Backtest · XAUUSD 5-minute momentum candles · Oct 2021 to Sep 2026

Rizki Aditama’s momentum candle scalp, tested on five years of gold

I coded the momentum candle rules from two Sekolah Trading videos and ran them on 1.75 million one-minute gold bars with real spreads. The base rules took 3,625 trades and lost 836.7R. Every variant loses, including his time windows, the second Fibonacci layer and the news filter, and a $200 account at 0.01 lot goes to zero.

Verdict: loses money in the long run

Net result−836.7R3,625 trades in 5 years

Per trade−0.23R95% range −0.26R to −0.20R

Win rate36%claimed 80–90%, on a 1:1 target

Typical stop$1.04in 2023, against a $0.35 spread

$200 at 0.01 lot$0blown by May 2022, 516 trades in

R = profit or loss divided by the stop-loss distance. His Fibonacci setup is 1:1, so before costs a win is +1R and a loss −1R.

Source

What the videos say

Two videos from Rizki Aditama’s channel Sekolah Trading: Cara trading scalping (momentum candle) dari nol sampai mahir (24 min) and Cara Menentukan Momentum Candle Di XAUUSD (10 min). He builds the method on four foundations: capital, strategy, “momen” (timing) and psychology.

The strategy is free and fully rule-based: find a momentum candle on M5 gold, draw a Fibonacci on it, and trade the pullback with a 1:1 target. He says it works best on XAUUSD and worse on forex. He also says his own tiny high-risk scalping account is for “dopamine”, and when he asks himself whether it can work long term, he answers “mungkin tidak” (probably not).

Claimed in the videos

Win rate: 80–90%
Best gains: +385,000%, +7,000%
Viewer account: $100–200
Viewer lot: 0.01
His own lot: 0.1 on any balance
Market: XAUUSD, M5

His 7-step checklist (on screen)

1	3–5 short candles Ada 3–5 candle pendek
2	The momentum candle is relatively long Candle momentum relatif panjang
3	It has closed outside their high/low Sudah close di luar high/low
4	Choose manual or Fibonacci entry Pilih manual atau Fibonacci
5	Fibonacci levels placed correctly Level Fib terpasang benar
6	Total lot and risk calculated Total lot & risiko sudah dihitung
7	Ready to accept the stop-loss Siap menerima stop loss

The Fibonacci map, drawn on the momentum candle

Level	Role	Buy example
−0.27	Take profit, beyond the tip	high + 0.27 × range
0	Tip of the candle	high
0.236	Entry (limit)	high − 0.236 × range
0.618	Optional second layer, same lot	high − 0.618 × range
0.745	Stop-loss	high − 0.745 × range
1	Base of the candle	low

Entry to TP and entry to SL are both about 0.51 × the candle’s range: 1:1.

The method

Step by step, from chart to risk

The routine as he teaches it, in order. The quotes are his own words.

Open gold on M5

M5 is the main time frame; M15 also works. M1 has too much noise and spread, and H1 or H4 is too slow for scalping. Gold only: he says the win rate drops on forex pairs.

“M5 adalah time frame yang utama.”M5 is the main time frame.
Spot the momentum candle

Look for three to five small candles, then one candle that is suddenly much longer. It needs a long body and short wicks; a long wick is not a momentum candle.

“Ada kecil kecil kecil kecil, tiba-tiba langsung panjang banget.”Small, small, small, small, then suddenly very long.
Wait for the close

It only counts once the candle has closed outside the high or low of the small candles. While it’s still forming, it can turn back and leave a long wick.

“Valid kalau candle close di luar range.”Valid when the candle closes outside the range.
Follow its direction

A bullish momentum candle means buy; a bearish one means sell.

“Ketika market itu naik, kita ngikuti aja.”When the market rises, we just follow it.
Draw the Fibonacci on the candle entry 23.6 · SL 74.5 · TP −27

Draw it in the candle’s direction (low to high for a buy). Place a limit order at 23.6%, the SL at 74.5% and the TP at −27%, beyond the tip. The alternative “manual” entry is a market order when the candle closes.

“Iya, ini area entry, area SL, area TP. Enak kan?”Yes, this is the entry area, SL area, TP area. Easy, right?
Optionally add a second layer

A second limit at 61.8%, with the same lot as the first, the same SL and the same TP.

“Lotnya harus sama, jangan 0,01 tiba-tiba 0,05.”The lots must be equal; don’t do 0.01 and then 0.05.
Pick the moment, avoid the news

The best setups come when a quiet market suddenly speeds up, which in November was around 08:00 and 22:00 Jakarta time, for about two hours. Never trade the news itself; wait 30 minutes after.

“Setidaknya 30 menit setelah news is oke.”At least 30 minutes after the news is fine.
Get out fast

A scalp lasts five minutes to an hour. Longer than that is intraday or swing trading.

“5 menit, 10 menit, mentok 1 jam sudah selesai.”5 minutes, 10 minutes, an hour at most, and it’s done.
Size small, cut losses, withdraw profits

Viewers should start with $100–200 at 0.01 lot and count progress in percent per month. Take every stop, stop trading after a stop-loss, withdraw profits and keep the same lot.

“Kalau udah terkena SL, ya wis jangan dilanjutin.”Once the SL is hit, that’s it, don’t carry on.

Implementation

How each rule became code

“Small” and “long” need numbers before a computer can test them. I picked one base reading and tested the others as variants.

Video rule	What the backtest does
Momentum candle	An M5 candle whose body is at least 2× the average range of the previous 5 candles, at least 1× ATR(14), and at least 70% of its own range, closing beyond the previous 5 candles’ high (buy) or low (sell). Variants: 3 candles, 3× size, M15.
Direction	The momentum candle’s colour.
Entry, SL, TP	Limit order at 23.6% of the candle, SL at 74.5%, TP at −27%. Setups whose own wick already passed 23.6% are skipped. Variant: market order at the close with the same SL and TP.
Second layer	Extra limit at 61.8%, same lot, SL and TP. The pair is scored in units of the first layer’s risk.
Scalping time	Order and trade live at most 60 minutes after the candle closes. Variants: 30 minutes, 4 hours, and only his 08–10 / 22–24 Jakarta windows.
News	No setup within 30 minutes of a high-impact USD event (4,165 events from the MT5 calendar). Variant: no filter.
Risk rules	One setup at a time. Variant: no more trades that Jakarta day after a stop-loss.
Costs	Real per-minute gold spread from an MT5 export (about $0.35). On the minute the order fills only the SL is checked, and if SL and TP fall inside the same minute, the SL counts. The optimistic reverse is tested too.

Base rules

Results

Cumulative R, trade by trade

Shaded area: out-of-sample period (Oct 2024 on)

The curve never gets above zero. It falls steeply until 2024, then more slowly as gold’s candles grow larger and the spread matters less. It flattens out in 2026 but doesn’t recover.

How setups ended

Outcome	Setups	Share	Avg
Limit never filled	611	16.9%	0R
Hit TP	995	27.4%	+0.99R
Hit SL	1,814	50.0%	−1.00R
Closed after 60 min	205	5.7%	−0.06R

Filled trades: median hold 8 minutes.

Tuning period vs check period

Period	Setups	Win	R/setup	Total
Oct 2021 – Sep 2024in-sample	2,241	31%	−0.314	−704.5R
Oct 2024 – Sep 2026out-of-sample	1,384	44%	−0.096	−132.3R
All, optimistic fills	3,625	37%	−0.216	−781.9R
All, spread ×2	3,615	28%	−0.346	−1,250.9R

Net R by year

Net gainNet loss

Show the numbers

Year	Setups	R/setup	Net R

Diagnosis

The spread eats small setups, and big ones still don’t win

The stop sits 0.51 × the candle’s range from the entry. For most of 2021–2023 that was about $1, while every trade pays around $0.35 of spread, roughly a third of the risk. As gold climbed, M5 candles grew and the typical stop reached $6.44 in 2026.

R per trade by stop size

GainLoss

Typical stop by year

Year	Median stop

With costs small (stop $3 or more), the setups come close to breakeven but stay negative: 1,039 setups at −0.051R each, 95% range −0.10R to +0.00R. Ordinary candles of the same size, with no momentum conditions, lose −0.089R per setup. So the momentum pattern is worth about +0.04R over a random candle. That isn’t enough to cover a 1:1 target’s costs.

Robustness

What it does to a $200 account

Equity at 0.01 lot, starting Oct 2021

Line at $200: starting balance

At his recommended 0.01 lot, $200 is gone by 25 May 2022, 516 trades in. Starting in January 2025 instead, when gold’s candles are larger and the spread matters less, it lasts until 3 September 2025 (456 trades): bigger candles mean bigger stops in dollars, so each loss costs more. At the 0.1 lot he uses on his own small account, $200 lasts 63 trades from October 2021, or 50 trades from January 2025.

Diagnosis

Why it doesn’t hold up

A 1:1 target on small candles can’t pay the spread. Stops under $1 lost −0.60R per trade, with only 20% winners.
The pullback fill picks the fading moves. Even big setups win only 46% of the time on a symmetric 1:1 target. When price comes back to 23.6%, the burst has often run out.
The second layer adds size when the trade is going wrong. Layer 2 only fills when price has retraced to 61.8%, close to the stop. Two layers lost 1,166R against 837R for one.
His timing and news rules don’t change the sign. His momen hours: −0.20R per setup over 556 setups. Without the news filter: −0.22R.
Stopping after a loss halves the damage only by halving the trading. Per setup it is still −0.21R.
Forex is far worse, as he says. The same rules on 9 major pairs: 18,621 setups, −0.54R per setup.

Forex side test by pair

Pair	Setups	R/setup	Net R

Conclusion

Verdict: not applicable as a long-run system

Taken as taught, the momentum candle scalp loses about 0.23R per trade on five years of gold, and none of the 14 versions tested makes money. Most of the loss is spread on small candles. Once that is stripped out by trading only big setups, what remains is roughly breakeven minus costs: the pattern beats a random candle by a little, not by enough.

This matches his own caveat about his small scalping account: when asked whether it can work long term, he says “mungkin tidak” (probably not). The 80–90% win rate and the large percentage gains can’t be reproduced from the rules in these videos.

If you still want to trade it

Check your real spread first. At $0.35 a trade, any setup with a stop under about $3 starts deep in the red. His chart shows a different broker; a lower-spread account helps, but even at near-zero cost the big setups only break even.
Forward-test on demo for at least 100 trades and log every momentum candle you took or skipped, so your judgment calls can be tested.
Never trade 0.1 lot on $200. In this test that account lasted 50–63 trades.

Caveats

Limits of this test

“Small” and “long” were fixed as numbers (2× the previous 5 candles, 1× ATR, 70% body). A trader who skips marginal candles by eye could do better; the 3× and $3-stop variants approximate a stricter eye and still lose.
The Fibonacci is anchored on the momentum candle’s high and low, as in his drawings and slides.
Spreads come from one broker’s MT5 export; his own broker and live fills may differ. The 2× spread and optimistic-fill tests show how much this matters.
He sometimes enters before the candle closes, layers more than twice, or closes by hand. Those calls aren’t modelled.
Only gold was tested in full; forex was a side test.

What the videos say

His 7-step checklist (on screen)

The Fibonacci map, drawn on the momentum candle

Step by step, from chart to risk

Open gold on M5

Spot the momentum candle

Wait for the close

Follow its direction

Draw the Fibonacci on the candle entry 23.6 · SL 74.5 · TP −27

Optionally add a second layer

Pick the moment, avoid the news

Get out fast

Size small, cut losses, withdraw profits