should be opened. In darwintIQ, there are fifteen distinct entry logic types, each with a different way of reading price action. Some are trend-following in nature — TrendFollow and SmaCross look for directional momentum. Others are mean-reverting, such as RangeBounce and RegressionBandTouch, which look for price to return toward a central value after deviation. Still others respond to specific patterns or statistical conditions: MACD, BollingerBands, SqueezeBreakout, and BreakoutExtrema all identify structural moments when momentum is likely to extend.

The critical point is that no entry logic type is universally superior. A TrendFollow entry that performs strongly in trending conditions will often fail in a range. A RangeBounce entry that thrives in flat, oscillating markets may produce a string of losses the moment a directional move takes hold. Entry logic tells a model when to get involved — it says nothing about how much risk to take, or whether the current market is even suitable.

Position Management — What Happens After You're In

Once a trade is open, the position manager takes over. Its job is to define where the stop loss sits, how the trade is trailed if it moves in favour, and ultimately when the position is closed. This is where much of the performance difference between models emerges.

In darwintIQ, there are four position manager types. The Absolute manager uses fixed pip distances — the stop and target are set at a constant offset from the entry price. This is predictable but doesn't adapt to how volatile the market currently is. The ATR manager does adapt: it sizes stops and targets relative to the Average True Range, meaning it widens in volatile conditions and tightens in quiet ones. The SMATrail manager trails the stop behind a moving average, locking in profits as price trends. The SupRes manager uses identified support and resistance levels as natural stop and target zones — letting price structure define the trade's risk parameters.

Position management is where the difference between a model that survives a volatile week and one that doesn't is often decided. An entry that's directionally correct but paired with a position manager that exits too early — or too late — will still underperform. As we've explored in Why Position Management Matters More Than Entry, the signal gets you in; position management determines whether that matters.

Regime Filter — Knowing When Not to Trade

The regime filter is arguably the most underappreciated component of a trading model. Its purpose is to define the market conditions under which the model is allowed to participate at all. A model with no filter trades in every condition. A model with a well-calibrated regime filter trades only in conditions where its entry logic has historical validity.

darwintIQ supports several regime filter types. The SMA filter assesses trend direction using a moving average relationship. The RSI and RsiBand filters use RSI readings to determine whether the market is in a mean-reverting or momentum state. The SupResFilter uses structural levels to gauge whether price is in a high-information zone. The TrendRegimeFilter detects the presence or absence of a sustained directional bias. And NoFilter means the model participates regardless of regime.

A regime filter doesn't improve a model's entry logic — it decides when that logic is applicable. Pairing a range-entry strategy with a TrendRegimeFilter that deactivates the model during directional moves is a basic but powerful form of context awareness. The filter doesn't predict the future; it enforces discipline based on what the market is doing right now.

Final Thoughts

Entry logic, position management, and regime filter — three components, each doing a distinct job. In darwintIQ, these three layers are combined by the genetic algorithm and evaluated continuously over a rolling window, with each combination scored on live performance. The models that rise in the rankings are rarely those with the most sophisticated entry logic alone; they're the ones where all three components are suited to the same market environment, working in alignment rather than against each other.

When you review a model in the dashboard, look at all three components together. Knowing what each one contributes helps you understand not just how a model has performed, but under what conditions it's likely to continue performing.