The Reversal Radar
The shape it looks for
A trend that is about to pause usually shows the same fingerprint: the higher timeframe still looks strong — that is the part everyone sees — while the lower timeframes have quietly stopped confirming it. Buyers who were going to buy have bought. The daily candle still looks impressive; the last few hours do not.
Scanning for that by eye across 31 instruments, around the clock, is not realistic. That is the only job the radar does.
The exact rule
The official preset, the one pushed to Pro subscribers in real time, is written 1d←4h+1h: a 1-day anchor confirmed by 4-hour and 1-hour counter-moves.
| Component | Condition |
|---|---|
| Anchor (1 day) | |σ| ≥ 1.2 — the pair is stretched |
| Slow confirm (4 hours) | ≥ 0.5σ against the daily direction |
| Fast confirm (1 hour) | ≥ 0.6σ against the daily direction |
| Coverage | 28 FX pairs + BTC/USD, XAU/USD, XAG/USD |
All three must hold at once. A stretched daily reading with no lower-timeframe turn is just a trend, and the radar stays silent. A one-hour wobble inside an intact trend is noise, and it stays silent for that too. Requiring two independent timeframes to agree is what keeps the signal count at a few per day rather than a few per hour.
The experimental weekly preset
A second preset shifts everything up one shelf: 1w←1d+4h, a weekly anchor confirmed by daily and 4-hour counter-moves. It targets position-sized swings rather than multi-day ones.
These are pushed too, tagged 🧪 in the message, and their record is tracked separately in the ledger. Mixing an experimental preset's statistics into the official ones would flatter or damage both, so the site does not do it. Lower-timeframe combinations (4h and 1h anchors) are visible on the page for your own screen-watching but are deliberately never pushed — at that frequency the signal-to-noise ratio does not justify a notification.
What a signal does not claim
A radar signal is an observation that a trend may be exhausting. It is not a forecast, not an entry, and not advice.
Our own backtests found that raw strength-reversal signals carry no standalone positive expectancy. We publish that on the ledger page rather than burying it, because the alternative — implying an edge we cannot demonstrate — is how most signal services work and is exactly what we did not want to build.
So why run it at all? Because the useful function is not prediction, it is attention allocation. Thirty-one instruments move continuously. The radar reduces that to the two or three currently showing a specific, defined structure, so the time you have for analysis goes to the instruments where something is actually happening. What turns that shortlist into a decision is the layer described in the workflow guide.
A useful way to keep yourself honest about any signal source — including this one — is to benchmark it against pure chance: the random forex pairs order generator deals random LONG/SHORT orders across the same 28 pairs. Anything that cannot beat that baseline after spread and swap is decoration.
The public ledger
Every signal the engine fires is written to the public ledger by machine, with its follow-up performance at +1h, +4h, +1d and +1w attached. Green marks the price moving against the prior daily trend within that horizon; red marks the trend continuing.
There is no curation step. New signals are locked for 24 hours on the public page — Pro subscribers already received them live — and then unlock automatically for audit. The reversal rate you see is whatever the machine recorded.
How to actually use it
- Treat it as a shortlist, not a trigger. The signal says "look here", not "act now".
- Check the positioning behind it. A reversal into a crowd that is already maximally positioned the other way has fuel; one into flat positioning does not. See the COT guide.
- Check what is on the calendar. An exhaustion signal four hours before a central bank decision is a statement about the last few hours, not about the next twelve.
- Prefer confluence over frequency. Several C-grade signals do not add up to one A-grade signal.