Trend Following depends on Prediction

[1a2b3cppp]

Quote from dom993:

Even in that very case you are taking for example, there is no prediction involved, the trade is taken based on:

- the observation of the past,
- the assumption that the future will *statistically* reflect the past.

One has to assume that the outcome of any particular trade is a random event.

You are not taking a particular trade because you are predicting it is going to be win, you are taking it because it is one in a series of hundreds of trades with similar circumstances, which have a statistical edge (taking *all* the trades that have similar circumstances, you will make money - *THAT* is what the prediction is about, *NOT* the current trade)

That's what prediction means.

If you know that 70% of the time a certain situation will result in price going up 5 points, then when you see that situation in the future you can predict that price will go up 5 points.

Without that statistical knowledge they would just be randomly trading (ie. gambling).

Why would you open a long position if you couldn't reasonably predict that price was going to go in your favor?

Prediction doesn't mean you have a 100% chance of being right.

If you flip a coin 10 times I predict 5 times it will be heads. It might actually be 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 times, but I would predict 5 times. If I could make a bet such that the closer the results were to 5 times the more money I would make, I would take that bet because of my prediction.

For example:

Heads 0 or 10 times = -2
Heads 1 or 9 times = -1
Heads 2 or 8 times = 0
Heads 3 or 7 times = +1
Heads 4 or 6 times = +2
Heads 5 times = +3

I would take that bet ever time because of my prediction that you would get heads on 5 out of 10 flips.

 

[dom993]

Quote from 1a2b3cppp:

That's what prediction means.

If you know that 70% of the time a certain situation will result in price going up 5 points, then when you see that situation in the future you can predict that price will go up 5 points.

Without that statistical knowledge they would just be randomly trading (ie. gambling).

Why would you open a long position if you couldn't reasonably predict that price was going to go in your favor?

Prediction doesn't mean you have a 100% chance of being right.

If you flip a coin 10 times I predict 5 times it will be heads. It might actually be 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 times, but I would predict 5 times. If I could make a bet such that the closer the results were to 5 times the more money I would make, I would take that bet because of my prediction.

For example:

Heads 0 or 10 times = -2
Heads 1 or 9 times = -1
Heads 2 or 8 times = 0
Heads 3 or 7 times = +1
Heads 4 or 6 times = +2
Heads 5 times = +3

I would take that bet ever time because of my prediction that you would get heads on 5 out of 10 flips.

If prediction has a statistical meaning to you, without any specific expectation from the current instance, then sure.

But I guarantee you that for most people, prediction implies specific expectation from the current instance - and all kind of frustrations when the current instance doesn't meet that specific expectation.

Realize that to make money from a system with an edge, you must faithfully take hundreds of trades with that system, so that its edge can play out, statistically.

In another thread, I took the example of my system CL always-in, which after the 50th live trade, was down net -7655 (-8610 from its early P&L peak) (trading 1 contract). Would *you* have kept trading it?

 

[themickey]

If price is sitting on support and rises and I go long, then falls below and I go short, these aren't predictive trades.
I'm trading the level.

If my computer is automated and places those trades, my algo isn't trading on prediction, it trades on maths, it trades on reaction, it trades on an algo, 'IF' this, then do that....

A prediction is a mindset. (Surf's method of trading)
Good traders don't have a mindset, they go with the blows.

Now what's so hard about this concept you guys don't understand?

 

[marketsurfer]

Quote from themickey:

If price is sitting on support and rises and I go long, then falls below and I go short, these aren't predictive trades.
I'm trading the level.

If my computer is automated and places those trades, my algo isn't trading on prediction, it trades on maths, it trades on reaction, it trades on an algo, 'IF' this, then do that....

A prediction is a mindset. (Surf's method of trading)
Good traders don't have a mindset, they go with the blows.

Now what's so hard about this concept you guys don't understand?

It's prediction regardless of your semantical argument.

You are predicting price will continue in the same direction as your entry--- otherwise you wouldn't waste the commission.

 

[dtrader98]

Quote from themickey:

If price is sitting on support and rises and I go long, then falls below and I go short, these aren't predictive trades.
I'm trading the level.

If my computer is automated and places those trades, my algo isn't trading on prediction, it trades on maths, it trades on reaction, it trades on an algo, 'IF' this, then do that....

A prediction is a mindset. (Surf's method of trading)
Good traders don't have a mindset, they go with the blows.

Now what's so hard about this concept you guys don't understand?

Your applied maths and rules are both constructed around prediction. How do you think those rules-- 'IF this then do that'-- are devised? There must be some underlying reason that compels you to go long if price sits on support, then rises-- whether you know the reason or not, it's based upon prediction that over enough trials, some favorable (to you) behavior will follow that action... more often than not. The rules are models that are built based upon an expectation or prediction that the model will continue to predict some relationship will continue to exist over some uncertain future.

I have a harder time understanding how traders don't seem to understand they are predicting.

 

[eurusdzn]

Is it because Dunn can't move as fast as I can with my 100 shares of SPY ?
Do THEY stick to rigid trend following where about 30% are winners.
Itis uncanny ... This seemingly market truism of about 30-35% winners if you hold through large retracement s.
And, isn't everyone's MR and chop system about 60% - 70% winners?
The summation of the two equal to 100% and therefore reflective of the nature of the market through a cycle of chop/range to trend and back.
I toss aside the 30% system, choke up on the bat, and exit quicker without enduring
a large retracement. In my testimg that is. I find it interesting when I vary the exits in
System testing that the net profit is pretty much the same across a wide range of exit parameters. Of course the avg win and loss, number trades, % profitable change to something that may not even be trend trading. What is it long term swing trading?
Incidentally, I am referring to a large basket of stocks not the standard fixtures markets.
By the way Slopetrader Maverick74 was spot on the nat gas trend early and the yen etc..
Not a bad guy to check in on for major changes/moves.
Last thought, fast entries and slow exits work best when major trends exist. No where does "let winners run" better apply. Does it apply anywhere else?

 

[dom993]

Let's take a hypothetical trend following system, with a 20% win rate, and a 5:1 reward:risk ratio.

What is the "prediction" one is making when taking one trade based on that trend following system?

 

[1a2b3cppp]

Quote from dom993:

If prediction has a statistical meaning to you, without any specific expectation from the current instance, then sure.

Prediction is possible because of statistics. Otherwise it's just guessing.

But I guarantee you that for most people, prediction implies specific expectation from the current instance - and all kind of frustrations when the current instance doesn't meet that specific expectation.

You should never enter a position without a specific expectation. That's the same as random entry.

Realize that to make money from a system with an edge, you must faithfully take hundreds of trades with that system, so that its edge can play out, statistically.

I agree. And you also must take every trade the system gives you because you never know which one is going to be a winner, but you know that statistically speaking you will win over time.

In another thread, I took the example of my system CL always-in, which after the 50th live trade, was down net -7655 (-8610 from its early P&L peak) (trading 1 contract). Would *you* have kept trading it?

What was the time frame? What was the account size? If it flips all the time and that was only two days' worth of data, that's one thing. If that was 6 months' worth of data, probably not.

Then again, I've never seen an always-in trading system that is profitable. I think the idea is fascinating but I don't know that anyone can do it*.








*except Jack Hershey who makes 3x the daily ATR every day with his always in system.

 

[1a2b3cppp]

Quote from themickey:

If price is sitting on support and rises and I go long, then falls below and I go short, these aren't predictive trades.
I'm trading the level.

What? Yes they are. You are predicting price will do a specific thing and that's why you chose to go long or short.

If my computer is automated and places those trades, my algo isn't trading on prediction, it trades on maths, it trades on reaction, it trades on an algo, 'IF' this, then do that....

A prediction is a mindset. (Surf's method of trading)
Good traders don't have a mindset, they go with the blows.

Now what's so hard about this concept you guys don't understand?

No.

Quote from marketsurfer:

It's prediction regardless of your semantical argument.

You are predicting price will continue in the same direction as your entry--- otherwise you wouldn't waste the commission.

Yes.

Quote from dtrader98:

Your applied maths and rules are both constructed around prediction. How do you think those rules-- 'IF this then do that'-- are devised? There must be some underlying reason that compels you to go long if price sits on support, then rises-- whether you know the reason or not, it's based upon prediction that over enough trials, some favorable (to you) behavior will follow that action... more often than not. The rules are models that are built based upon an expectation or prediction that the model will continue to predict some relationship will continue to exist over some uncertain future.

I have a harder time understanding how traders don't seem to understand they are predicting.

Good post.

Here's a flowchart that may help people understand it:

 

[1a2b3cppp]

Quote from dom993:

Let's take a hypothetical trend following system, with a 20% win rate, and a 5:1 reward:risk ratio.

What is the "prediction" one is making when taking one trade based on that trend following system?

If you do not believe price is going to go in your favor over time you should not take the trade. So the prediction for each trade is that it probably won't go in your favor but if it does it will make up for the last 4 losing trades. You can make that prediction based on the statistical analysis that gave you those numbers.

Your prediction is that over 10 trades you will win 2 and lose 8 but come out ahead 2. It's similiar to the heads/tails example I posted; you are using statistics to predict outcome over time.