I'm not informed about forex so no comment on your second reason. But your first reason was already addressed in my comment. I said "with attention to common biases, left a safe margin of error..." This means that for example not using closing prices in your backtests but the middle price in the inside spread, taking into account commissions, per-trade and quantity costs in your P&L formulae, etc. It's not too hard to make your simulation environment realistically similar to the live environment unless you are trading on very small time frames.
It can be hard...I'm not saying it can not be...but there are doable setups for retail traders if they know what they are doing. My own strategy is a simple moving average crossover on 5-min time frame with some simple rules to avoid whipsaws. I'll never dare step into anything shorter term than that with my current limited technology because that's when things can get out of control quickly. But at five min interval, there is enough time to take control manually if things break down. As an example, my code will send me a text message when account balance falls more than x%, refreshed every second or so. Worst case is I log into the mobile app or call brokerage wherever I am and flatten everything if I feel confused. One can even put server-side emergency conditional orders that can go off automatically. Again, it boils down to having margin of safety that is big enough to offset any live-simulated discrepancies. If your strategy is making 15% annually with 20% max drawdown in simulated environment, yes, it's madness to implement it live. But if those numbers are instead 200% and 15%, even if things are twice as bad in live trading than in simulated trading, you will be left with 100% return and 30% max drawdown in live...not too bad. Btw, "Margin of safety" was either coined or popularized by the great Ben Graham.
If you figured out how to trade off 5 min chart with ema all I can say, is wow. I ran countless back tests over a decade testing all usual shit: ema, bollinger, rsi, Macd and etc nothing showed positive expectancy in the long run. Good luck
Markets change, boys & girls! Let's get over it. Let's get over any fantasy that markets will/should behave consistently in the present or into the future as they might have done in the past. Ain't gonna happen. Harvest what you can, when you can, and then get the hell out. T/A tells you what just happened. How likely is it to continue into the future? That is a statistical question, and *not* a question of fairness or justice or methods. This round&round reminds me of "Cut your losers short; let your winners run." This is a formula for both trend exploitation *and* risk control/position maintenance. If you put up a chart (or a data array) of 1min data going 4 hours along, and get a strong trend "indication" (from whatever set-up you care to name), and you expect it to last more than 5 minutes going forward -- you're gambling, not trading. Similarly for set-ups of 1hour/48hour or 1day/3month or whatever else: "In the long run," trends will change: that's a fact. If you can step in, harvest what you can, have risk management pull you out when (historically) deemed necessary? That's joy. That's (earned) money. To step in and expect no change? To look back and expect no change? That's foolhardy and a waste of time. ...Just not that hard: "Enter on a shout; exit on a whisper." Let your winners run; cut your losers short.
Well, duh. There's nothing to chew on here. Anything that is publicly available, including many of his own, is no longer a secret and will become a prey to AI. So the takeaway is to not depend on others' code and never ever publish it in the public domain.
%% Exactly right on gamblers .Sorry I don't know Mr Chan.........................................................
My take on all this is that people want to go straight to algo trading. Probably going to be a lot better to learn to trade then develop an expert system based on what you do.
The simple answer is it works. The secret is, if you enjoy the search and exploration, you will definitely get there .