After the 仲値 [MT5 version]: a USD/JPY EA that only sells after 9:55

Backtests look clean, but they crumble when actually run. This was made for people who want to avoid such EAs. This EA will perform one retrial within the period defined after deciding the logic (January 2025 to September 2026) and has produced results of the same level as the created period (Profit Factor from 1.67 to 1.68), so it is being offered. Trading occurs only after the Tokyo time rule “Gotō-day mid-rate (9:55).” Once per day, stop loss 20 pips, no averaging, no martingale, no hedging. Past performance was a backtest with initial capital of 1,000,000 yen and fixed 0.5 lots (January 2015 to September 2026), with an average annual return of 11.6% (CAGR 7.6%), maximum drawdown 7.7%, and 11 of 12 years were positive (past results do not guarantee future performance).

What this EA does: It uses Tokyo time behavior that makes USD/JPY rise toward the mid-rate on Gotō days (5, 10, 15, 20, 25, 30 days and month-end) when importers buy dollars, and after the mid-rate is decided, buying pressure subsides and prices tend to rebound. - On Gotō days at 9:55 (when the 9:50 5-minute bar closes), sell USDJPY (1 position only) - Close at 11:30 (time-based; independent of take profit or stop loss) - Stop loss fixed at 20 pips, set at entry - Do not trade except on Gotō days. Do not hold positions overnight - Logic parameters are fixed. Users only set lot size, maximum spread, and magic number - For MT4 users, a version with the same logic is prepared as a separate productTrades: 893 / Net profit +1,365,350 JPY (initial 1,000,000 JPY → about 2,360,000 JPY) / Profit Factor 1.67 / Win rate 56.1% / Maximum drawdown 76,800 JPY (7.7%) / Maximum consecutive losses 6 / Average annual return 11.6% (net profit ÷ initial capital 1,000,000 JPY ÷ 11.75 years simple interest) / CAGR 7.6% (1,000,000 JPY → about 2,360,000 JPY per year) with lot 0.5; if trading costs were 0.5 pips worse each time, average annual return would be 9.7%. Of these, the logic was developed over 2015–2024 (10 years), and after January 2025, only one retrial was conducted. Breakdown is as follows:Simulation of changing trading order (Monte Carlo 1,000 runs) Median maximum drawdown 75,475 JPY, 95th percentile 114,762 JPY (11.5%). Please prepare funds for a drawdown of about 10–12% in actual operation.Required funds and lot size - Initial setting: 0.5 lot (50,000 units). Required margin at USD 1 = JPY 157 with leverage 25x is about 315,000 JPY - Recommended funds: 1,000,000 JPY (same conditions as backtest). If operating with 500,000 JPY, reduce lot to 0.2–0.3 - Lot can be fixed or percentage of fundsSpecifications Currency pair: USDJPY (works even if your account name is “USDJPY-” or “USDJPY.pers”) Time frame: M5 (please open chart in M5) Max positions: 1 / Hedging: none / Averaging: none / Martingale: none Stop loss: 20 pips (set at entry) / Take profit: time-based (11:30) Trading frequency: 6–7 times per month (Gotō days only) Server time: automatic (9:55 and 11:30 Japan time. GMT offset can be set manually) External communication/DLL: none Parameters: Lot / Risk-based lot / Spread limit / Slippage / Magic number / Time difference (auto) Compatible brokers: domestic MT5 brokers (as long as USDJPY can be traded). Not restricted to a specific broker. In netting accounts (where multiple positions on the same currency pair are consolidated into one), please do not run this EA concurrently with other USDJPY EAsThe top-left of the chart displays Japan time, the next Gotō day, positions, spread, and whether automated trading is allowed【Forward】 Live trading has started on 2026-10-03 with a domestic broker (Finatex MT5 demo). Trading results are published as REAL TRADE.【Note】 - Backtest and forward results do not guarantee future performance - During market shocks (economic data, remarks by officials), stop loss orders may be filled at prices worse than the set values - If Gotō day falls on a weekend, trading occurs on the previous Friday. It trades on Gotō days even if it is a Japanese holiday (we do not determine whether it is a holiday). Results vary with broker’s trading hours, spreads, and execution conditions - Support covers installation and operation guidance. It does not provide trading instructions or individual advice - No logic changes or parameter distribution (fixed logic)

Development background: This EA was not made by polishing one idea. It systematically creates a large number of trading rules, tests them all under the same evaluation (2015–2024 backtest for 10 years → re-tested once for a different period from 2025 onward → tests with worsened spreads → rearranged trade order), and discards the ones that fail. More than 170 rules were tested before listing it; even among popular high-win-rate scalping styles, only the rule “rebound after the mid-rate” maintained performance in a different period. The remaining one is this EA. The three parameters of the logic are entry time, exit time, and stop width. Shifting exit time to 10:30–14:00 or stop width to 15–30 pips yields a profit factor between 1.5 and 1.75 over 10 years. Shifting entry time to 9:50 or 10:00 drops to 1.2–1.3. The source of performance is the rule itself, “immediately after the mid-rate is decided,” not minor numerical tuning. Once the product is released, the logic will not be changed. An EA that cannot maintain performance without changing it is discarded without modification.Breakdown 1: Development period used (2015 January–2024 December, 10 years)

Trades 762 (74–77 per year) / Net profit +1,104,850 JPY / Profit Factor 1.67 / Win rate 56.4%

Average profit 6,419 JPY (about 12.8 pips) / Average loss -4,986 JPY (about 10.0 pips) / Maximum consecutive losses 6 / Maximum drawdown 76,800 JPY (7.7% of initial capital) / Risk-reward ratio (net profit ÷ max DD) 14.4 / Average holding time 1.5 hours / Carryover 0 / Swap 0 JPY

Backtest results (January 2015–September 2026, 11 years 9 months, real ticks)Breakdown 2: Period not used in development (2025 January–2026 September, 1 year 9 months)

Annual profit by year (0.5 lot)2015: -2,000 JPY (74 trades) 2016: +84,700 JPY (76 trades) 2017: +143,500 JPY (77 trades) 2018: +14,000 JPY (77 trades) 2019: +72,550 JPY (76 trades) 2020: +172,650 JPY (76 trades) 2021: +55,150 JPY (76 trades) 2022: +205,400 JPY (77 trades) 2023: +116,750 JPY (77 trades) 2024: +242,150 JPY (76 trades) 2025: +207,200 JPY (74 trades) 2026 Jan–Sep: +53,300 JPY (57 trades) (21 months from 2025-01 to 2026-09: 131 trades, PF 1.68, win rate 54.2%, max DD 51,700 JPY)

If costs are worsened (2015–2024): - Trade cost +0.5 pips: net +914,350 JPY, PF 1.53 - Trade cost +1.0 pips: net +723,850 JPY, PF 1.40 - Trade cost +2.0 pips: net +342,850 JPY, PF 1.17 - Fixed spread 1.0 pips (tested since 2010): PF 1.56 → With brokers and time zones that have wider spreads, performance declines. If the spread limit (initial 2.0 pips) is exceeded, no entry occurs.