TTrades' Fractal Model (4H to 15m) on Gold

Method

4H to 15m, with daily bias. A higher candle sweeps the previous candle's extreme and closes back inside it, or a failed one is engulfed, or a strong next candle continues it; a change in the state of delivery on the lower timeframe confirms it. In the first half of the next higher candle, after it has traded against the direction, the first new change in the state of delivery gives a market entry at the next bar's open. The stop is at that candle's extreme so far, the protected swing, and the target is 2R. No time exit and no session filter. Every rule was written down before the test and nothing was optimised.

Data: the broker's MetaTrader 5 bars, 1 October 2021 to 26 October 2026. A signal fills at the next bar's open; a bar that reaches both stop and target counts as stopped.

The test is a rolling walk-forward with five folds; only trades in the test windows count. Results are in R, the amount risked per trade, before trading costs.

This is one of 75 tests. With that many, three or four can look good by chance alone.

The Fractal Model is taught by the trader and educator TTrades, on ttrades.com and on YouTube. The rules here are IDFX's mechanical reading of what TTrades teaches publicly, written down before the test. TTrades reads context by eye, including correlated markets, and has not seen or endorsed this test.

Verdict: reject. Not this time. Out of sample, this reading of the Fractal Model averaged -0.338R per trade over 68 trades on XAUUSD, -23.0R in total, and 0 of 5 test windows made money. All figures are before trading costs. This is one of 75 tests; the series page shows them all side by side.

Run run_fbc49172e5, IDFX engine, 30 October 2026. The sample-trade review was deferred by the owner; the trades were checked candle by candle instead.

The idea

Price cannot turn without a swing, so number the candles around one: candle 2 is the swing, candle 3 should expand away from it. Let the wick form, then trade the body. The same picture repeats on every timeframe, which is why it is called fractal.

The Fractal Model is taught by the trader and educator TTrades, on ttrades.com and on YouTube. The rules here are IDFX's mechanical reading of what TTrades teaches publicly, written down before the test. TTrades reads context by eye, including correlated markets, and has not seen or endorsed this test.

Data and test

• XAUUSD (gold), M15 bars from the broker's MetaTrader 5 feed, 1 October 2021 to 26 October 2026.
• Rolling walk-forward, five folds; only the test windows count. Every rule was fixed.

Results (out of sample, before trading costs)

The verdict

Reject. Rejected: expectancy of -0.338R per trade before trading costs; 2% probability the true Sharpe is above zero (one fixed rule set was tested, so there is no search to deflate for); p = 1.000 against block-bootstrapped price series. Also failed: sample size, fold consistency, robust to resampled trades.

• Sample size: failed. 68 out-of-sample trades; under 100 the statistics are fragile
• Positive expectancy: failed. expectancy of -0.338R per trade before trading costs
• Walk-forward efficiency: not applicable. not applicable: every parameter is fixed, so nothing was fitted on the training windows and there is no fit to decay; the fixed rules are judged by their out-of-sample expectancy and fold consistency
• Fold consistency: failed. 0% of walk-forward folds were profitable
• Sharpe significance: failed. 2% probability the true Sharpe is above zero (one fixed rule set was tested, so there is no search to deflate for)
• Beats shuffled price: failed. p = 1.000 against block-bootstrapped price series
• Robust to resampled trades: failed. 1% of trade sequences resampled with replacement end profitable
• Survivable drawdown: passed. 0.0% chance of halving the account at the configured risk
• Drawdown: passed. maximum drawdown of 24.4%

Reshuffled price series scored as well or better in 40 of 40 tries (p = 1.00).

75 tests. This is one of 75 run together, so no single one should be read on its own.

Limitations

• Before trading costs. Spread, commission and slippage would make the result worse, not better.
• One of 75 tests; a single good-looking result among them can be luck.
• The out-of-sample result covers about 3.3 years, not the whole five.
• A mechanical reading. TTrades reads context by eye and uses correlated markets; TTrades' own trading can differ.
• CFDs on one broker's feed; TTrades trades futures. A candle's sweep can differ between feeds.
• A backtest, not live or funded-account trading.