Manual Strategies

Fractals, RSI leaving 30/70, Bollinger re-entry and the engulfing candle on 16,926 EURUSD M5 bars. Bollinger sat at the real top or bottom 33.8% of the time, three times the random rate, but none of the four won the 10-pip race more than a coin, and all lost 1.04 to 1.70 pips per trade after costs.

An H1 trend filter against slow EMAs on the 5-minute chart, with the same four entries and exits on 16,926 EURUSD bars. The H1 EMA 50 traded within 0.04 pips of an M5 EMA 600. The short M15 EMA 50 did best for three of four entries, and requiring H1 and H4 to agree made results worse.

Four 5-minute entries on 16,926 EURUSD bars, filtered by the M15, M30, H1 or H4 trend. Read from the last closed bar, the filter beat no filter in 67 of 144 setups and made 4 profitable. Read with the unfinished bar’s final close, as repainting MTF indicators draw it, it showed 114 better and 32 profitable.

Keltner (EMA 20 ± ATR), Bollinger 20/2, Donchian 20 and a plain ATR impulse on the same 16,926 EURUSD M5 bars with the same exits. All 50 tests lost after costs. The ATR channel was the best to follow and the worst to fade. As a filter, ATR cut an EMA cross’s loss from 2.60 to 0.22 pips per trade.

Four ATR breakout rules on 16,926 EURUSD M5 bars: impulse, range expansion, Keltner channel and the daily open breakout, with ATR 7 to 50, multipliers 1 to 3 and five exits. Following the break lost before costs in 289 of 310 setups. Fading it was positive before costs in 297, and a 1-pip spread, half an ATR on M5, took most of that.

One entry rule on 16,925 EURUSD M5 bars, traded with a fixed 0.01 lot and with martingale. With 10-pip targets and stops the fixed lot lost $82.27 and never broke; the martingale ran out of margin on 57 of 66 start dates. With 20-pip targets the fixed lot made $39.02 and the martingale $191.73.

Martingale, grid, Asian breakout, EMA cross and RSI return, each run on 16,925 EURUSD M5 bars with a $1,000 account, 0.01 lots and 1 pip of cost. The grid won 146 of 147 baskets and hit a stop-out on 22 of 66 start dates. The martingale ran out of margin on 57. The three robots with a stop never blew up and never made money.

Every Stochastic 14/3/3 cross on 16,925 EURUSD 5-minute bars, traded with a 1-pip cost. Held 30 minutes, the crosses made nothing before costs and lost 1.00 pip after, the same as random entries. 53.9% were reversed within 3 bars. Crosses in the 20/80 zones came after a 4.17-pip move and delivered 0.11.

13,631 matched EURUSD and gold 5-minute bars, June to August 2026. The correlation of bar-to-bar changes was 0.42 on M5, 0.52 on H1 and 0.57 daily, strongest at 12:00 UTC. But gold’s last bar correlated only 0.04 with EURUSD’s next one, and trading EURUSD after a big gold bar lost in 15 of 16 setups after a 1-pip cost.