Support and Resistance Trading — What the Levels Actually Measure

A support and resistance trading strategy uses price levels where the market has previously stalled or reversed. Support is a level where falling price has tended to find buyers; resistance is a level where rising price has tended to meet sellers. Trades are placed on the expectation that the level holds, or on the expectation that it breaks.

Support and resistance appear across darwintIQ in three places: the SupRes entry logic, the SupResPM Position Manager and the SupResFilter Regime Filter. That reach reflects how much of market structure the idea captures.

What a level represents

A level forms where price turned several times. Each turn is evidence that supply or demand was concentrated there, perhaps because of prior orders, round numbers or earlier highs and lows. The more often a level has held, and the more recently, the more weight it carries.

It follows that levels are statistical tendencies. They describe where reactions have been likely, not where they are certain. This is the same logic that governs every metric in the platform: a pattern is a probability, and the useful question is how large it is and how stable.

Two ways to trade a level

The first is the bounce: enter near the level, expecting a reversal, with a stop just beyond it. It resembles range trading and suits Range Dominant conditions. The second is the break: enter when price closes decisively through the level, expecting continuation. It suits trending conditions and sits close to the breakout strategy family.

Both use the same level and opposite expectations. Which is right depends on the market state, so the regime matters as much as the level itself.

Where support and resistance trading fails

Levels fail in three common ways. They are mis-drawn, with too many or too few touches counted. They are broken, with price pushing through and not returning. Or they are faked: price pierces a level briefly and reverses, trapping traders on the wrong side. That last pattern is the subject of the FailedBreakout entry logic.

Because of this, a SupRes model's value depends on what surrounds the entry. A structure-based Position Manager such as SupResPM places exits relative to levels, so risk is defined by the same structure that justified the trade. A SupResFilter can hold the model back when price sits in the middle of nowhere, far from any level worth trading.

Evaluating a SupRes model in darwintIQ

In the Trader Detail view, read the usual metrics in the context of the logic. A bounce-style model should show a respectable Win Rate with contained Drawdown; a break-style model may show a lower Win Rate with a stronger Risk/Reward. The Robustness Score and Stability Score indicate whether the levels it uses keep working across the window or only worked during one episode.

Check the symbol's regime in the Market Panel as well. A SupRes model ranked highly during a Range Dominant phase may fall back as the market starts trending through its levels. That movement is information, not noise.

How many touches make a level

A level touched once is an observation, and a level touched five times is a pattern, but there is a limit. Each additional touch also uses up resting orders at that price, so heavily tested levels can become more likely to break. This is one reason SupRes parameters are evolved rather than set by rule: the right sensitivity differs between instruments and between regimes. A model that counts every minor turn will trade constantly and carry high Exposure with little edge, while one that demands many touches will trade rarely and may lack the sample size for its metrics to mean much. The evaluation window shows which compromise is currently working.

Final thoughts

Support and resistance endure because they describe real concentrations of market behaviour. They are also the most over-trusted idea in technical analysis, because a level that has held says nothing certain about the next touch. The sensible stance is the one a rolling evaluation enforces: treat each level-based model as a hypothesis, check how it is performing now, and be ready to see it replaced.