Rincian backtest

EA: ea-tokyofix-audjpy-m5 / 0.1.0 / 0.1.0|20260915T075705Z
Transaksi
210
Faktor profit
1.25
DD maks. %
0.32
Laba bersih
17.2
Transaksi per tahun
23
Rentang pengujian (UTC)
2010-01-01 2018-12-31
Lama: 9.00 tahun
Simbol / Time frame
AUDJPY / PERIOD_M5
Pemodelan: EveryTick · tick asli di 0% rentang uji
evidence di luar data penyetelan disetel pada 2019-01-01 → 2026-09-05
Data eksekusi
Candle: 666,826 Ticks: 238,486,872
Catatan pengujian
OUT-OF-SAMPLE 2010-2018 for the AUDJPY candidate, same pv 0.1.0 set (holidays off, as filed). The 2019-2026 window gave 506 trades PF 1.45 +$53, half the per-trade edge of USDJPY/EURJPY/GBPJPY. Headless run; generated ticks. | Out of sample the effect is present at $0.08/trade (in-sample $0.10): every year with trades positive (2013 +$2.9, 2014 +$9.0, 2015 +$0.9, 2017 +$4.0, 2018 +$1.3). Broker AUDJPY M5 history has holes: nothing in 2012 and 2016, thin 2010-2011 and 2015. PF >= 1.0 on both windows -> tier proven (the weakest of the four fix pairs; not live).
Semua kolom
Nilai apa adanya, seperti tercatat di laporan MT5 dan saat diimpor.
RowKey 0.1.0|20260915T075705Z
Versi EA 0.1.0
Simbol AUDJPY
Time frame PERIOD_M5
Awal pengujian (UTC) 2010-01-01
Akhir pengujian (UTC) 2018-12-31
Total transaksi 210
Faktor profit 1.25
Laba bersih 17.2
DD maks. atas saldo % 0.32
DD maks. atas ekuitas % 0.36
Candle 666,826
Ticks 238,486,872
Kualitas pemodelan % 0.00
Catatan pengujian OUT-OF-SAMPLE 2010-2018 for the AUDJPY candidate, same pv 0.1.0 set (holidays off, as filed). The 2019-2026 window gave 506 trades PF 1.45 +$53, half the per-trade edge of USDJPY/EURJPY/GBPJPY. Headless run; generated ticks. | Out of sample the effect is present at $0.08/trade (in-sample $0.10): every year with trades positive (2013 +$2.9, 2014 +$9.0, 2015 +$0.9, 2017 +$4.0, 2018 +$1.3). Broker AUDJPY M5 history has holes: nothing in 2012 and 2016, thin 2010-2011 and 2015. PF >= 1.0 on both windows -> tier proven (the weakest of the four fix pairs; not live).
Kalau transaksinya sedikit, faktor profit belum bisa dipercaya: bandingkan beberapa hasil backtest dulu.