Editorial note: this conversation is a curated transcript from one of our internal agent debriefs. The agent's "voice" is a writing convention — under the hood it's algorithms and tools — but the way it reasons about its job is real.

xris team: Tell us what you do.

S/R agent: I read price and volume. That's it. I don't read news, I don't track macro, and I have no opinion on whether a token is a good investment. My job is to look at the chart and tell you, with as much precision as the data allows, where price has actually reacted in the past — and where it would have to reach to be a real target or a real stop from here.

xris team: "Actually reacted" sounds like a small word. Why does it matter?

S/R agent: Because most "support and resistance" you see in trading content is drawn, not measured. Someone looks at a chart, picks a level that feels important, draws a line, and tells a story about it. That's fine as a discussion device. It's a terrible thing to trade off.

I work the other way around. I count touches. I check whether the touches had volume. I cluster them into zones with a tolerance based on the token's ATR. Each zone gets a tier — Major if it's defended at least five times with at least one volume-confirmed touch and one in the last 30 days; Strong if it's defended but missing one of those criteria; Notable if it's only been touched twice or three times and isn't strongly held.

When you see a level in the dashboard, you can click through and count the touches yourself. I never invent.

xris team: What's the most common mistake you see traders make when they read levels?

S/R agent: Treating all levels as equal. A Major level with five body touches and volume confirmation is not the same as a Notable level with two wick touches eight months ago. The first one is a wall. The second one is a memory.

Also: forgetting that distance matters. A level five percent away matters less for an entry decision than one one percent away. The Near tag in the table tells you which tokens have price sitting right at a support or resistance — those are the ones where the next candle matters most.

xris team: You don't pick the take-profit and stop-loss directly — there's a cascade rule. Walk us through it.

S/R agent: Right. The naive thing to do is: pick the closest resistance as the target, the closest support as the stop. Done. But that gives you garbage setups, because the closest level is often inside a single ATR — meaning the move from here to that level is statistically noise, not signal.

So I cascade. For the take-profit side, I walk through resistances above price in order. If the first one is within one ATR of the entry, I skip it and look at the second. I stop at the first level that's at least one ATR away. For the stop-loss side, the rule is asymmetric — 0.75 × ATR — because well-defended supports just inside one ATR are useful as stops, even if they're not far enough to be targets.

After picking, I apply a final filter: if the level I picked is closer than 0.5 × ATR after the walk, I drop it. I'd rather give you no setup than a bad one.

xris team: That's where the "Target (R2 ↗ from R1)" label comes from?

S/R agent: Yes. R1 means the literal nearest resistance above price. R2 is the second-nearest. If the cascade walked from R1 to R2 because R1 was inside one ATR, the dashboard shows you both. You see what I tried to do, what I rejected, and why. No magic.

xris team: What about a setup you got wrong?

S/R agent: Plenty. The honest answer is: I'm right about the levels — they exist, they're measurable — and I'm right about the cascade math. What I can't do is predict whether the level will hold under whatever exogenous shock is coming.

If I tell you "support at $78,700, Major tier, volume confirmed, five touches" and the next day a regulatory headline drops and price goes through it like it isn't there, I wasn't wrong about the level. The level was real. The market just had a more important input.

That's why I work in a team. The News agent and the Macro agent and the Risk agent are there to flag the exogenous stuff I can't see. If they're quiet, my levels are more likely to hold. If they're flagging something, the same level is statistically less reliable.

xris team: Anything you wish traders did differently?

S/R agent: Look at the tier. It's the single most-ignored thing in my output. People grab the closest level and trade it without checking whether it's Major or Notable. A Notable level breaking is unsurprising. A Major level breaking is news.

Also: read the cascade label. "Target (R2 ↗ from R1)" means the math walked past the obvious target. The setup is taking a longer move. Position size accordingly.

xris team: Last question. What changes when you ship — what should we expect from you in the next six months?

S/R agent: Multi-timeframe cascading. Right now I work on the daily. I want to publish levels and setups on the 4h and the 1h too, with the same rigor, so intraday traders get the same treatment. I also want to expose more granular touch metadata — body vs wick, time-of-day clustering, average rejection size — for the analytics view.

But the core promise won't change. You ask me where the levels are, I tell you. I count, I don't invent.

The S/R agent's full output is visible on every token in the dashboard. For the methodology in long form, see how the S/R agent finds levels. For interviews with the rest of the team, start with the News agent.