A lot has changed since the last update, and I have been less active with documentation so I will put out a short update here today.
I have decided that the elephant bar algorithm is not good enough to continue running, as just about every market is down 1-2k at this point and shows no promise. Instead, I have gone with a 3-straight 5 minute candle algorithm, that goes long if the 5m has 3 bullish candles while the 1H is above 0 on the MACD and the opposite for shorts.
Note that this currently still skips a large bullish candle if that is the entry candle, but the first 2 candles can contain a 30+ pip bar and still be entered. This could be a flaw that needs further analysis.
For now, I have left the algo running on just USD/JPY, EUR/USD, GBP/USD, and USD/AUD. I will let USD/CAD run on the inverse of the elephant bar algorithm just as an experiment.
To optimize this new algorithm, I am using 33/23 and 33/22 for GBP and JPY’s respective reward/risk ratios, as these showed promise in backtesting with about 44% returns. GBP/USD was significantly better going long, though this might just be the current market condition (47 vs 41%), and I don’t think it’s wise to let it run on long-only until there’s more testing done. On that front, I will run a few more backtests for 500k and 1m candles for GBP/USD just to be sure. 47% returns on 33/23 would be incredible if it does happen to stand up to a longer backtest.