We don't exactly know when new rally starts (sure we can use some patterns for identification of probability of starting new rally). Also we can't exactly know that rally which we joined still will go on. So this is talk about probability of "working" some patterns. And statistics of ATR of particular instruments. Or you mean something else?
Why compare with ATR. You want to find a favourable risk/reward ratio. If your set-up loses 100% of the time then it doesn't work. If it works 50% of the time and has a 1 to 2 risk to reward ratio then you make a profit. If your system isn't working then you have to determine why; is it a poor system or poor execution.
May be beneficial to think a bit more of the metrics that are helpful and well understood in quantifying strategies. This can aid in reduction of noise (nonsense) in your quest for improved strategies. You may find you prefer a number of metrics to properly quantify a strategy. I like Profit Factor, Kelly Percentage, overall return, to name a few, as well as balance of sequential wins to sequential losses, etc. Only you will be able to properly determine your comfort with max drawdown as limiting it will typically impact overall return, but is also not to be ignored. The set of metrics to quantify a strategy will likely vary with the intended sizing of the trades, so small size trades may tolerate more "shoot from the hip" set of metrics, where your large sized trades may be restricted to a very well thought out set of metrics and back testing. For larger sized trades, the metrics should continue to be reliable out of sample.
. Been testing strategies for many years now and my experience has indicated that one can still use different combinations of indicators to create profitable strategies. There appear to be lots of moderately profitable strategies that have around 50% wins and profit factors of about 1.5 (meaning that one's gross gains from winning trades are 1.5 times higher than gross losses from losing trades). I swing traded strategies like this off of daily charts for years. I would mention that real world profitability was never quite as good as the backtesting results. Also, my backtesting always includes at least 3 futures markets tested over multiple year periods and only systems that gave acceptable results in all three markets were used for real world trading. Despite these precautions, a couple of strategies flopped in real world trading and had to be discarded, while others survived. Strategies with higher win rates and consistent profit factors of 2 or more have been more difficult to design, but these systems exist as well. I know it's a cliche, but it really is better to design systems that trade with the current trend. Whether one is an indicator-based trader or a price action trader or support/resistance trader or market profiler or Elliott waverider or whatever, only taking the entry signals produced by those methods that are in the direction of the current trend is likely to improve the win rate and possibly make an inconsistently profitable system into a consistently profitable one. Just to go beyond the cliche, a couple of specific suggestions are shown in the picture below. Consider designing entries/exits that only buy when prices are above a long term MA or only when the slope of a shorter term MA has been up for X price bars, or both. To put it another way, don't buy when price bars are below the long term MA and/or don't buy when the slope of a faster MA is against you. Opposite rules for sell signals. View attachment 243613
Thank you for advice. I compare with ATR to understand volatility and avoid of not accurate SL. I didn't say that my strategy doesn't work but I'm looking for new ideas. And tested a lot of examples I found that having small SL is worse for executions and final result in comparison with bigger (for intraday trading for most liquid futures end stocks , include: ES, NQ, 6E, est). So idea: having small SL and big take profit usually doesn't work due the market noise.
Thank you for message. Profit Factor, Kelly percentage est are just methods how to check some options of strategy. Kelly percentage might develop final results in case of good management of size and risks. But they are no able to make strategy profitable if it is not. Or I lost something?
Thank you for answer. I agree with your opinion. Trend Following system (especially when we don't have particular Take profit) always show better results than trading in range because length of price movements in trends mostly much bigger than in range. Using any indicators (like MA, levels est) as filters of directions might be used, but also that don't give good predictable probability but give bigger average profit. It's true. Thank you for example and explaining. But on my mind you use MA as filter of direction and perhaps some patterns for entry/closing positions. I also agree that it works if we have trailing SL. And we were lucky enough to take good movement.
Metrics can aid in comparing strategies and modifications to strategies. For some of us, we prefer to understand why something is working as well as why something is not working so we can make better decisions. Each person's preference may vary wildly on what constitutes adequate for them. My suggestion for using metrics that suit you, is so you can more easily discriminate among various strategies and trade tweaks! -- How will you know if a trade/strategy matches your requirements!
Having a strategy that works puts you ahead of a lot of traders. I don't trade futures or intraday so I won't be of any help to you. Good luck in your quest.
Volfix, I think everyone has there own path that suites their personality, educational background, and resource constraints. TomMorton is correct in that he is focused on the essentials. It might be helpful to step back and outline all the parts you need, and then tear them down to the most simplest solutions. Put them together and optimize, never loosing sight of what is important. I bet you can come up with a custom solution that suits you. I bet it could be elegant too. Factiously, I can say, "all you have to do is press the right button at the right time". How hard can that be? How complex you make it is up to you? Lastly, I just stare are ES all day long and AH. Two years down the road, it is pretty easy to see high probability setup that I need. Hope that helps.