Wow, you are a legend. Can multicharts backtest from / off trendlines ?

327 views in 6 hrs with 33 likes. Thats got to be a record. Ppl must luv doji's. Well done.

Great content as always Jarrod. I think could be very interesting a video about reversing rules in strategies that have "nice and smooth" negative equities.

Thanks!

Great video! It's always interesting to see different strategies being tested and compared. Have you tried this strategy on other markets besides Forex? How did it perform?

Not the Yankee Candlestick that I expected but as always Jarrod, excellent strategies and great charts. 👍 Really finding my trading style now thanks to your videos. ☺️

Hi, Jarrod. Why not test other intraday time frames? I noticed that Oxford's website doesn't indicate a time frame of strategies. Or Doji pattern strategy is only suitable for daily or 1440minutes time frame?

I don't think that the backtesting is sufficient. here is why: you test practially all available data, and therefore you search for a best fit for all currently known data. But, markets vary, and therefore a future market may not resemble a past market. So you should split your testing up, to see if past performance is roughly equal to more recent performance. kinda like in machine learning, where you have a train and a validation test set. if performance on the past is better than on recent data, its just overfitting. secondly net profit could just be a random flucuation. you'd be better of taking samples of all available trades, compute net profit of those, do it again and again, and see how the net profit fluctuates. the more it fluctuates, the worse the performance. Sure you can look at the equity curve, but a) it is subjective to a degree, b) could a random system have an equally impressive equity curve just by chance and c) using statistical methods you can automate this, have an objective measure, and do not need to look at equity curves. Historical price data can be very biased. which means that, if we do not try to take measures to reduce the influence of said bias, could make a system look way better than it actually is. But this bias tends to change over time, which would show in a trading system performing well on historical data, but not on live data. And lastly, you should always compare a system with a random system on the same data. if it is similar, or not outperforms the random system with a p