Simple Mean Reversion | Page 26 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

A Simple Mean Reversion Strategy 392 replies

Pairs Trading: Reversion to the Mean 22 replies

Forward test of my new mean reversion strategy 11 replies

Synthetic hedges, cointegration, mean reversion and similar stuff 419 replies

Cointegration, Synthetic hedges, mean reversion in R, Tech Thread 0 replies

Attachments: Simple Mean Reversion
Exit Attachments

Simple Mean Reversion

  • #501
  • Mar 17, 2019 6:29am Mar 17, 2019 6:29am
Quoting alphaomega
THINK! PATTERNS! MARKET MODELS! TOTAL NUMBER OF VARIATIONS??? CYCLES! TIME DIVIDERS! RELATIONSHIPS BETWEEN TIME AND PRICE! PRICE OVER TIME! PRICE MOVING OVER TIME! (Not time over price!) TIME IS THE KEY! TIME DETERMINES PRICE! (Not the other way) NUMBER OF MARKET MODELS BASED ON FIXED TIME CYCLES? HOW MANY? 3 ! 1. HORIZONTAL RANGE (Many variations) 2. RANGE WITH DIRECTIONAL DRIFT (Many variations) 3. SMOOTH DIRECTIONAL TREND (Many variations) WHAT ARE THE PERCENTAGES FOR EACH MODEL TYPE? 1 - DOMINATES! 2 - APPEARS FROM TIME TO TIME. 3 - VERY RARE!...
Ideally an EA should trade randomly buying and selling. Driven by some Momentum Driver. I have done it manually and currency strength in pure form (not averaged)
worked well. 100 trades , each of 10 pip size , are shot between daily high low range of 150 pips. As a result of screen-watching and executing trades manually my back suffered and now I can't sit for more than 10 minutes on the PC. Manual trading for me is out for ever.
If EA is fed with currency strength logic, Momentum is triggered in instalments. You have to be direction neutral. Direction bias only when you are sure of timing.
Ideal candidates will be mean-reverting symbols. Should correct by 40% at least in a week. Hedged broker platform will be ideal . 95% negatively correlated pair also should be traded to control drawdown.
Practice makes a person perfect
  • #502
  • Edited 2:29pm Mar 17, 2019 11:26am | Edited 2:29pm
Quoting MonkeyPlanet
{quote} Have you tried to apply this heuristic principle in trading? looks legit..
Yes, I have applied this principle and proven the theory works.
Staying in my lane...
  • #503
  • Mar 17, 2019 5:38pm Mar 17, 2019 5:38pm
Quoting simnz
{quote} Ideally an EA should trade randomly buying and selling. Driven by some Momentum Driver. I have done it manually and currency strength in pure form (not averaged) worked well. 100 trades , each of 10 pip size , are shot between daily high low range of 150 pips. As a result of screen-watching and executing trades manually my back suffered and now I can't sit for more than 10 minutes on the PC. Manual trading for me is out for ever. If EA is fed with currency strength logic, Momentum is triggered in instalments. You have to be direction neutral....
can you elaborate more on the "randomly buying and selling" vs "momentum driver".

perhaps if you can write the strategy in clear forms so we can understand the logic in it.
  • #504
  • Mar 31, 2019 6:08pm Mar 31, 2019 6:08pm
AO, are you still trading the strategy in post 1?
  • #505
  • Apr 12, 2019 5:32am Apr 12, 2019 5:32am
Hi AO,

Same question as alpha.
And if you do still trade it, how did you do today?
I was doing ok, but today was a bad day, a lot of runners in oposite direction from initial trade (aud and eur crosses).
All were cut off at some point but they still cost me a big chunck of profit i made last few weeks.
Persistence will get it. Consistency will keep it.
  • #506
  • Apr 12, 2019 6:36am Apr 12, 2019 6:36am
Quoting alphaomega
{quote} Nothing happens! But it will take you awhile (or forever) to understand what I mean by nothing. And why this is important. I strongly suspect very few people know about this. For example, I started trading a long time ago and I found the most important things only in the last few months. And that's not even funny. Now I know that I wasted many years walking in the wrong direction. Most people don't even know...


Outstanding & very true information you shared my friend. I too traveled the wrong way for many years before finding my way. It is right before our eyes, but not always clear. Your post are much enjoyed by this Trader. Keep sharing my friend. Others will soon see the light too.
  • #507
  • Apr 14, 2019 7:13am Apr 14, 2019 7:13am
Thank you Davit for a very interesting thread. A number of people seem to use it very successfully.
  • #508
  • Apr 27, 2019 3:36pm Apr 27, 2019 3:36pm
Attached Image (click to enlarge)
Click to Enlarge Name: e-u.PNG Size: 67 KB
eur usd custom 360m
  • #509
  • Edited 2:28pm May 9, 2019 4:34am | Edited 2:28pm
Correlated currencies revert to mean when paired.

Currency strength meter highlights correlations. The meter is most predicitve after a market'event' (a type of stress test):

Example:

After London open (event) it became clear CHF and JPY were highly correlated:

Attached Image


Allowing for multiple scalping opportunites:

Attached Image (click to enlarge)
Click to Enlarge Name: Chart.JPG Size: 47 KB
  • #510
  • May 27, 2019 9:46am May 27, 2019 9:46am
Quoting alphaomega
Hi, and thanks for stopping by! Regardless if you're new or old member of the forum, keep reading here and I promise you will not be disappointed! From my point of view this thread is the natural continuation of the thread "Higher Edge within a Single Candlestick” and here I intend to share with you some of my discoveries regarding this concept. As I commented in the original thread, I believe that we can create and apply multiple strategies only by using the information which is...
It's been a while since we've heard from you AO. Are you still trading this strategy? Keen to find out your progress!
  • #511
  • Edited 2:47pm Jun 10, 2019 6:27am | Edited 2:47pm
It's relatively easy to achieve gains from random movements if there's only a slight tendency to revert to the mean. Simulating an instrument with a mean exchange rate of 1.0 and random variations and applying a very simple "sell when quote rose by 0.3%" and "buy when quote fell by 0.3%" yields a result like the one below.

Attached Image (click to enlarge)
Click to Enlarge Name: rand_meanret1.png Size: 44 KB

(btw it's intersting to see how many well known patterns appear in such a randomly simulated price series: trend channels, s+r lines, shs, ...)

Doing it 1000 times and overplotting the equity curves shows that it is actually quite a safe approach, more so when averaging ovear a number of instruments.

Attachment g


The disadvantage is that you can't leverage too much, otherwise the thing explodes.
Attached Image (click to enlarge)
Click to Enlarge Name: 1000equities.png Size: 55 KB
  • #512
  • Jun 10, 2019 7:11am Jun 10, 2019 7:11am
Quoting mfeldt
It's relatively easy to achieve gains from random movements if there's only a slight tendency to revert to the mean. Simulating an instrument with a mean exchange rate of 1.0 and random variations and applying a very simple "sell when quote rose by 0.3%" and "buy when quote fell by 0.3%" yields a result like the one below. {image} (btw it's intersting to see how many well known patterns appear in such a randomly simulated price series: trend channels, s+r lines, shs, ...) Doing it 1000 times and overplotting the equity curves shows that it is actually...
Looking at the dips in equity - are you applying a grid-based approach similar to what AO does?
  • #513
  • Jun 10, 2019 8:16am Jun 10, 2019 8:16am
The original rules that AO set in the beginning of the thread has an edge. I'm not sure how many of you have been able to successfully adopt some of the ideas here into your own strategies. If you find yourself stuck, do not give up. It does not seem like the thread has any meaningful discussion left, I have taken my own studies outside of this forum but for whoever is still on this, keep at it. You can DM me if you would like to discuss further.

All the best guys!
Attached Image (click to enlarge)
Click to Enlarge Name: eu.JPG Size: 590 KB
  • #514
  • Jun 11, 2019 6:19am Jun 11, 2019 6:19am
Quoting carddard
{quote} Looking at the dips in equity - are you applying a grid-based approach similar to what AO does?
It's really as simple as I wrote above and not optimized in any way. Also, it's not tied to any specific hour as AO's original scheme, it keeps going on forever whenever price moved by another 0.3% in any direction. So yes, it's grid-based somehow, but the grid is newly set up each time a trade has been made.
  • #515
  • Jun 14, 2019 1:48am Jun 14, 2019 1:48am
I made it to Page 16 tonight. I have to get to bed though. I have been testing Money management using a coin flip simulator. I think people get it wrong by only focusing on R:R and going 1:2 1:3 etc. That can maximize an effective system, but you have to have effective control over money flow itself regardless of increased R:R. All that to say.. I'll catch up on the thread, and maybe we (Mfeldt and Carddard) can add some better money management to this and see what it does to the equity curve?

If you give me a 50/50 system, I can make money with it without increasing R:R.
  • #516
  • Jun 14, 2019 3:29am Jun 14, 2019 3:29am
To me, this is the key to success. Making money from coin flips is safe + sound. Imagine betting on a stock market index: You can be sure it will not fall to zero. You can flip a coin at each point and decide whether it will rise by a certain percentage, or fall. Now you can e.g. always enter with 1% of your equity when the index drops by 1%, and pull out 1% when it rises by 1%. This system will always make money and never fail! The problem, is you can not leverage, and most of the time you'll make less money than with a simple buy+hold!

Now you can try to come up with something smart that allows shorting, but it will naturally be less safe as upwards there's no limit for the index! Or you can go to forex, where you essentially trade (at least) the ratio of two indexes (aka currencies). The ratio will also never fall to zero, but here you can be sure that also the reverse ratio will also not fall to zero, i.e. there now is effectively also a limit to any upward movement.

Now you need to find a good model to generate movements from coin flips, and then play with MM parameters (e.g. leverage, entry+exit sizes, hedging if you like, ...). Once you find something that makes decent money, you can pretty safely enter the market without going through any test on real data (which, in contrast to thousands of generated series will always be only one specific realization!).

Attached Image (click to enlarge)
Click to Enlarge Name: alLo2Zj.gif Size: 78 KB
  • #517
  • Jun 15, 2019 12:40am Jun 15, 2019 12:40am
Are you guys not using those cones anymore? Did anyone ever code those for mt4?
  • #518
  • Jun 24, 2019 7:28am Jun 24, 2019 7:28am
I ve an other mean reversion coded....if any hardcore testers are here....
Vucking good EA coder... https://t.ly/AZjRM
  • #519
  • Jun 24, 2019 9:16pm Jun 24, 2019 9:16pm
Sure, send it over. Working on a couple of related projects atm, so we might be able to use it or make it better.
  • #520
  • Jun 24, 2019 9:32pm Jun 24, 2019 9:32pm
Quoting reteid2222
I ve an other mean reversion coded....if any hardcore testers are here....
Share it here, reteid
0 traders viewing now