Reducing Trading Strategy Drawdowns with a Logical Stop Loss

[@aurelien221]

Great content as always. Can't wait to see more!

[@Darwinexchange]

More to come! Thanks, Martyn

[@Martian4x]

Hey Martyn, thank you for the work you are putting in.

[@Darwinexchange]

My pleasure! Thanks, Martyn

[@danielracovitan9779]

excellent

[@Darwinexchange]

Thank you! Cheers!

[@benceszabo9706]

Hello Martyn, Although I found this a qualitative content I have a reservation regarding the following: Could you please elaborate why you are using CAGR/MDD while you are using min lot size? As far as I know the logic behind the CAGR/MDD supposes that position sizing is relative to equity. You also mentioned this concept in one of your video dealing with the performance metrics. Thank you for clarifying this!

[@Darwinexchange]

Hi Bence. You are absolutely right, and I know I have stated this myself in past videos. I think the reason I did this was because I knew that once I had implemented the stop loss and would be using risk based position sizing I would of course be using the CAGR metrics and so it provided a comparison from 'pre risk based' to 'post risk based' pos sizing. But if I knew I was going to stick with fixed pos sizing then I would probably used a profit factor-based metric instead. Good call and thank you for picking me up on it :) Martyn

[@benceszabo9706]

​@@Darwinexchange, anytime haha. :) Btw it absolutely makes sense. I was thinking about that probably using CAGR with min lot doesn't ruin the results in this example. A rhetorical question springs to my mind while writing this comment: could it be a way to compare CAGR and profit factor-based values (how can these be standardized)?...it's not worth the time right now to research this but it's an interesting question on brainstorming level.

[@kylemaharaj9563]

Loving this series! The TK crossover signal that you are using in this backtest, did you include the criteria such that the TK crossover has to be above/below the cloud for buy/sell signals? Or are your buy/sell signals here just purely from TK crossovers? The reason is Ive tried setting up the same strategy but getting slightly different results. Thank you again

[@Darwinexchange]

Hi Kyle. Firstly thank you for the feedback. Secondly to answer your question, I took an alternative approach (simpler implementation) and used 'price being above the kumo' as the rule instead of the 'crossover being above the kumo'. I expected the results to be very similar, but this will account for the slight discrepancy you have. Thanks Martyn

[@leonjbr]

Very Nice. Correct me if I am wrong, but I allways read that the stop loss tends to lower the profit. Nevertheless in your video it happens the oposite. Am I right?

[@Darwinexchange]

Hi Leon. The effect of stop losses are very dependent on the type of strategy. For Mean Rev. Strategies the stop loss almost always reduces performance. And it is easy to understand why this is: if you are buying when over-sold them if price goes more over-sold (against the position) then there is even more reason to be in the position. But for trend following it is different. Here if price goes against your position then it could simply mean the prediction of the start of the trend was wrong. So you want to get out. In this scenario sometimes you see an improvement, something you see a degradation and sometimes the effect on performance is negotiable. So you have to consider stop losses using this type of reasoning. Hopefully that makes sense. Martyn

[@leonjbr]

@@Darwinexchange Martyn: your answer makes perfect sense. Now I understand the problem. Thanks.

[@karenkrohn8003]

@@Darwinexchange That was a very helpful answer, thank you!

[@Martian4x]

What are your thoughts on this? I have run all the pairs individually and cherry-picked the ones with better results (9 pairs), some Yen pairs made into the list some didn't, and then run them together for 5 years duration (I couldn't do 10 years bc I have a sh*t PC). I got slightly better results (5.9 CAGR/MeanDD) but something tells me I'm overfitting.

[@Darwinexchange]

Hi Martian. I think you raise a very valid point, and I'm inclined to agree with your hunch about over-fitting. This is the reason why I included some Yen pairs in my test even though they did not perform very well, and also left out two non-Yen pairs that did perform well - my own fear of over-fitting. However, if I notice that these two non-yen pairs perform well for other Ichimoku strategies then I would consider leaving them in, which would hopefully improve results. But yes, I am with you all the way. It is always prudent to be cautious. Thanks for the question Martyn