Every Zelos strategy runs on the same rule-based engine underneath — a market read, a pattern scan, a score, and a publish-or-pass decision. Nothing here is a black box: this page walks through exactly how a setup earns an alert, what the scoring actually weighs, how the three strategies differ, and what the jargon in an alert means.
Agentic trading is when a software agent applies a fixed set of trading rules on its own initiative — screening the market, scoring candidate setups, and deciding whether to publish an alert — without a person re-running the analysis by hand each session. The agent decides what qualifies; a person still decides whether to act on it. It's distinct from automated trade execution, which places orders on its own — an agentic trading system can (and, in Zelos's case, does) stop at the decision and the write-up, never the order itself.
Zelos is a rule-based agentic trading system: the same scoring engine described below runs every session, unattended, and either publishes a qualifying setup with its full reasoning or says plainly that nothing cleared the bar. See how the engine works for the exact steps, or the glossary for the terminology it uses.
What changes between strategies is the shape on the chart each one is hunting for. What doesn't change is the process that shape has to survive before it becomes an alert.
Before looking at any single stock, the engine classifies the broad market as trending up, trending down, or sideways. That read sets how strict everything downstream gets — the bar for a "good" setup rises when the backdrop is weak, and eases slightly when it's healthy.
Each strategy screens hundreds of stocks for its own specific shape — a pullback holding in an uptrend, a breakout that's actually holding above resistance, or a same-day setup already published by another strategy. Anything that doesn't match the shape gets thrown out immediately, no matter how exciting the ticker looks.
Every candidate that survives the scan gets scored, not just flagged. See the scoring system below for exactly what's weighed. Most days, most candidates don't clear the bar — that's by design, not a bug.
If something clears the bar, the alert page updates with the setup, the numbers, and the reasoning behind it. If nothing clears the bar, the page says so plainly. Nothing is ever forced onto the page to make a session look more active than it was.
A setup that matches the right shape still has to earn its score. These are the factors that make up that number on every alert page.
Is the stock's broader direction actually working in the trade's favor, or is this setup fighting the tape?
Is there a real level the setup is holding above (or breaking through), or is the "level" just a line drawn to fit the story?
Is participation confirming the move, or is this happening on thin, unconvincing volume that could reverse on nothing?
How much upside is on the table versus how far away the stop has to sit? A great pattern with a poor reward:risk still gets marked down.
Is this stock outperforming or lagging the broader market right now? Strength relative to peers matters as much as the chart in isolation.
Typically around 60 out of 80 to qualify. It's set high on purpose — "no trade today" is meant to be the most common outcome, not an edge case.
That's not inconsistency — it's the market-regime check working as intended. The broader backdrop changes how strict the scan is before a single stock is even screened.
Bullish continuation setups (pullbacks, breakouts) get a slightly easier bar — the wind is at their back.
The bar tightens across the board. Fewer setups qualify, and reversal or short-biased shapes get closer scrutiny.
Neither trend nor breakout patterns get the benefit of the doubt — setups need to prove themselves on their own merits.
Every row below runs through the same four-step engine above — this is where they diverge.
| Swing Trader | Breakout Rider | Options Scanner | |
|---|---|---|---|
| Pattern targeted | Pullbacks, bull flags, breakout+retests, and higher-low reversals in an uptrend | Clean breaks above resistance that actually hold, not spike-and-fade moves | Whichever directional setup (from either strategy above, or its own scan) clears an options-specific check |
| Direction / instrument | Long equity | Long equity | Single long call or put — never spreads, multi-leg, short options, 0DTE, or weeklies |
| Typical structure | Entry, stop, target 1/2, reward:risk | Entry (at the hold, not the spike), stop, target, reward:risk | Direction, ~1 strike OTM, 30–45 DTE, plus the underlying setup's reward:risk |
| Track record | Live — publishing real sessions | Live, newly launched — no long-run history yet | New — illustrative walkthroughs only so far |
| Price | $20 one time | $20 one time | $20 one time |
| See Swing Trader → | See Breakout Rider → | See Options Scanner → |
Plain-English definitions for the vocabulary used across every alert — no finance-degree required.
No account connection, ever. Zelos never asks for your brokerage credentials and never links to any account. There's nothing to authorize and nothing for it to touch — it publishes an alert page, full stop.
Nothing here places a trade. Every run only researches and reports. If you decide to act on an alert, you place that trade yourself, in your own brokerage account, on your own judgment.
It's rules, not a mysterious model. Every factor described on this page — trend, support, volume, reward:risk, relative strength, market regime — is an explainable, standard technical concept. Nothing here is a proprietary black box.
"No trade today" is a normal outcome. The bar exists to be missed most days — that's what keeps it a strategy instead of a habit.
Nothing here is a guarantee. This approach hasn't been backtested against a long track record. It's rules, applied consistently — not a promise of returns.