What are the most discretionary elements when you trade or you can think of, which are the hardest if not impossible to backtest and make it fully systematic ?
Discretionary is an interesting term, in trading. Actually, in trading, everything has to be reproducible, scientifically reasonable, etc. Then there are traders who say discretionary and price action. So, what does that mean?! I think, when you have a trading technique, TA, volume, etc (anything in the chart), You always trade in some way discretionary as well, as no trade is the same. Just if you have very clear TA rules you obey 100%, disc. is not part of your trading. But honesty, who does that? Discretionary means to get into the trade early, before a TA signal is being triggered. Discretionary means you are good at short term trading, or scalping. But actually, a trader needs to trade discretionary, to be ahead of the masses. TA with clear entry rules can be back tested, but there is no algo/expert advisor that works successfully enough on TA only. - otherwise we wouldn't have traders. News, volume trading in the short term is discretionary trading, if you will. You obey rules you learned, but you always adopt to what you see (and feel) as well.
One simple example I can think of is the breakout of a level. If you look at index futures for example, you can have either have a straight away successful breakout that never looks back or you can have an initial breakout which is reversed and then chops around a bit before either reversing or continuing. I would think this is very difficult to code, i.e., make clear rules for what's a successful breakout and what's a false breakout. There are other key levels I like to use for clues intraday. Often they work to the tick. Other times they will overshoot a bit. Some times price will linger there and zig zag a bit before taking off. Other times it's a straight away reversal. This is also stuff that seems difficult to code precisely and where I think discretion is required or useful.
Fundamental data (weather, inventory, interest rate, financial, etc), geopolitics, government policy, sanction, etc. Basically anything that obviously influence the price in the long term. You need to have vast knowledge, experience and good sense to interpret all these variables correctly. It's pretty subjective and the reason why I solely use price for trading. I only use all those subjective-to-interpret variables for investing.
Context 100%. There are recurring patterns and setups that present themselves over time on charts - but they won't be of any benefit to you if they aren't utilized under the correct conditions. And guess how one excels in identifying context? Yep - screen time. The thing you can't magically have handed over to you and not something that can be coded as far as my experience goes.
In my opinion, discretionary trading would be anything that is not automated trading from entry to exit. Also, rule base traders (not automated) are discretionary traders too. wrbtrader
One definition I have seen of discretionary trading is: “Before we delve into the specifics, let’s start with a clear definition of discretionary trading. In a nutshell, discretionary trading refers to a trading approach where decisions to enter or exit trades are made based on the trader’s own analysis and judgment, rather than relying solely on predetermined rules or algorithms. As a discretionary trader, I have full control over my trading decisions, and I am not bound by rigid trading systems. This flexibility allows me to adapt to changing market conditions and take advantage of unique opportunities that may not be captured by automated trading strategies.” Source: https://www.morpher.com/blog/discre... trading,on predetermined rules or algorithms. Therefore, I Volpri would submit that if it could be coded and automated and made very systematic then it likely is not discretionary. There are just too many variables and possible judgement calls, in and to make in multiple environments, to fully automate the process, if that is what you are getting at?? I can’t see coding as interpreting new scenarios live and making decisions on the fly. Maybe someday automated learning may do that but presently! in my opinion, a good disciplined discretionary trader is leaps and bounds more ready for changing market conditions and adaptability than algos. Algos will work until they no longer do. The human brain is much more powerful in many ways than say a computer.
Depending on your programming skill, everything can more or less be programmed. However, the problem is this: Is what you programmed profitable? If what you programmed isn't profitable, what good is it? That is, my friend, the most discretionary element of trading.
Suprise can’t be backtested. Anything psychological can’t be backtested. Well … if you don’t have (enough) data … Then it cannot be backtested. If you don’t know the algorithm (discretionary decision making?) then it cannot be backtested.