ICC pullback (USDJPY)

AUDJPY, GBPJPY, USDJPY · 1H chart · strategy s29_icc_pullback · 4 variations

Si's ICC idea on USDJPY: when a 4-hour candle closes above the last swing high, buy the pullback halfway back down the move, with the stop under the last 4-hour swing low.

What it is about

ICC = Indication, Correction, Continuation, taught by "Si" in a 14-video course. The indication is a candle CLOSING beyond the last swing: the market has shown its hand. Price then usually corrects part of that move before it continues, and the trade is taken in that correction.

His own entries wait for a confirmation candle on a lower chart. Our tests found a plain limit order halfway back down the move beat every confirmation and rejection candle (engulfing, pin bar, inside bar, rejection wick), so the order simply rests there.

Tested on 7 markets with about 5 million settings. Only USDJPY buys beat random entries with the same stops and exits (90-100% of the time); on gold and the indices the profits were just the market rising. It also passed on the yen crosses AUDJPY and GBPJPY.

How it trades

Chart & market

USDJPY (also AUDJPY and GBPJPY) on the 1-hour chart, with swing points from the 4-hour chart. Buys only.

Setup

A swing high on the 4-hour chart is a candle body higher than the 2 bodies on each side, known 2 candles later. The setup starts when a 4-hour candle closes above the last swing high (the previous close was at or below it).

Entry

From the next 1-hour candle, a buy limit sits halfway between the broken swing high and the highest price of the move since. As the move makes new highs the order moves up with it (using the high known before each candle). The first candle that touches it fills the order, at the level or at a lower open. The order is cancelled after 72 hours (~3 days) or when a new 4-hour indication appears.

Stop loss

The last 4-hour swing low. The distance from entry to stop is 1R (R = the amount risked). If the fill candle also reaches the stop, the trade counts as a loss. Trades with a stop smaller than 0.15 x the 1-hour ATR or smaller than 3 x the spread are skipped.

Exit

A: no target; once the trade is +1R the stop follows 1R behind the best price. B: a fixed target at 10R. Either way the trade is closed after 120 hours (~5 days) at the latest.

Filters

One trade at a time. No trend filter: a Daily EMA 50 filter made it weaker. Sells are off: they lost in every version tested.

Any time of day, Sunday evening to Friday. About 4 trades a month per account; trades are held overnight and up to 5 days.

What to expect

A wins about 6 times out of 10 with small winners; B wins about 4 times out of 10 and needs the occasional big run.
Backtest at zero cost, 2018-23 / 2024-26: A +81.3R / +54.0R (largest drop 7.5R / 4.2R), B +90.2R / +50.4R (largest drop 13.2R / 12.1R); 2026 so far A +17.0R, B +22.4R.
A made money in every calendar year from 2018 to 2026; B lost a little in 2019.
On AUDJPY and GBPJPY the same rules earn less (about +53R to +64R over 2018-today) and lose in some years.

Risks and caveats

Only one market (USDJPY) and one direction (buys) survived the tests. That can be luck from testing many markets, and it is a buy rule on a pair that rose for most of the period.
Si does not trade currencies himself; the USDJPY version is our finding, not his.
Trades stay open for days, so overnight financing (swap) applies; it is not included in the backtest.
A and B take the same entries; they are two ways to manage one trade. The AUDJPY and GBPJPY accounts often trade the same days, since all three are yen pairs.

Where it comes from

From the ICC course by "Si" ("Trading Course Day 1-14", 14 YouTube videos) that Victor shared on 2026-10-23. The indication / correction / continuation idea and the 4-hour -> 1-hour pairing are his; the halfway limit, the swing stop, the exits and the market (USDJPY buys) were chosen in research and checked against random entries. Paper accounts from 2026-10-25.

Main variation: s29-A ★ Main

Live result

+1.8R

1 closed · 1 open · zero cost

Backtest total

+135.3R

407 trades · +0.33R a trade · win 59%

Backtest max drawdown

−7.5R

total ÷ drawdown 18.2

Years won

9/9

3.9 trades / month

Backtest R per year (zero cost), s29-A

The variations

Each variation is the same strategy with a different approach, trading its own paper account. Click one to see exactly what it changes.

s29-AUSDJPY, trail 1R from +1R★ Main

The default: halfway limit, stop under the last 4-hour swing low, and from +1R the stop trails 1R behind the best price. Many small winners, the best profit for its drawdown.

s29-A-AUDJPYAUDJPY, trail 1R from +1R

A on AUDJPY, with the minimum stop set to 3 x that pair's spread. Added after it beat random entries in every run and 98% of nearby settings were profitable in both periods.

s29-A-GBPJPYGBPJPY, trail 1R from +1R

A on GBPJPY, minimum stop 3 x its spread. It passed the same checks as AUDJPY but with a thinner margin.

s29-BUSDJPY, 10R target

Same entries as A, but no trailing stop: a fixed target at 10R, so the trade waits for a big trend leg or the 5-day limit.

Variations side by side

Live result of all 4 together: +1.7R (zero cost).

			Live paper account (R, zero cost)				Backtest 2018 – today (R, zero cost)

s29-A★ MainICC pullback	USDJPY	1H	+1.8R	1	100%	1	+135.3R	+0.33R	9/9	3.9	59%	−7.5R
s29-A-AUDJPYICC pullback	AUDJPY	1H	+0.1R	1	100%	1	+64.0R	+0.15R	7/9	4	54%	−8.7R
s29-A-GBPJPYICC pullback	GBPJPY	1H	−2.0R	2	0%	1	+53.4R	+0.12R	6/9	4.2	54%	−13.0R
s29-BICC pullback	USDJPY	1H	+1.8R	1	100%	–	+140.6R	+0.39R	8/9	3.4	42%	−13.2R

Main = the variation with the most backtest R per R of max drawdown (2018 – today, zero cost). It is picked on that history, so the live accounts are the real test.