Stop Hunt Radar [GBB]STOP HUNT RADAR
Have you ever placed a stop loss under a swing low, watched price come down, take out your stop to the tick, and then run exactly where you said it would? That's not bad luck and it's not a conspiracy. Below every obvious low sits a cluster of sell orders — your stop, my stop, the breakout traders' sell-stops — and for anyone who needs to buy size, that cluster is the only spot on the chart with guaranteed forced selling waiting at a known price. Your stop wasn't hunted out of malice. Your stop was the liquidity.
This indicator maps those clusters in real time, classifies what happens when they get hit, and keeps honest records. That last part matters more than you think.
How is works
The shaded bands are stop pools. "Sell stops · 60,621" means stop losses from longs are probably resting under that level; the shading shows the pocket where they sit. Color is a ranking, not a direction: gray is a minor fresh swing, pink is the most loaded level currently on your screen. Hover any level and it tells you in plain words why it's ranked — "3 equal lows · yesterday's low · 4.2× volume". No codes to memorize. (If you prefer compact BSL/SSL labels, switch Label Style to Pro.)
When price hits a pool, the radar decides what happened by fixed rules, on closed bars, with no repainting:
⚡ swept — price wicked through and closed back. The marker stays on the chart with the measured raid depth ("⚡ 2.4 ATR").
✕ broken — price closed through and stayed.
⌛ undecided — price closed through but might come back. The radar waits a few bars before calling it instead of guessing. Capitulation V-reversals get correctly labeled as sweeps because of this.
The dotted lines near ranked pools are the part I'm most proud of: depth guides. They're measured from this chart's own history — "typical sweep · 63,955" marks how far the median raid ran past that level, "1-in-20 sweep" marks the bad case. Now look where the textbook stop placement is. Usually inside the typical zone. That's the whole point of this tool in one picture.
IMPORTANT
Every channel that shows you stop hunts follows up with the same pitch: buy the sweep, ride the reversal. Before publishing this, I tested that. Properly — the engine was rebuilt in Python, verified bar-for-bar identical against this script, and run through a pre-registered study on 24 months of BTC, ETH and SOL. The final test ran exactly once, on 21,813 sweeps of held-out data the tuning never touched.
Result: after a sweep, the average forward move is statistically indistinguishable from entering at a random moment. The strongest-ranked pools did not reverse harder — if anything slightly worse. The dramatic panic-volume wicks leaned toward continuation, not reversal.
So no, the lightning bolts are not buy signals, and I won't pretend otherwise. The live tally on your own chart will show you the same thing — reclaim rates around 55–60%, a coin flip with commission. This is why the radar prints the losses too. If someone shows you this pattern with only winning examples, you now know what got cropped.
WHAT IT'S ACTUALLY FOR
Stop placement, mostly. The depth guides answer "is my stop sitting in the feeding zone?" before you find out the expensive way. Either place it past the typical-sweep line or size down knowing the risk. Second: when a level gets hit you get a rule-based verdict — swept or broken — instead of arguing with yourself, and alerts fire on those events so it watches the levels while you don't. Third: the receipt trail audits your beliefs. You think "they always sweep the lows here"? Scroll back. The chart kept score.
ANY MARKET, ANY TIMEFRAME
The geometry is ATR-scaled and the tolerances are percentage-based, so it self-adapts. The stats are measured per chart — a Gold chart shows Gold's raid depths, not Bitcoin's. Auto-Adapt handles the asset-specific details: forex gets proper big-figure/half-figure round numbers (159.50 on USDJPY, 1.0850 on EURUSD), and symbols without volume data get the volume factor switched off. When an adaptation is active, the legend says so. One honest caveat: the validation study was crypto on 5m. Other markets run the same mechanics but ship with their own live base rates instead of borrowed claims. Stocks gap — gap-throughs get adjudicated by the same waiting rule, but read markers around opens with some skepticism.
GOOD TO KNOW
Colors are relative: a pool's color can shift as stronger or weaker pools appear on screen. That's ranking, not repainting — the absolute score lives in the tooltip and never changes retroactively. Depth guides appear once the chart has logged at least 10 classified sweeps per side, because statistics from three events aren't statistics. If another indicator ever squashes your chart, right-click the price scale and enable "Scale price chart only" — worth doing in general. The optional dashboard (off by default) adds the nearest pools and the rolling base rates in a corner panel.
THE SETTINGS
You don't need to touch any of this — the defaults are the validated configuration and what I run myself. But it's all there if you want it. The settings are grouped the same way the panel is.
Display — the stuff you'll actually use. Min Score To Display hides weak pools (0 shows everything). Display Radius (default 3%) hides pools too far from price; they're still tracked, they just reappear when price comes back. Show Pool Labels, Show Radar Dashboard (off by default — the corner panel with nearest pools and base rates), Show Legend, Show Sweep Base Rates. Label Style is the big one: Beginner spells everything out, Pro uses compact BSL/SSL codes. UI Text Size and Label Size — bump these up for screenshots and video. Sweep Depth Guides plus "Hide Guides For Gray Pools" (rank 0–10, default 3): raise it to only annotate the strongest levels, lower it to 0 to guide every pool. Auto-Adapt To Asset Class — leave this on unless you have a reason.
Swing Detection — how a level is found. Pivot Left / Pivot Right define how many bars each side make a swing (8/3 default — bigger = fewer, more significant levels). Max Armed Life retires a level that's gone untested for too long.
Pool Geometry — ATR Length drives all the scaling. Pocket Depth sets how thick the shaded stop pocket is. EQ Merge Tolerance controls how close two swings must be to count as the "same" level and cluster together — by % of price (default) or ATR.
Sweep / Outcome — the classification rules. Min Sweep Penetration is how far past a level a wick must go to count as a touch. Multi-Bar Sweep Grace (default 10 bars) is the window where a close-through can still turn back into a sweep — this is what catches capitulation reversals instead of mislabeling them as breaks; set it to 0 for strict single-bar sweeps only. Reclaim Confirmation and Outcome Watch Window define what counts as a confirmed reclaim and how long the radar watches before giving up.
Scoring & Heat Weights — what makes a level rank high. Four weights: EQ Cluster Size, Untested Age, Confluence, Formation Volume. The "Norm" values are how much of each earns a full score (e.g. 3 equal pivots = full cluster score). Session Confluence toggles the prior day/week levels; Round Number Confluence and its step (0 = auto, and auto is asset-aware) toggle round-number weighting. Fair warning: the study found the heat score doesn't predict sweep outcomes, so retuning these changes what looks prominent, not what works. I left mine at default and I'd suggest you do too.
Heat Palette & Receipts — looks. Heat Color Scale is Relative by default (the ramp stretches across the pools currently on screen); switch to Absolute if you want a fixed 0–10 meaning. The four color stops (Cold / Warm / Hot / Prime) are the ramp. Outcome Receipts: show Wins + Losses (default — the point), Wins Only, or None.
Alerts — High-Rank Pool Score is the threshold for the "approaching a strong pool" alerts (it's on the absolute score, where pools typically sit around 1–4). Proximity is how close counts as "approaching". The sweep and break alerts themselves fire on the events directly — add them from the alerts dialog.
Pine Script® indicator
