I understand. I was troubleshooting trading method for my client and suggested adding filter to the method to bring down number of trades and increase profit . It worked and everyone was happy. This Friday he emailed me your post which he misinterpreted saying """ We better increase the number of trades to get bigger profits "" Thanks for clarifying this for everyone who might also misinterpret this.
In your example you're showing 1-2 daytrades per-year. No matter how good they are I think that would be a waste of time. I wasn't planning on posting about edges in this thread however I'll give you the basic process for mining them since you're so persistent. First find out what is going on in the market you want to trade in the timeframe you plan on holding a position. If you want to daytrade with one trade per-day then find out all the different ways the day has played out in the past. ex. trend day, two-way day, reversal day, etc. Once you've done this you should have an idea of which type of day is most common and which is most profitable. Then define something which could be of value to trade one of the market types. An example might be in a reversal day to find out how often the market makes a low of the day in the first 15 min. of the session. If it happens often enough to be of interest then you go on to the next step. Take every period for which the target is found and create a table of outputs with 1 for the target and 0 for non-targets. Then pre-process all the inputs into the target and convert them to binary inputs. (A common mistake is to take open, high,low, and close data -- analog and assume you can find relationships with the target). For ex. yesterday close > day before yesterday close. If found mark the input as a 1 if not present mark it as a 0. Do this for as many identifies as you can. This may present a hundred or more binary inputs leading to the target for each day of the data. Then you'd pass the data into a backprop neural net and have it train on the data. (you'll need to set aside some data for out of sample testing). Once it's trained to hit at least 90% correctly test the NN on the out-of-sample data. If you hit at least 85% correctly then you can do one of two things. If you're a discretionary trader, setup the NN and preprocess the inputs every day and use the net to predict whether tomorrow has the target (in this example the low of the day is within 15 min. of the start of the session). If so use it to trade to the upside as long the net remains 85% correct. If you're a systems trader then go back to the net and look at the weights of the net to see which of the binary inputs were most important in hitting the target. Use the inputs to create a backtestable system based on the patterns. A system might be when xyz pattern exists then buy next bar above the lowest bar as long as the time is within the first 15 min. of the day. Set the stop to one tick below the low. If the system tests profitable enough to be of interest then move on to the next step. Next, take the trades and test them against random trades pulled from the same year (the edge test). Rank the trades versus random for each year of the backtest. If the trades score consistently above the 70th percentile then you can guess you've found a edge-based system. If not, then you have to assume you've found a temporal characteristic in the data that can be exploited for some period of time. If it's edge based then all you need to do is adjust the trades for market volatility and apply a money management strategy. Check the trades on a periodic basis to ensure the edge continues and plan what to do with your next million. If it's not edge based you can still trade it but you need to setup a objective bailout method such as running a monte carlo sim and determining the bailout point to be say the 95% level of the predicted max drawdown point. Your trading would be more defensive using a non-edge based method as well. Maybe you'd split the trade size in half and have a 15 min. or 10% of daily range as a filter to adding the second position (letting the position prove itself) as long as the volatility was large enough to justify the scaled entry.
I haven't given it much thought. I'm just unwinding right now. I think what I'll probably do is go back through the thread to get a sense of what I was doing and then go back to the notebooks to see what info. I can post that I wouldn't mind being in the public domain. Then I'll start posting with some sort of organized method. Eventually I do plan to post some sort of trading plan template that puts many of the pieces together, though I don't think it will resemble the plan I made for others.
I'd like to throw my nickel into this pool but perhaps I'd be creating to much of a ripple. Just to get a better gauge of things, what is the mean number of trades for you acrary? Accordingly, I will rehash thru the thread to become proficient with the level of dialogue in order to contribute to the thread in a productive manner in addition to overlay the analysis with the automated trading paradigms that I run... Kind Regards, MAK!
Great explanation, thank you. I think I mostly understand what you are saying except for this "target" definition. For instance, say you are building an Opening Range Breakout system. What would constitute a target of 1 vs a target of 0? -bulat
Hi Acrary, Thanks for the great thread! I'm still quite new at trading and am learning a lot from your posts. As such, a lot of the guys who reply to your posts are experienced experts and follow quite well. I, on the other hand am a bit slower. Hope that you could consider my request to post more examples, and/or attachments of files that show calculation samples w/ maybe some explanations. Again, this is just for my benefit (if you have the time), since I somehow doubt many others would need such. I've also come across some traders who trade multiple systems that negatively correlate each other (ie. either different markets or same markets but have a system for scalping, reversal, swing that trade simultaneously). Again, being new at this, reading your posts, I now understand the value of having all of them at the same time, though it takes a while to get used to. However, since you've touched on the topic in earlier posts. I'd appreciate it if somewhere down the road you get the explain in detail how to actually calculate the negative correlation between the markets and/or models so as to come up with an optimized or at least for guys like me, a good enough working system that is able to work several non-correlating setups that comliment each other with regards to smoothing the equity curve since the markets change their behavior every now and then. Thanks again.
i have never come across systems that were truly negatively correlated. sure that is the ideal: positive expectation, yet negatively correlated. IMHO reducing positive correlation below 0.2 or 0.1 is already doing a great job in terms of diversification. (hence that correlation as such is not necessarily always delivering intuitive results.)
but jack improved, didn't he? i remember two years ago i read posts of him, didn't get if he really tried to say something and had a headache. recently i read his posts, didn't get if he tried to say something ... but had no headache. (jack, if you read this ... no real offense intended, just kidding ) sorry alan, couldn't resist to pollute your thread a little. something more on topic. many questions in trading systems boil down to whether you face an overfitted backtest or not. i tried the edge test concept, but i was frustrated because i had a "system" which clearly passed every requirement (i hope i did no drastic mistake ) but lost money almost from day one. so i searched for something else to give more clue on overfit. we recently started to use a tool that is an optimise-walkForward-optimise-walkForward. usually results are worse than in the (presumably overfitted) best paramter set. if they are not too far off, we think that the fit did not stretch too far. which leads to the second issue: which paramter do you fit to. so far i was (well somehow still am) a defender of sharpe ratio opposed to sortino. my rationale was that it is conservative and good practise if you assume you can loose as much as you can win, thus your distribution of returns is (more or less) equally shaped on both sides of the mean. recently it seems that those ctas who use the sortino approach seem to do better. names that come to my mind are Winton and Bluetrend. i am re-considering now to move towards sortino as the main figure (always given that you finally look at a set of figures). the rationale could be that in trend following you bet on a notEqually shaped distribution by cutting looser and let winners go. thoughts? peace