Quant basket trader? I need your help! | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

BASKET TRADE Management: Using multiple trade management EA's on basket trades 11 replies

Basket Knights - A Basket Trading Round Table 1,313 replies

Need Help on EOD Basket Close Codes 0 replies

TMS basket: Dashboard for a basket of MA-Filtered RSI Signals 135 replies

Quant - I will run a forward walk for you! Analyze your risk! 0 replies

Attachments: Quant basket trader? I need your help!
Exit Attachments

Quant basket trader? I need your help!

  • #21
  • Aug 30, 2015 12:03pm Aug 30, 2015 12:03pm
Hello everyone,

First let me put this out there, my degree is in health & pe so I'm pretty far away from being a able to do a lot of the math you talk about, but I research it and can basically understand the concepts behind it. Now when I am reading these different threads that guys post in I see conversations going on with something like this is good in a ranging period but not in trending and vice-versa and debate goes on. Here is my dumb question for everyone. Can't you just turn your different programs on and off when the market is changing? I would think that with the numbers and the ability to manipulate them that you guys could do something Malkovian and determine state changes and know when the probability of the change will take place.

Next question. Now in my low level work I"m trying to determine what works best in a range on my spread sheets. Any suggestions on where to look on how to mathematically determine what a range and trend is? Because I think I found a good concept for range trading, break out, and trends. I would just like to do some back tests and look at what happens to each "system" during each of those market types.
A woman drove me to drink, and I forgot to thank her
  • #22
  • Aug 30, 2015 1:44pm Aug 30, 2015 1:44pm
Quoting Atokys
...a trading system needs to be objective in defining a trend, identifying a potential start of said trend and defining a point where, if reached, the trend can be said to have ended.
I've read a paper where the authors do exactely this. They break down (in hindsight) the market into pieces of linear trends. I personaly can't see the trend as something that starts and later stops. Up trend and, all the sudden, down we go. A candle is in a up trend and the candle right after is in the next down trend. I find this too artificial. Take the picture post 17 where would you pin the turning points? I see the trend as the general drift of the market (in the bandwidth I consider). It simply oscillates between positive and negative values, close or far from zero.

Quoting simpleguy
Can't you just turn your different programs on and off when the market is changing?
This leads me to the remark by simpleguy. If some method considers the market is in a range then we play the range strategy. If the market is found to be in a up trend, we play the up trend strategy. ON/OFF. We can easily extend this. If the method considers the market is very likely in a up trend but it may still be in a range, we can run both the strategies but with adjusted money for each to trade. Yet this doesn't tell what are the best strategy to be run in a range or in a trend. Can we infer it from the data without my own and certainly wrong beliefs on how the market sall behave ? If the market shifts to uptrend what to do with the running trades? Are their SL and target still valid? Shall an extremely strong down trend be traded with the same strategy as a mild down trend?
No greed. No fear. Just maths.
  • #23
  • Aug 30, 2015 9:21pm Aug 30, 2015 9:21pm
Quoting PipMeUp
I've read a paper where the authors do exactely this. They break down (in hindsight) the market into pieces of linear trends. I personaly can't see the trend as something that starts and later stops. Up trend and, all the sudden, down we go. A candle is in a up trend and the candle right after is in the next down trend. I find this too artificial. Take the picture post 17 where would you pin the turning points? I see the trend as the general drift of the market (in the bandwidth I consider). It simply oscillates between positive and negative values,...
If you focus on a small number of specific trades, you have already curve-fitted your system. There is no one correct way to discern your own beliefs about markets, but you can try this:
- Come up with a hypothesis: It can be something simple like: "A trend is said to have established in one direction when the 50 EMA crosses the 100 EMA in that direction"
- Test the hypothesis across a wide range of markets and time span (perhaps 20 different markets across a 10-20 year period)
- See if any of it makes sense to you
Quote
This leads me to the remark by simpleguy. If some method considers the market is in a range then we play the range strategy. If the market is found to be in a up trend, we play the up trend strategy. ON/OFF. We can easily extend this. If the method considers the market is very likely in a up trend but it may still be in a range, we can run both the strategies but with adjusted money for each to trade. Yet this doesn't tell what are the best strategy to be run in a range or in a trend. Can we infer it from the data without my own and certainly wrong...
Personally, I would design systems that each capitalize on a certain type of market structure and fire off when conditions signal the possible start of such a period.
  • #24
  • Aug 30, 2015 10:37pm Aug 30, 2015 10:37pm
Quoting PipMeUp
{quote} I've read a paper where the authors do exactly this. They break down (in hindsight) the market into pieces of linear trends. I personaly can't see the trend as something that starts and later stops. Up trend and, all the sudden, down we go. A candle is in a up trend and the candle right after is in the next down trend. I find this too artificial. Take the picture post 17 where would you pin the turning points? I see the trend as the general drift of the market (in the bandwidth I consider). It simply oscillates between positive and negative...
I've done something like what Atokys suggested using the zig-zag indicator as an "optimal" representation of the trend or trade-able swings. This isn't too different from what you described earlier, perhaps just a different way of visualizing "trend". There were a couple of threads that have since gone into the trash bin that related to this idea. One benefit to the idea is that after an up swing you know the next swing will be down and vice versa. From a systematic standpoint, this simplifies things considerably. You can say that an optimal entry/reversal strategy will enter/reverse on all the peaks and troughs in the zig-zag line. Any deviation in terms of exit/entry (too early, too late) decreases the optimality of the strategy and could be summed up as "error".

Quote
Imagine two traders who see this trend about to end. One of them is already long since the beginning and the other one is not in the market. The first one has no reason the close, even partially...
In this case the one who has no position will be better off waiting for and taking the next entry signal. Also, the one with the profitable open trade will be better off taking the next entry signal, because doing so constitutes the implied positive advantage (alpha) of that trader. If the trader who has an open profit ignores the next reversal trade signal, that trader is operating sub-optimally because their advantage says to exit and take the next trade. You're correct that their optimal exposure will be different because there is a cost for entering the market for the first (flat) trader in terms of alpha decay for a given opportunity. If the first trader's expected profit per trade is X, and the market has already moved X or more then there is no positive expectancy for an entry at that point. If the market has moved less than X, then there could still be some positive expectancy in that trade, but it may be less than some minimum acceptable level.
  • #25
  • Aug 31, 2015 3:07am Aug 31, 2015 3:07am
Quote
Come up with a hypothesis: It can be something simple like: "A trend is said to have established in one direction when the 50 EMA crosses the 100 EMA in that direction"
This is exactely what I want NOT to do. I don't want to guess random rules out of the blue and check them against a backtest. I want the rules to be infered from the data. That means without me involved. Please understand that I'm NOT looking for a profitable system to trade. I'm way more ambitious. I want to make the system emerge from the data. Yes I'll certainly fail but in the process I would have learnt and discovered very importants things that others just overlook.

Quote
You can say that an optimal entry/reversal strategy will enter/reverse on all the peaks and troughs in the zig-zag line.
But I showed it isn't true! If you have for example a zigzag that goes 6 units up, 1 unit down and 4 untis up, it is better to not short to get the 1 unit retracement and stay long all the time. Of course, in hindsight, it's idiotic since you know what the price did. But in real time the zigzag is only the best estimate you have. Entering the short (any position) comes with a risk. When you factor in this risk the short is no longer good to take.

Should I give you the next hour of ticks in E/U you couldn't take all the up's and down's just because of the trading fees. How would you trade this to maximize your profit? You'll quitely realize that it depends on if the next swing is smaller than, between 1 and 2 times or greater than two times your trading costs.

Quote
In this case the one who has no position will be better off waiting for and taking the next entry signal. Also, the one with the profitable open trade will be better off taking the next entry signal, because doing so constitutes the implied positive advantage (alpha) of that trader. If the trader who has an open profit ignores the next reversal trade signal, that trader is operating sub-optimally because their advantage says to exit and take the next trade. You're correct that their optimal exposure will be different because there is a cost for...
This is exactely what I meant. This implies that you can't have a strategy that gives the exposure you should have only given the PA. Yet a very used mean reversion strategy under OU process assumption is exactely this: "have a leverage linearly proportional but opposite to the distance to the mean". I'm not saying it won't work. I'm saying it is not the best possible method; even if/when the price is really OU.
No greed. No fear. Just maths.
  • #26
  • Aug 31, 2015 3:11am Aug 31, 2015 3:11am
Quote
One benefit to the idea is that after an up swing you know the next swing will be down and vice versa.
I built a pSAR like zigzag that says UP, DOWN or "well I'm not sure I won't tell". So I can have UP, don't know, and UP again.
No greed. No fear. Just maths.
  • #27
  • Aug 31, 2015 4:41am Aug 31, 2015 4:41am
Quoting PipMeUp
This is exactely what I want NOT to do. I don't want to guess random rules out of the blue and check them against a backtest.
Not asking you to guess random rules but rather, come up with a belief, then validate or discard the belief based on observation.

You first need a framework of beliefs upon which you base your statistical testing and optimization on. Even objective Sciences and Math are based on a philosophical framework.
Quote
I want the rules to be infered from the data. That means without me involved. Please understand that I'm NOT looking for a profitable system to trade. I'm way more ambitious. I want to make the system emerge from the data.
I wish you all the best then. Do beware that even random data can produce 'profitable' trading systems that are curve-fitted and have no application to the market. Personally, I don't even see the point of jumping through hoops to make the best technical system of entry and exit. Successful technical system traders I know have awfully simple entry/exit systems but highly optimized and fine-tuned risk management systems.
  • #28
  • Aug 31, 2015 5:05am Aug 31, 2015 5:05am
Quote
Not asking you to guess random rules but rather, come up with a belief
It's just wording. What you call belief is what I call random rules out of the blue.

Quote
Successful technical system traders I know have awfully simple entry/exit systems but highly optimized and fine-tuned risk management systems
Repeating myself "I'm NOT looking for a profitable system to trade".
No greed. No fear. Just maths.
  • #29
  • Aug 31, 2015 5:07am Aug 31, 2015 5:07am
Is there a valid meaning of these strange clusters forming an ellipse or are they just an artifact?
Attached Image (click to enlarge)
Click to Enlarge Name: strange_ellipse.png Size: 91 KB
No greed. No fear. Just maths.
  • #30
  • Aug 31, 2015 5:17am Aug 31, 2015 5:17am
Quoting PipMeUp
It's just wording. What you call belief is what I call random rules out of the blue.
A belief is a belief. A random rule out of the blue is a random rule out of the blue. They clearly mean different things to me but not to you, but then again, our perception differs because we hold different beliefs.
  • #31
  • Aug 31, 2015 5:47am Aug 31, 2015 5:47am
Definition of belief in English by the Oxford dictionary:

belief: noun. An acceptance that something exists or is true, especially one without proof; a firmly held opinion
No greed. No fear. Just maths.
  • #32
  • Aug 31, 2015 5:58am Aug 31, 2015 5:58am
Not the definition of 'belief':
Random rules out of the blue.
  • #33
  • Aug 31, 2015 6:14am Aug 31, 2015 6:14am
Quoting Atokys
Not the definition of 'belief': Random rules out of the blue.
What else can that be "A trend is said to have established in one direction when the 50 EMA crosses the 100 EMA in that direction"?
Attached Image (click to enlarge)
Click to Enlarge Name: EURUSD-D1.png Size: 14 KB
No greed. No fear. Just maths.
  • #34
  • Aug 31, 2015 6:20am Aug 31, 2015 6:20am
>
Quoting PipMeUp
{quote} What else can that be "A trend is said to have established in one direction when the 50 EMA crosses the 100 EMA in that direction"? {image}
Its a belief.
  • #35
  • Aug 31, 2015 6:37am Aug 31, 2015 6:37am
Then it validates that I'm not interested in believes either.
No greed. No fear. Just maths.
  • #36
  • Aug 31, 2015 6:53am Aug 31, 2015 6:53am
Quoting PipMeUp
{quote} But I showed it isn't true! If you have for example a zigzag that goes 6 units up, 1 unit down and 4 untis up, it is better to not short to get the 1 unit retracement and stay long all the time. Of course, in hindsight, it's idiotic since you know what the price did. But in real time the zigzag is only the best estimate you have. Entering the short (any position) comes with a risk. When you factor in this risk the short is no longer good to take.
I'm not sure why this is a problem. Can't you just set the minimum move amount to be greater than the threshold required to make such a trade worthwhile? For instance, increase 1 unit to 3 units on the zig zag minimum move setting and it will always be worth taking the shorts, and this will also filter out all the 1 unit retracements.
  • #37
  • Aug 31, 2015 9:07am Aug 31, 2015 9:07am
Attached Image

The New Joy of Sim. Work on your trading using safe simulation studies, thereby neatly avoiding the messy, emotional, organic part. Also, by the same author: Mathturbation - How to Avoid the Emotional Issues by Tinkering Endlessly with your System.
  • #38
  • Edited 9:45am Aug 31, 2015 9:16am | Edited 9:45am
Quoting FXEZ
I'm not sure why this is a problem
It's not a problem per se.
Quoting FXEZ
For instance, increase 1 unit to 3 units on the zig zag minimum move setting and it will always be worth taking the shorts, and this will also filter out all the 1 unit retracements.
Reducing the sensitivity of the ZZ indicator makes it trade a higher TF. Your uncertainty increased. So does your SL. Doing this you change centimeters to inches. Up to the scale nothing changes.

Attached Image (click to enlarge)
Click to Enlarge Name: schema.png Size: 52 KB
No greed. No fear. Just maths.
  • #39
  • Aug 31, 2015 4:55pm Aug 31, 2015 4:55pm
Quoting PipMeUp
Is there a valid meaning of these strange clusters forming an ellipse or are they just an artifact? {image}
Wow this is interesting. Can you tell us more about the heat map?
Thanks.
What's it all about? It's all about money.
  • #40
  • Aug 31, 2015 9:31pm Aug 31, 2015 9:31pm
PipMeUp, yes I see what you mean - the thresholds between L/S increase which decreases the profit potential on each swing. The thing is, I'm not certain that SL alone can provide any edge. That said, Adal posted an idea a while back that tends to contradict this and may be of interest to you regarding using a volatility slider as a way to dynamically alter TP/SL in some "optimal" way according to the current market volatility. The pictures are missing but I think the idea is fairly clear from the text:

http://www.forexfactory.com/showthre...01#post7552801
http://www.forexfactory.com/showthre...05#post7552905 (middle section)

Quote
So the results are conditional on market volatility.
Maybe this is the answer you've been looking for?
0 traders viewing now