Are you really making money in the Forex market? - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I'm still actually in favor of automated trading. I can see you're truly dedicated to it yourself.
Yes, for sure.
Quants fail in an inefficient manner.
I have to argue the opposite. A competent quant can write EA code that fully executes in mere milliseconds. This includes all relevant data analysis and trade execution. In contrast, the manual trader has to perceive various data sources by way of manually reading, listening, and/or watching. Unless we are living the realm of the Limitless movie with NZT, the quant's EA is faster... and more precise, by the way.
Some models which are trained on past data make losses when market conditions shift, and they're no stranger to overfitting.
Undoubtedly, some quants will and do overfit their algo's. The solution here is to increase the sample size of the test data. The overfitting argument loses steam where a 20+ year backtest shows consistent and acceptable profit. As I mentioned earlier, adaptive code and machine learning can be used to exploit shifting market conditions.
If they do factor in qualitative context, social media, black swan events...then they might have an edge similar to a manual trader, but any latency in execution will be a big problem.
I haven't delved into coding many scalping EAs. I could make something statistically profitable, but infrequent trading was inevitable.
Although I gave up on scalping (mostly due my own retail grade setup and multiplying transaction costs) decades ago, infrequent trading certainly fits within my definition of scalping. I imagine 2 or 3 trades per week or so, executed at or near a session open. HFT is just not possible without a full-on professional quant setup. My most profitable algo's are MFT swing trading algo's. Ironically, my best MFT swing trading also today averages 2.5 trades per week. I would be hesitant to adopt any strategy simply because it trades more often than another strategy. A natural human desire to rapidly compound a small account can lead to a total loss.
The manual trader has the advantage of instinctually closing trades sooner with any arbitrary take profit. Automating an efficient (and dynamic) take profit is quite difficult if the market conditions can shift in the moment.
Anything that operates arbitrarily in a trading strategy would have me concerned. I prefer to code a fixed "emergency" server-side stop with each entry and then code a dynamic exit condition that floats nearer to current price. If power goes out, internet service dies, pc dies, or software crashes, the emergency stop is sitting on the trade server as a safety. Otherwise, the algo's dynamic exit does its job as intended.
HFT isn't in the realm of retail. Trading millions of trades in a second is not going to work on normal brokers, and the transaction costs would be huge. I would never bother with order block clusters as I deal with normal brokers, and none of that matters when I'm looking to react to the macro trend, or at least a strong momentum. I also like to put down rigid rules and stick by them, but I have to correspond with these rules. The top scalpers in the world are not settling for less than about 400 pips per day, and constantly juggling different sources of information. Try to make that successful with automated trading. There are also ideas that conflict with other ideas. There's risk management protocols that conflict with the trades as a whole. How many times is an EA hitting a tight stop? There are still some things that will work well in automation, but I know from trial and error that many things people do in manual trading work terribly in automation. Bollinger bands is something that people love, but wow, after I put the classic BB strategy into automation, the amount of losses I saw wasn't good enough. Clearly people have a better edge with that indicator in manual trading.
I agree with @Conor Mcnamara and would like to add the following, in response to posts by @Ryan L Johnson ...
Most of the strategies that are discussed online or in books, are usually directed at manual trading, where the trader's own discretion comes into play. Most of these strategies will fail when completely automated.
Strategies that are good for full automation or quantitative trading, are usually not discussed as readily, because there is a much smaller audience for them. A much higher level of knowledge and skill is required for such strategies, especially in mathematics, statistics and programming, but sometimes also in economics and finance.
Most truely active retail traders do not fall into this category. It is a small subset of retail traders, and even less if you only consider the successful ones. So, these types of automated or quantitative strategies are usually not so openly discussed, and it is up to the individual retail trader/programmer to come up with their own.
Also, as clarification, I am not a manual trader. I only use EAs to trade, but I have to give credit where it is due, and consistently successful manual traders deserve my respect.
In order to convert a manual trading strategy into an algo, the trader must not only have the programming skills necessary for conversion of each element of the strategy but must also have "high resolution" logic and analysis skills. Let's reexamine your BB indicator example...
Of course, this is not an exhaustive list of BB observations and we have not even delved into the most obvious indicator--price bars, themselves. And then naturally, the interaction of price bar elements with BB elements must be analyzed.
The point here is that the human eye might be seeing things that the human brain behind it can fail to adequately quantify in the process of conversion to code. For example, simple "close above" and "close below" logic is much too low of a logic and analysis resolution.
Strategies that are good for full automation or quantitative trading, are usually not discussed as readily, because there is a much smaller audience for them. A much higher level of knowledge and skill is required for such strategies, especially in mathematics, statistics and programming, but sometimes also in economics and finance.
Apparently, I have agreed with you in long form while we were posting at the same time.👍
A quote from Investopedia ...
Investopedia is a great finance reference source, verified by industry professionals and academics. It's totally free, and I've pulled math formulas from there for use in coding. Even though I'm not a member, they accepted an article correction from me.
I just want to clarify... In this thread, we've been using the word, "quant," as a synonym for an algorithmic trader (except where I said, "professional quant"). Interestingly, the bold heading in Post #26 seems to do so as well that is, until we click on the Investopedia link and see that "Algorithmic trading" is merely one key aspect within the greater field of "Quantitative Analysis."
Even if a trader has only one straightforward yet profitable manual trading strategy and then precisely, efficiently, and fully automates that strategy; that trader is actually an algorithmic trader--without having to be a professional quant. If that trader already has sufficient trading industry/markets knowledge, all that is likely needed is one book (there's another thread in this forum for that) to become an algorithmic trader.
[L]arger movements...
I can certainly vote for that where long-term volatility is high, e.g., XAUUSD, EURJPY, or GBPJPY.
In order to convert a manual trading strategy into an algo, the trader must not only have the programming skills necessary for conversion of each element of the strategy but must also have "high resolution" logic and analysis skills. Let's reexamine your BB indicator example...
Of course, this is not an exhaustive list of BB observations and we have not even delved into the most obvious indicator--price bars, themselves. And then naturally, the interaction of price bar elements with BB elements must be analyzed.
The point here is that the human eye might be seeing things that the human brain behind it can fail to adequately quantify in the process of conversion to code. For example, simple "close above" and "close below" logic is much too low of a logic and analysis resolution.
Institutions outsmart all of this. No matter how perfectly the code has encapsulated the bollinger bands indicator as a whole, there are other things to deal with:
- Liquidity sweeps
- Iceberg orders and hidden volume
- Stop Hunting
- Quote Stuffing
- Cross-Asset Correlation Trading
A manual trader watching the order book might spot a large bid that repeatedly appears and disappears (spoofing) which an EA focused on Bollinger Bands wouldn't register. They could also notice a sudden price push through a resistance level right before a major data release suspecting stop hunting, while an EA would only react to band rules and price crossings, or a dynamic stop loss that would absolutely not update fast enough. Humans also have the ability to spot patterns holistically which are too difficult to automate correctly. We can identify anomalies that don't fit predefined rules. Basic algorithms rely on rigid rules or statistical thresholds (e.g. Bollinger Bands deviations), which can miss subtle context-dependent patterns.
In terms of automated trading, I am still learning. I haven't delved into tools that use adaptable algos, will do eventually, but as of now I'm sceptical for it to be that easy. The only way I can see profitable trades in automation, is to use a very generous stop loss initially, a stop which is very unpredictable, and then modify it intelligently as the trade progresses. Some people disagree with the concept of not setting stop loss straight away with the trade, but I believe it can be wise to set the stop loss later after a few bars pass from the trade open time.