Exhaustion + Band — fading the giant candle
Discovered in round 10 of our studies: an entry orthogonal to Double Reversal that survived out-of-sample on both JPY pairs.
The rule: hunt the "giant" candle
Exhaustion + Band trades the exhaustion of an explosive move. It enters when, on the same candle:
- The candle range is ≥ 2× the previous ATR(14) — a giant candle vs the recent norm;
- The body is ≥ 60% of the range — directional move, not a wick;
- The candle closes OUTSIDE the Bollinger band — the final overshoot.
Giant down candle closing below the band → CALL (fade). Giant up candle above the band → PUT. The logic: a move 2x bigger than normal that still closes outside the band tends to be a climax — everyone who had to enter already has.
How to measure the three conditions
Everything is measured on the closed M1 candle. Mid-candle, a bar can look giant and outside the band and then close back inside — that is not a signal.
- Range = high − low of the candle. The yardstick is ATR(14) as it stood before this candle, so the giant candle does not inflate its own reference. Condition: range ≥ 2 × ATR.
- Body = |close − open|. Condition: body ≥ 60% of the range. This is what separates a real push from a spike that was already rejected.
- Close outside the Bollinger band: a down candle closing below the lower band gives a CALL; an up candle closing above the upper band gives a PUT.
All three must be true on the same candle. Two out of three is a different setup — and the table below shows why those near-misses are left alone.
Worked example — illustrative candles
The prices below are invented to show the mechanics — not a real GBP/JPY chart and not a trade result. Suppose ATR(14) before the candle is 0.030 (3 pips) and the lower band sits at 198.489, a Wednesday at 21:14 UTC-3.
| Candle | Range ÷ ATR | Body ÷ range | Close vs lower band | Signal? |
|---|---|---|---|---|
| A: open 198.540, high 198.548, low 198.462, close 198.470 | 0.086 ÷ 0.030 = 2.9× | 0.070 ÷ 0.086 = 81% | 198.470, below | CALL |
| B: same range, but close 198.528 after a long lower wick | 2.9× | 14% | Above | No — a pin bar, which measured 47–49% on M1 |
| C: range 0.050, full body, close 198.480 | 1.7× | 90% | Below | No — not giant; that is Double Reversal territory if RSI turns |
| D: range 0.090, body 85%, close 198.495 | 3.0× | 85% | Above (inside the band) | No — no final overshoot |
For candle A the order goes in as the candle closes, entry ≈ 198.470, with a 2-candle expiry (120 s). If the second candle after entry closes at, say, 198.492, the CALL wins: +$4.25 on a $5 stake at 85% payout. If it closes at 198.455, the loss is −$5.
The numbers
| Pair | Overall | Test | FRESH (out-of-sample) |
|---|---|---|---|
| USD/JPY | 60.8% | 60.0% | 56.6% (n=488) ✅ |
| GBP/JPY | 57.6% | — | 58.5% (n=340) ✅ |
Bonus finding: with a 2-candle expiry the same entry rises to 60.8% test / 58.7% fresh — the fade needs a bit more time to complete. We tested "looser" variants (exa25, exa_solto, RSI confirmation): none improved — the edge lives in the exhaustion+band COMBINATION, not in loosening parameters.
The perfect partner for Double Reversal
Most valuable of all: Exhaustion+Band is orthogonal to Double Reversal — it fires on different candles (candle anatomy vs distance from mean). Running both in individual mode (union) doubles the signal count at the same accuracy: 59.9% overall, 58.4% fresh (n=1,597 and 498). It's the champion config our bot ships with.
Why fading a giant candle works on M1 — and when it does not
During liquid hours, a 1-minute candle twice the normal size that still closes beyond the band is usually the tail end of a burst of orders: stops being triggered, late traders chasing. Once that flow is spent, the resting orders on the other side pull price back toward the mean. It is the same mean-reversion behaviour that made trend strategies lose on M1 in our study (~46–49%) and made reversal strategies the only ones approved. The condition is liquidity: overnight there is nobody on the other side, and hours 00, 04 and 05 (UTC-3) fell below break-even.
Do not run it in these cases:
- Around high-impact news — the biggest risk for this strategy in particular. A scheduled release (central-bank decisions for USD, JPY or GBP, the US jobs report) produces exactly the giant, full-body candle the rule looks for. But it is new information being priced in, not exhaustion. Our study did not isolate news candles, so there is no number here; the precaution is to keep the bot off around those releases.
- OTC pairs — ~50% for every strategy (test).
- Fridays — 48.8% in the fresh window, below break-even (hours study).
- Trending sessions. When giant candles keep printing in the same direction, the market is directional — the regime where reversion lost in our data. Use
pausar_apos_perdas+cooldown_minandstop_loss_diato cap the damage. - Payout under 80% — see the next table.
Expiry and payout, in money
Expectancy per $1 staked = accuracy × payout − (1 − accuracy); break-even = 1 / (1 + payout), which is 55.6% at 80% payout and 54.1% at 85%. For reference, 58% accuracy works out to +0.073 per $1 at 85% and +0.044 at 80%. With this page's numbers:
| Scenario | Accuracy | Expectancy at 80% | Expectancy at 85% |
|---|---|---|---|
| USD/JPY, fresh window | 56.6% (n=488) | +0.019 | +0.047 |
| GBP/JPY, fresh window | 58.5% (n=340) | +0.053 | +0.082 |
| 2-candle expiry, fresh | 58.7% | +0.057 | +0.086 |
| Study comparison: 1-candle expiry | 57.5% | +0.035 | +0.064 |
| Study comparison: 2-candle expiry | 60.8% | +0.094 | +0.125 |
At a 70% payout the break-even jumps to 58.8%, and every row above except the last one turns negative. On a $5 stake, USD/JPY at 80% payout expects about +$0.09 per trade; the 2-candle fresh figure at 85% expects about +$0.43. Those are averages over hundreds of trades, not what one session will show. The estimates are also uncertain: 56.6% on n=488 has a 95% interval of roughly 52.2%–61.0%, and the low end is under break-even. The expiry choice is the cheapest improvement available: same signal, ~3 points more accuracy.
Enabling and adjusting it in the bot
estrategias = reversao_dupla,exaustao_bbwithmodo = individual(orexaustao_bbalone).periodo = 60and 120 s expiry.pares = USDJPY;pares_sombra = GBPJPYif you want GBP/JPY logged without orders — the bot holds one order at a time, and an extra live pair can take the slot from a better signal (pairs study).horas_boas = 09:00-10:59,18:00-23:59(UTC-3, i.e. 12:00–13:59 and 21:00–02:59 UTC),dias = mon-thu,payout_min = 80,conta = demoto start.
Should you tune the 2× ATR / 60% body thresholds? We already tried: looser variants (exa25, exa_solto) and an added RSI confirmation — none improved on the original. If you want more trades, add Double Reversal in union rather than loosening this rule. Any change should go through simular_csv.py first, judged on n ≥ 150 candles the settings never saw.
Why not majority mode with Double Reversal? In maioria with only these two strategies active, the automatic minimum (half + 1) is 2 votes, so the bot waits for both to fire on the same candle. Because they are orthogonal, that rarely happens — and the one confirmation we tested on this entry (RSI) did not help. The tested combination is the union in individual mode: 58.4% fresh on n=498. Details in the Double Reversal guide.
Common mistakes with this setup
- Counting the wick. A giant range with a small body is a pin bar — 47–49% on M1, even inverted.
- Deciding before the close. The candle that looks like a signal at 40 seconds often closes back inside the band.
- 1-candle expiry. It throws away the ~3 points that a 2-candle expiry adds.
- Loosening the rule for volume. Tested, no gain. Combine with Double Reversal instead.
- Doubling after a loss. Martingale creates no edge.
- Judging after 20 trades. Even at 150 demo trades, a true 58% can read anywhere from ~50% to ~66% — see the demo guide before drawing conclusions.
🤖 Get the free trading bot
Open-source bot for Quotex & IQ Option (13 strategies). Edit it with ChatGPT/Claude — no coding. Free, no spam.
FAQ
Does it work on OTC? No — like everything: OTC is ~50% for all strategies.
What about pin bars / long wicks? Tested: 47–49% on M1 (a wick there means continuation, not reversal). Even inverted it went nowhere.
Where do I enable it? In the panel: estrategias = reversao_dupla,exaustao_bb, 120s expiry, USDJPY+GBPJPY, 9–10:59am / 6–11:59pm UTC-3 windows.
⚠️ Binary options carry a high risk of losing all your capital. Educational content based on demo-account data — not investment advice. Past results do not guarantee future results. No bot or strategy guarantees profit.