Making an almost profitable system profitable

For what it's worth, when I was using another MA + price system from another thread, the only way it was profitable was to stay in every trade until the MA slope changed. I tested with a few different fixed targets and each of them resulted in net losses. It is my conclusion, then that a trend following system needs to to for home runs every time in order to make up for the chop. I needed the occasional 30+ tick winner to offset the losses.

I was only profitable when I stayed in every trade until the MA changed and only when taking long trades after HH/HL and short trades after the opposite.

Occasionally you miss a up trend after a LL, for example, but overall this was the way to go.

I may make a thread revisiting that system at some point in the future.
 
I wonder if time of day matters. Is there any data on when price tends to be the trendiest?

Seems like big trendy moves can start anytime, though.
 
Well if you want to remove some bad trades at a glance I would suggest testing a few possibilities. To start with maybe a filtering system.

filter out ranges:

Seems like the problem occurs around ranges for obvious reason. So filter them out. I am sure there is an indicator for that. If not take a look at some of Jack Hershys complete charts with no annotations that show price ranging when ever we hit a long term channel trend line. Just stay out of those situation ? (Assuming his charts are a good representation of markets.

Another filter could be yet another more long term moving average. As the market gets in to a range mode, it will not be long before the longer term moving average flattens. A bit late, but you may be able to take out some of the last bad trades in ranges.

Looking at your original chart I notice that (based on little data) that the system is about as good (or worse) then simply looking for doji/hammer type bars (long tails)

I wonder if the rest of the data is similar but suggest to run a test to look for optimum long candle parameters. Maybe the data on your chart is a good representation and it will show better results? If not, you can combine the two and only go short when the candle is present of take profit when it is present along with another rule of max allowable retracement based on optimization.
 
On the subject of not having more then 3 or 4 of these in a row. I am not sure of its statistical significance in CW math. Also when do you start to count if you do not yet know if you are in such a situation.

Maybe look at where price intersects the MA and if you see that it is done at price levels that are not far apart then trading stops. But this is similar to the longer moving average idea i guess.

Again I am looking at limited data, but I assume the chart represents most the data.
On your second chart I notice the 4 bigger bars, 2 green followed by 2 red.

It is interesting that the every bar in your circle has a part of it that touches the the price range of the first big green bar. but this idea continues through the chart every where to some degree. Visually it is more in the part that is circled. maybe we can look at the next bars and see if they hit the mid point of the first big range big bar we stay out?
 
The part circled is full of ugliness, you see (going from left to right) a progression occur.
bars get smaller, and develop have smaller bodies 2/3rds of the way in. Then they get back to normal. also during this time we see opens and closes of single bars and neighboring bars are have a small difference in price. maybe filter that out ?

You may be left with too little to trade for your liking. these criteria will give too many false signals to stop trading. You can make rules that if one trigger is met the out put is 1. if 2 are net the out put is 2.

you may find an optimal number of triggers that gives the best profit. You can even make criteria 1 only add .5 to the out put as it is not as good and criteria 3 add a value of 2 to the output.

This will keep the programing simple once you have your added filter(s) programed. now all you need to do is give value to each EA.

Sorry If these posts take over the thread. I shoot things out there that are dumb, but some times people see something in them. Unfortunately my next post will qualify how serious you should take what i write.
 
On the subject of position size. We may not be flipping a coin and unless you have flipped on personally it is better to not have any preconceived ideas about it or relating it to trading.

I always wonder how consistent max draw downs are and if there is a black swan that can be avoided simply by stopping trading after X amount of consecutive loses and reentry after y amount consecutive positive trades (that were not actually real trades)

Anyways the idea of doubling up once you have lost seems risky at a glance.
But there exists a maximum number of times you would have to double up based on historical data. lets say the max is y. now you know it is Y and if you can cover teh cost of Y you have positive outcomes all the time.

What if the Y increases in future trading.

Each set of trades in this case (a set being consecutive number of wrong trades where we doubled up) should be seen as a single trade. in this case I will call them string.

so each string yields a certain profit 100% of the time historically.
each string yeild a profit that ranges from L to H

If Y (max number of consecutive bad trades) increases to Y+1, we close the trade. since we are looking at multiple trades as single strings. each string can be said to have a stop loss of Y+1

Also an actual traditional stop loss may be needed on each trade.

The idea of doubling up is too scary, it actually is pushing risk down the road to a point that historically we end up in the green 100% of the time.

There are much better variations of this.
 
Quote from 1a2b3cppp:

Disagree. I am profitable because I average into trades.

A fixed sizing scheme would not work for me, unless I misunderstood what you are saying.
More...

Continuing to variations of the doubling up, here you mention one.
There is also another way to go about it I was playing around with.
As i understand scaling in means adding to a trade, I assume a wining trade but this may not be so.

Instead of adding funds to the trades. you can also add to the risk. allow for larger stop losses.

Dom993 mentions "You are right this system is never on the wrong side for long" maybe allow for more leeway so that you don't have to exit and enter during those times. this will mean that your profitable exits will also be smaller.

This is one reason I like the idea of of the exit strategy being a different system then the entry system. it allows for some leeway on the entry without compromising the exit. well kinda..

Again sorry for taking over the thread. I should of told you to read no more then one post a week lol. Or better yet I should of stayed on topic and spread out my posts through the thread. It is what it is.

For what it is worth your thread is one of the best I have read. Look me up when you become a guru :)
 
Quote from IronFist:

What if we were to add the following condition:

Only take long trades after HH HL.

Only take short trades after LL LH.

That was one of the necessary components for the variant of jjrvat's sytem that I was using to be profitable. Taking every trade was bad. Taking trades following the appropriate high/low criteria made the system profitable (over my test months).
More...


To follow this interesting quest would you - or someone else - please let me
have the indicator's (and possibly the funtion's) formulas.

Thanks for taking your time!

GG
 
Quote from Goalgetter:

To follow this interesting quest would you - or someone else - please let me
have the indicator's (and possibly the funtion's) formulas.

Thanks for taking your time!

GG
More...

There weren't really any indicators, it was just a WMA and an HMA. I think I've posted the entire system before.
 
You must log in or register to reply here.
Back
Top