The Myth of Averaging | Page 5 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

1% Money Management is a myth, fib is a myth and so on.. 45 replies

'Averaging Down Is for Losers' - valid or not? 27 replies

Fast Averaging versus Slow Averaging 1 reply

Averaging 7 replies

Martigale and Averaging Down 4 replies

Attachments: The Myth of Averaging
Exit Attachments

The Myth of Averaging

  • #81
  • Edited 12:49pm Jan 18, 2024 10:37am | Edited 12:49pm
Quoting yoriz
{quote} Thank you for your support, sharingan9, but just put him on the ignore-list as he is a troll here on FF. Just check the last two threads he started: one implying that real traders hedge and Martingale (yes, really!! And he is upset about my averaging down thread?!?!), and in another thread he is bashing other FF users in rhyme?!
Hey yoriz,

With all due respect, and I know that nub doesn't need to be defended (please don't bash me nubcake!), but his martingale thread was "tongue and cheek"... to his credit, nubcake has been a long-time (since at least 2009) vocal critic of approaches that advocate martingale, same-pair hedging (or "nedging" as merlin called it back in the day) and blindly averaging down (nanningbob and the like)... and his BS detector pops up when someone claims that they have been long-term profitable by trading those approaches...if you read his older posts, he is one smart trader, and the traders that he bashed are really never heard from again after making their claims of success...

IMO, the point that nubcake makes is the bane that every technical system confronts... what happens in the past doesn't pertain to what will happen in the future, and that mining past data using random entries (or even non-random entries) and averaging down without a true "edge" will get killed at some point when the trend runs strongly opposite of a basket of open trades (i.e. the death trade). I believe that what you have shown is fascinating from the standpoint that averaging down may work even with random entries IF you close a losing basket at some point before the size of the loss does irreparable damage to your account.

The approach that you've presented, and that I believe that nubcake is in disagreement, is the opening of trades at highly specific TL/SL points that were optimized based on data mining. The question becomes what happens to the equity curve when you do a walk-forward test with those same parameters? In other words if you data mined from 2010-2016 and then walked forward from 2017-2023, do you generate the same equity curve? Or better yet, if you optimized the TP/SL points for 2010-2011, what does the walk-forward curve look like in 2012? Do a yearly (or bi-yearly if you need more data points) rolling period optimization and compare the parameters for each time period. Did the parameters need to be readjusted every year (or every two years)? If they did, then the approach will most likely fail miserably in real time... but if the parameters are stable, then you might be onto something!

Just my 2 cents,

Dave
  • #82
  • Jan 18, 2024 12:34pm Jan 18, 2024 12:34pm
Dave, I actually wanted to stop putting energy into this thread, but I respect you so let me reply.

My bad if the man is not a troll, but his communication style is certainly a bit "unusual". His first post came down to:

"I did not read the thread, but here is my opinion why you are wrong".


His next post literally contained sentences like:

"I'm annoyed at your arrogance"

"How dare you"

"it's a waste of time"

"Shame on you"


Sorry, but I my book that is simply rude. Moreover, he seems to reads things that are not there or puts words in my mouth:

  1. "Throwing shit at the all and seeing what sticks". I merely documented my experiments in averaging down. This is what a public forum is for, right?

  1. "If you've done something astonishingly novel and new you should have mentioned what that is.". No, I don't think I have. I was merely surprised, as I clearly mentioned. I did not claim the next Nobel price.

  1. "Try to save face, or whatever, but this thread isn't what you dress it up to be.". Uh? Again, I merely documented my experiments. Nothing to "dress up".

  1. "anyone who knows a lot more than you seems like a troll because you just can't keep up. You do you, boo.". Or could it have been the communication style that gave me the impression I was dealing with a troll?

  1. "complained when nobody continued to post here." When did I complain? This thread had already ended begin of July (apart from some users that later posted questions or remarks).

  1. "You've posted your results, people have looked it at, and moved on. What more do you want? What more is there to discuss?". Uh? Nothing. The thread was already done in July, but your post reopened it.

I am not a native English speaker so perhaps it's just me, but this comes across as hostile or belittling. Not fitting for a friendly, public forum where traders can exchange ideas and discuss differences in opinion in a normal, civilized manner.

  • #83
  • Jan 18, 2024 1:27pm Jan 18, 2024 1:27pm
Hi Yoriz,

Understandable.

But aside from that, and if you have the time, what about repeating the analysis on one pair with the walk forward idea- preferably, the rolling walk forward?

Best,

Dave
  • #84
  • Jan 18, 2024 9:03pm Jan 18, 2024 9:03pm
Quoting davelansing
{quote} Hey yoriz, With all due respect, and I know that nub doesn't need to be defended (please don't bash me nubcake!), but his martingale thread was "tongue and cheek"... to his credit, nubcake has been a long-time (since at least 2009) vocal critic of approaches that advocate martingale, same-pair hedging (or "nedging" as merlin called it back in the day) and blindly averaging down (nanningbob and the like)... and his BS detector pops up when someone claims that they have been long-term profitable by trading those approaches...if you read his...
Your translation of my impatience is accurate.

I'm pretty sure Merlin died of a stroke after rolling his eyes too far back into his head after reading posts on this forum.

I would raise an eyebrow in curiosity if the entry time was changed (if I read correctly it was LO or some specific time chosen) just to see how much it altered the results (good or bad). It's otherwise not an interesting topic for obvious reasons.

Unless I've mis-skimmed, the "positive results" weren't even particularly positive at all, so even a curve fit is squeezing blood from stone in this configuration.

Quoting yoriz
His first post came down to: "I did not read the thread, but here is my opinion why you are wrong".
Yes, precisely, because I've been around long enough to spot the same horse being beaten again and again, without having to read the entirety of every newbies "experiments" and fallacious arguments. I've been around the block enough times to know ahead of time, skim for the important bits, and filter out the rest

I don't have the desire to read every single thought-crumb of every excitable newbie who is going through the exact same motions as every other one who has come before, including myself and my OWN "experiments" etc. I did more than just skim, but I'm not going to read every little thing, so I say I skimmed.

I know enough to know when someone DOESN'T know!

Quote
His next post literally contained sentences like: "I'm annoyed at your arrogance" "How dare you" "it's a waste of time" "Shame on you"

And it's a waste of time and arrogant because:

Quote
I merely documented my experiments in averaging down.

Which is one of many dead horses that KEEP GETTING BEAT regularly. Bring something new, or have people become annoyed that you are just repeating what's already been repeated ad nauseam while fishing for more discussion on a dead topic.

Quote
Or could it have been the communication style that gave me the impression I was dealing with a troll?

Yes, because you read with your feelings. The hallmark of someone not actually wanting to learn, they just want to be excited even when what they've discovered isn't a thing.

Quote
The thread was already done in July, but your post reopened it.

I think I found my way here from you referring to YOUR thread in someone ELSE'S thread. You fished for input, and you got it but didn't like it.

Quote
but this comes across as hostile or belittling.

Because it is, by design. The topic, and your replies since, have received the amount of respect I believe they have earned.

Quote
Not fitting for a friendly, public forum where traders can exchange ideas and discuss differences in opinion in a normal, civilized manner.

That's your feelings and opinions not based in reality. You simply don't like the reality of the "discussion" you prompted. If you want to be congratulated go tell your parents about it, not a public forum where some people can point out where you've fucked-up

And to get ahead of the clowns who'll pipe-up: I'm not a slow typist, so replies like this aren't a time-hit. Try some other angle.
Congratulations ScalaFX for coming out of the closet!
  • #85
  • Jan 18, 2024 9:55pm Jan 18, 2024 9:55pm
This is what started it:

Quoting yoriz
{quote} Well, I indeed stumbled on your interesting thread :-) In my thread "The Myth of Averaging" I experimented with random entries and controlled exits. You can indeed achieve a positive expectancy using Averaging Down (adding to losing positions): {image} And also by Averaging Up (adding to winning positions): {image}...
You implied you can trade randomly and have positive expectancy. True, but in the same sense that water is air because it contains oxygen... so, not actually true.

You went into a thread that was active, caught my attention with the above post, now here we are. Seems pretty straightforward.
Congratulations ScalaFX for coming out of the closet!
  • #86
  • Edited 8:48am Jan 24, 2024 8:29am | Edited 8:48am
Quoting davelansing
what about repeating the analysis on one pair with the walk forward idea- preferably, the rolling walk forward?
Below is a walk forward experiment on AUDCAD that creates and runs 5000 strategies in parallel where each of these strategies had their input parameters chosen randomly from these ranges:

Inserted Code
EntryHour           = 1 .. 23
RandomSeed          = 0
L1_TakeProfitPoints = 50 .. 1500
L1_StopLossPoints   = 50 .. 1500
L2_TakeProfitPoints = 50 .. 2500
L2_StopLossPoints   = 50 .. 2500
L3_TakeProfitPoints = 50 .. 3500
L3_StopLossPoints   = 50 .. 3500

I track the simulated(*) performance of these strategies over the last 1260 trading days (=5 years) in a rolling window. Every 20 trading days I sort them by Profit Factor and select the best strategies in the 98.7% .. 99.6% percentile range. So with 5000 strategies that is strategies 4935 to 4980. The upper boundary is to filter out strategies that were extremely lucky by chance. These selected strategies are then traded in parallel for the next 20 trading days, after which I rebalance the portfolio again.
Attached Image (click to enlarge)
Click to Enlarge Name: screenshot.png Size: 28 KB


As you can see, it does have a positive expectancy, but even when we use up almost all margin at a 1:500 leverage the profits are still too small.

To conclude: it was a nice experiment to test conventional wisdom, but the results are not useful for live trading.

(*) P.S. For simplicity the EA merely internally calculates open-close price and subtracts the commission the broker would have charged. It does not simulate slippage or swap. For sorting the strategies by performance that is probably sufficient. The strategies that were selected in the portfolio are actually send to the MT5 strategy tester to be executed as usual.
  • #87
  • Sep 7, 2024 11:14am Sep 7, 2024 11:14am
Quoting sharingan9
I personally disagree with the mathematicians or at least the phrasing of their argument in the original post. "Mathematicians will tell you that a strategy without an edge can not be made profitable with "smart" money management".. This statement is rather contradicts itself as "smart" money management is an edge. In essence the statement reads, "A strategy without an edge can not be made profitable with an edge" which is to say that trading is not a profitable venture lol.
I think it depends which mathematician you ask

There is a way to get a mathematical ege in risk management. I'm not going to tell what it is since it took me long time, through a lot of failed experimetns and a lot of thinking and doing math to find it. But trading costs and slippage are still a problem, so you can't just "coin flip" it since you have this negative drift over a long time. Still need that additional edge.
  • #88
  • Last Post: Sep 7, 2024 11:24am Sep 7, 2024 11:24am
Quoting yoriz
Level 3 optimization gave: L3_ScaleInPoints=200, L3_TrailingStopLossPoints=375. {image} I am giving up... Since AUDCAD is fairly ranging, perhaps this is not the best symbol to add to winners and follow trends? Let's try EURUSD now...
Adding to winners is not the best idea. If the market retraces 50%, you're usually at zero, or if the price falls down to the starting point, you will be at a loss instead of zero since by averaging up, you also averaged up your entry higher up.
And the second reason is that trend never lasts really long. Every trend has an end point. We don't know where that point is, but we know that there is one, and we also know that by time passing by, candle by candle, you're only getting closer to that point, so risk is increasing.
Your best bet is always at the start, then it starts to degrade.
0 traders viewing now