Joovier Gems London Breakout (coded 10.5-year study)
Neither coded version made money over 10.5 years. In a challenge simulation, about 1 in 5 tries passed, and most funded accounts never got a payout.
▶ Watch the video- Trades
- 1,310
- Win rate
- 35.9%
- Profit factor
- 0.92
- Reward : risk
- —
- Biggest drop
- $87,790
- Net result
- -$49,770
A dash means the video did not report that number.
Two ways to check this test
My breakdown
Why I coded it
Joovier Gems' London breakout is extremely popular, and he teaches it in two videos with slightly different rules. I'd already tested it by hand twice: it passed in 2017 and stalled in 2026. This time I coded both versions and ran them on more than a decade of data.
The rules
- Mark the high and low of the London session, 3:00 to 9:00 AM New York time.
- On the 5-minute chart, a break above the high is a buy. A break below the low is a sell.
- Only trade after 9:30 AM. If the range breaks before 9:30, skip the day.
- Stop just beyond the entry candle. Target 2R.
- One trade a day, on whichever side breaks first. No trades after 11:00 AM.
Where the rules were fuzzy
To code a strategy, every rule needs an exact number. Some of his didn't have one:
- When does the range end? His list says 9:00 AM, but his charts repeatedly show 8:00. I tried both.
- Close or touch? The older video waits for a candle to close outside the range. The newer one enters as soon as price touches it. I tested both, as two versions.
- Breaks before 9:30. Only the newer video covers this. I used its stricter rule for both versions, to be fair.
- The stop in the newer version sits beyond a candle that's still forming, so the same candle can stop you out. I assumed the worst case, and that happened 38 times.
How I tested it
Real 1-minute NQ data from January 2016 to July 2026, one contract, no compounding, with $2.50 per side in commission and one tick of slippage each way.
What happened
Wait-for-close version (the older video):
- 1,310 trades, a 35.88% win rate
- −$49,770, about $0.92 back for every $1 lost
- biggest drop $87,790, longest losing streak 14
Enter-on-touch version (the newer video):
- 1,427 trades, a 24.25% win rate
- −$162,375, about $0.71 back for every $1 lost
- biggest drop $167,020, longest losing streak 20
Without any costs, they still lost $32,860 and $143,225, so fees aren't the reason. Over the same years, simply buying and holding NQ made about $506,890.
The one version that made money
The losses came mostly from very wide mornings. The widest quarter of days alone lost $96,915, while the calmer three-quarters made $47,145. So I added one filter: skip mornings wider than the 20-day median range.
That version made +$28,015 over 531 trades, with a 39% win rate, a $13,120 biggest drop, and 8 of 11 years profitable. But I tested 15 variations, and none of them passed the statistical check for testing that many ideas. Which side made money even flipped after the fix. So it's an idea, not a strategy.
Prop firm results
- $25K: 127 attempts, 22 passed (about 17%). 100 failed on the max loss limit. Only 6 of 22 funded accounts ever got paid, about $6,593 in total.
- $50K: 72 attempts, 16 passed (about 22%). Only 5 of 16 funded accounts got paid, about $11,500 in total. Every funded $50K account eventually hit its max loss.
These use one firm's 2026 rules on older data.
Takeaway
This is the clearest case on the site of a strategy that looks good in a short replay and doesn't survive a long test. My guide on whether YouTube strategies work walks through all three Joovier tests side by side.
The test
- How I tested
- Coded backtest
- When
- 2016-01-04 to 2026-07-14 (~10.5 years)
- Market
- NQ (Databento continuous), 1 contract, no compounding
- Trades
- See named results below
- Risk per trade
- Not said in the video
- Fees
- $2.50 per side + 1 tick adverse slippage each way
Trade log
I have not published a trade log for this test. The numbers on this page come from the results shown in the video.
More from the test
- Instrument
- NQ (Databento continuous), 1 contract, no compounding
- Coverage
- 2016-01-04 to 2026-07-14 (~10.5 years)
- Costs
- $2.50 per side + 1 tick adverse slippage each way
- Wait for close version
- 1,310 trades, 35.88% WR, -$49,770, PF 0.92, $87,790 max drawdown, 14-loss streak
- Enter immediately version
- 1,427 trades, 24.25% WR, -$162,375, PF 0.71, $167,020 max drawdown, 20-loss streak
- Zero fee control
- Still -$32,860 and -$143,225, so costs are not the cause
- Buy and hold benchmark
- +$506,890 over the same window
- Volatility diagnosis
- Widest-range quartile alone lost $96,915 while the calmer three quartiles made +$47,145 (Holm-adjusted p = 0.047)
- Best fix
- Skip mornings wider than the 20-day median range: 531 trades, 39.0% WR, +$28,015, PF 1.118, $13,120 max drawdown, 8 of 11 profitable years
- Statistical caveat
- 0 of 15 tested variants cleared the multiple-testing significance bar, and long/short profitability flips after the fix
- Prop lifecycle 25k
- 127 attempts, 22 passed (~1 in 6), 100 failed on max loss, 6 of 22 funded accounts ever paid, ~$6,593 total payouts
- Prop lifecycle 50k
- 72 attempts, 16 passed (~1 in 5), 5 of 16 funded accounts ever paid, ~$11,500 total payouts; every funded $50K account eventually breached max loss
A note
This uses one firm’s 2026 rules on old data. The 2025–2026 stretch was not a clean separate check.
Where the numbers come from
Checked Sep 22, 2026. These numbers come from the video. They are backtest or simulated results, not live trading, and not financial advice.
Original video ↗