Simple Mean Reversion | Page 15 | 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

  • #281
  • Aug 31, 2018 12:38am Aug 31, 2018 12:38am
Quoting Masterrmind
{quote} One word of caution I'll give you though Mizi is be sure you have a reasonably solid exit strategy, by that I mean one that protects the loss on your account when price trends away and does not come back. The strong trend will eventually strike and if the traders mindset is something like this; "It'll be ok price will reverse soon, its always reversed for me over a hundred times in a row... I'll keep the trades running..." Then the fact is that the account will eventually end very very badly. Long term account survival depends on having...
Kudos for a well written post. I am a huge fan of multi-pair trading so this resonates with me.

Just a quick comment on the following....
Reducing the risk by spreading margin across several pairs (Sum [Pair A floating P&L + Pair B floating P&L + ... + Pair N floating P&L]);

Even with single pair multi-position strategy, I am a firm believer that we should always base position size (for %risk position sizing) from "Available Margin". Open 1st position, x% of accountequity()...free margin depletes
Open 2nd position, x% of free margin

The beauty of this is a more balanced exposure and if you can visualize this, a far "optimized" use of margin..... you can literally squeeze every penny worth of free margin under a certain threshold to avoid MC. This model is far superior than a higher %risk bigger lot position sizing.
Staying in my lane...
  • #282
  • Aug 31, 2018 12:47am Aug 31, 2018 12:47am
Quoting VEEFX
{quote} Kudos for a well written post. I am a huge fan of multi-pair trading so this resonates with me. Just a quick comment on the following.... Reducing the risk by spreading margin across several pairs (Sum [Pair A floating P&L + Pair B floating P&L + ... + Pair N floating P&L]); Even with single pair multi-position strategy, I am a firm believer that we should always base position size (for %risk position sizing) from "Available Margin". Open 1st position, x% of accountequity()...free margin depletes Open 2nd position, x% of free margin The...
So all hope is not lost. This post is actually a good one! I was beginning to think you had become all jaded and scorned.
If you are good at something, never do it for free--Joker
  • #283
  • Aug 31, 2018 12:55am Aug 31, 2018 12:55am
Quoting jmn5611
{quote} So all hope is not lost. This post is actually a good one! I was beginning to think you had become all jaded and scorned.
hehe... I belong to the camp of "Risk increases astronomically the moment we enter in the market". Risk is not fixed even if we measure the range/distance of price in pips. We need to do more to contain risk without dependence on price itself. Price will always go wherever the f*ck it wants to go for one pair... but price will never go up or down for all pairs. Using this method builds a fabulous cushion or shock absorbers by considering what market HAS to DO.
Staying in my lane...
  • #284
  • Aug 31, 2018 1:00am Aug 31, 2018 1:00am
Quoting VEEFX
{quote} Kudos for a well written post. I am a huge fan of multi-pair trading so this resonates with me. Just a quick comment on the following.... Reducing the risk by spreading margin across several pairs (Sum [Pair A floating P&L + Pair B floating P&L + ... + Pair N floating P&L]); Even with single pair multi-position strategy, I am a firm believer that we should always base position size (for %risk position sizing) from "Available Margin". Open 1st position, x% of accountequity()...free margin depletes Open 2nd position, x% of free margin The...
Good points raised here.

What I was leaning towards was spreading the margin (not the risk) across a basket of trades and not just the one pair.

The interesting part though is that at some stage there will also come a time when more than 1 pairs floating loss will also align with at least 1 to 2 other pairs floating loss and create a larger cycle of DD.

This is accelerated (or should I say more commonly noticed) the longer the cycle of trades remains active in the market.

I've seen this more than once on several test runs, demo as well as several in live mode.

Regarding the higher lots, these were applied only in sections of the Asian trading session and only on certain favored pairs.
So its a unique setup and certainly not to be played out in general terms or the market will eat you up.

The weak link in all of this is running out of free margin (or the trader for allowing it to happen in the first place), then the choices are gone and so too the account thereafter unless the market reverses and takes away your pain but rarely does this occur.

Plenty of good points raised by you though.

ps. The font in red above should be re-written to say spreading the margin NOT spreading the risk.... ha ha.

Masterrmind..............
Master your Mind then Master your Trades
  • #285
  • Aug 31, 2018 1:07am Aug 31, 2018 1:07am
Quoting VEEFX
{quote} hehe... I belong to the camp of "Risk increases astronomically the moment we enter in the market". Risk is not fixed even if we measure the range/distance of price in pips. We need to do more to contain risk without dependence on price itself. Price will always go wherever the f*ck it wants to go for one pair... but price will never go up or down for all pairs. Using this method builds a fabulous cushion or shock absorbers by considering what market HAS to DO.
Aint that one so true.

Masterrmind...............
Master your Mind then Master your Trades
  • #286
  • Aug 31, 2018 1:08am Aug 31, 2018 1:08am
Quoting Masterrmind
{quote} Good points raised here. What I was leaning towards was spreading the risk across a basket of trades and not just the one pair. The interesting part though is that at some stage there will also come a time when more than 1 pairs floating loss will also align with at least 1 to 2 other pairs floating loss and create a larger cycle of DD. This is accelerated (or should I say more commonly noticed) the longer the cycle of trades remains active in the market. I've seen this more than once on several test runs, demo as well as several in live...
Thanks... are you a believer of "measuring currency correlations" in your trade entry decisions? Just curious because, being a non-believer of Price itself, I do not measure or calculate anything based on price. A price as a dot in infinite space that keeps moving every tick by tick simply cannot be measured or contained or captured or trapped or cornered or pivotted or fibbed or trendlined or bands or waved or... you name it.. using any calculation of price itself.
So the classic problem arrives with "how to balance or select pairs to extract more when they move in our direction and protect us from drawdowns when they move against us. We simply cannot rely on correlations to contain risk as part of our portfolio basketing. It will always results in ziggy zaggy shitty equity curve imo. Any suggestions? if you have one, I will share mine :-)
Staying in my lane...
  • #287
  • Aug 31, 2018 1:11am Aug 31, 2018 1:11am
Quoting VEEFX
{quote} hehe... I belong to the camp of "Risk increases astronomically the moment we enter in the market". Risk is not fixed even if we measure the range/distance of price in pips. We need to do more to contain risk without dependence on price itself. Price will always go wherever the f*ck it wants to go for one pair... but price will never go up or down for all pairs. Using this method builds a fabulous cushion or shock absorbers by considering what market HAS to DO.
There are only 3 effective ways to diversify for the purpose of risk reduction.

1. Apply the same strategy to multiple uncorrelated markets.
2. Apply multiple different strategies to one market.
3. Apply multiple different strategies to multiple different markets where each strategy is optimized individually.

I cant think of any other way to reduce risk and volatility in the equity.
Maybe we can use options as a form of insurance....like alternative to disaster stops. But this depends on the strategy.
  • #288
  • Aug 31, 2018 1:18am Aug 31, 2018 1:18am
Quoting alphaomega
{quote} There are only 3 effective ways to diversify for the purpose of risk reduction. 1. Apply the same strategy to multiple uncorrelated markets. 2. Apply multiple different strategies to one market. 3. Apply multiple different strategies to multiple different markets where each strategy is optimized individually. I cant think of any other way to reduce risk and volatility in the equity. Maybe we can use options as a form of insurance....like alternative to disaster stops. But this depends on the strategy.
I fall under #3 with a twist. The twist being "using the exact same entry applicable to a universal principle or rule at applies to all price movements". I am also not a believer of any such thing as a "strategy optimization"... optimization by definition means you are trying to fixate a circle inside a square or vice versa.. neither the circle (radius/range) nor a square (top or bottom) can be fixed as part of any strategy parameter optimization for it to last longer so that is ruled out in my research
Staying in my lane...
  • #289
  • Aug 31, 2018 1:25am Aug 31, 2018 1:25am
Quoting alphadude
{quote} Hi TRD, nice work there. you points are correct; and mine are too. The reason I asked you about live trading; is to ask for your experience on slippage. I have never traded GBPCAD myself. my experience live trading pairs other than EURUSD; is high slippage when you need it. Usually this occur during closing of positions; especially when the size is slightly large. In some occasions; I get slipped 20 pips and even 40 pips on pairs like AUDUSD, NZDUSD, USDCAD and EURJPY The backtest (using real ticks) shows fantastic results; yet live result...
Hi alphadude,

slippage is part of the game and it depends on your size. 1 pip for 10 lots trade is nothing to worry about. If you get slipped 20 pips, then you probably didnt notice the spread which was same size or you have to double check your broker or account type. Fx is liquid enough even on retail level.

"Not to mention high spread during quite times."
I think i clarified that in my last post

"the spread is about 4 times the spread of EURUSD. Opening and closing a position in GBPCAD will cost me 8 times the cost of trading EURUSD"
That is true, however GC has unique edge compared to EU. And probably because of spread most traders will refrain from analysing its behavior. Which is good for me

"In your case; you appear to trade medium term; where spread may not be a problem"
It depends on the market. Some trades close in 1 min, some in 1 month. So am i trading long or short term.
  • #290
  • Aug 31, 2018 1:28am Aug 31, 2018 1:28am
Quoting jmn5611
{quote} So all hope is not lost. This post is actually a good one! I was beginning to think you had become all jaded and scorned.
Nothing new. Actually, I was completed jaded trading manually. Going to automation route opened my eyes to the world of possibilities (but still at the mercy of the darned broker) It's all included in my bookmarked (or perhaps deleted) posts in the past. read up on my "TradingCostMultiple" and many other ideas all shared throughout my journey to get the F out of here one day from this revolving door :-)
Staying in my lane...
  • #291
  • Edited 3:21am Aug 31, 2018 1:32am | Edited 3:21am
Quoting RondaRousey
{quote} The avg loss is close to 40 times the average profit??? This would scare the shit out of me regardless of 99% winrate Drawdown is not problem if the recovery is quite fast however if it takes years then that is a problem Would not use mean reversion as a standalone strategy
By default its set to 100 times, but because of adaptive TP strategy, the algo managed to bring it down to 40. As mentioned recovery takes 2-6 months.
The obvious choice is usually wrong

TRD
  • #292
  • Aug 31, 2018 7:55am Aug 31, 2018 7:55am
Quoting VEEFX
{quote} Thanks... are you a believer of "measuring currency correlations" in your trade entry decisions? Just curious because, being a non-believer of Price itself, I do not measure or calculate anything based on price. A price as a dot in infinite space that keeps moving every tick by tick simply cannot be measured or contained or captured or trapped or cornered or pivotted or fibbed or trendlined or bands or waved or... you name it.. using any calculation of price itself. So the classic problem arrives with "how to balance or...
I have advocated using currency correlations on entries in some strategies but not in all.

I think you need to consider many other factors as well when it comes to entry techniques.

Being a transparent trader I have no problem outlining a trade setup before the fact and following it through to completion.
If this occurred in this thread then it would be based on at least some resemblance to simple mean reversion out of courtesy to the Th Op.

I tend to use the word, confluence alot, meaning that I'm more of a trader that seeks to make an entry from an assumed higher probability setup.

But I may apply a relatively high risk to some setups especially scalping type strategies where I closely monitor the SL pips as a % of account.

Since I trade multiple strategies it would probably not be the right place here in this thread to elaborate on them.

But I will say this, everything in trading is relative period.

When trades move against us and accumulate DD something needs to happen.
Either hedge, drop some positions, add additional positions apply some elaborate techniques or two or sit and hope that things change for the good.
I go for the elaborate techniques(s).

Using correlations to contain risk might offer some relief but it is limited in its effectiveness mainly because it can change in time and the trader might be in a period of time where the effectiveness of correlations isn't so effective and is basically out of sync.

If its related to trading multiple pairs in a basket then it may have its uses but its still not a simple matter.
There are numerous other competing factors at work in this space and when correlations diverge and converge from one another over time the account margin levels will also fluctuate.

An immediate 'band aid' effect is to drop several opposite direction trades to ease the escalating DD but this is only a temporary measure and to be very effective the account requires plenty of free margin otherwise its usually a less than favorable ending.

A worthwhile exercise is to intentionally run a demo account or more into serious DD say over 50% and then attempt to recover the account.
I wonder how many traders have tried to do this ... you will learn an incredible amount in a very fast time.

I believe there is a way to smooth out an equity curve but as I've mentioned in a previous post I'm not going to divulge it.
Testing continues and in any case I've been working on it for a very long time.
The greater the number of pairs the more one must rely on an automated approach because I've tried it in manual mode and its a beast to handle.

I would certainly be interested in your ideas about controlling DD in multiple pairs or singles for that matter but not sure you'll share it since I'm not about to share mine.

I am willing to drop some hints though and in any case I am searching for one more piece to the puzzle and my hope is that it might show up somewhere in this thread.

Do I really need this though for my trading ? The answer is ofcourse no.
However, I do need it to conquer this Quantum trading style if I wish to drive a 1K account into 10K .... thats why I need it, to finish this goal.

Sometimes I wish I never had made this goal but now its become an obsession, in fact I'm also hoping that some other trader does it.
One thing is forsure and that is plenty has been learnt along the way.

Perhaps the learnings can later be shared here once the goal is completed.

Masterrmind..............
Master your Mind then Master your Trades
  • #293
  • Aug 31, 2018 11:05am Aug 31, 2018 11:05am
Quoting Masterrmind
{quote} I have advocated using currency correlations on entries in some strategies but not in all. I think you need to consider many other factors as well when it comes to entry techniques. Being a transparent trader I have no problem outlining a trade setup before the fact and following it through to completion. If this occurred in this thread then it would be based on at least some resemblance to simple mean reversion out of courtesy to the Th Op. I tend to use the word, confluence alot, meaning that I'm more of a trader that seeks to make an...
I will look into your quantum thing but it is definitely possible under the right conditions. I did something similar in May across four 1K accounts shared here. I will remove the screenshot soon. Hint: two were long-only and the other two were short-only with two strategies (diverge and converge) across 28 pairs.

Edit- I am a demo trader for life for the same reason you mentioned above. I run 5 instances in parallel with live :-)
Staying in my lane...
  • #294
  • Sep 4, 2018 5:19pm Sep 4, 2018 5:19pm
And to backup my opinion about GBPCAD that its best mean reversion pair from all, i created this video. The style is to be seen from the equity and stats. From 1k to 20M in 9 years:

Inserted Video


TRD
  • #295
  • Sep 5, 2018 12:11am Sep 5, 2018 12:11am
Quoting TheRealDeal
And to backup my opinion about GBPCAD that its best mean reversion pair from all, i created this video. The style is to be seen from the equity and stats. From 1k to 20M in 9 years: https://youtu.be/94fZeM9oTXI TRD
The test is done with control points on 30 min. chart! This is not accurate result.
Find some real historical tick data, and make the test again.
Ideally you need real ticks + spread data. But if your trading model is based on higher timeframe above M15, then even M1 bars will produce a more realistic picture.
  • #296
  • Sep 5, 2018 11:44am Sep 5, 2018 11:44am
Quoting alphaomega
{quote} The test is done with control points on 30 min. chart! This is not accurate result. Find some real historical tick data, and make the test again. Ideally you need real ticks + spread data. But if your trading model is based on higher timeframe above M15, then even M1 bars will produce a more realistic picture.
Agee. Tick data has more accurate results, but for this kind of test deviation is minor. For the sake of short video I did that.
In fact I don’t use much mt4 anymore for automated testing as I find it less flexible and sometimes produces weird results. And as for test data I use fxcm and dukascopy high quality tick data.

Trd
  • #297
  • Sep 5, 2018 12:23pm Sep 5, 2018 12:23pm
Quoting TheRealDeal
{quote} Agee. Tick data has more accurate results, but for this kind of test deviation is minor. For the sake of short video I did that. In fact I don’t use much mt4 anymore for automated testing as I find it less flexible and sometimes produces weird results. And as for test data I use fxcm and dukascopy high quality tick data. Trd
dukascopy is quite good data actually. I use to test as well. It goes as far back as 2008 for some instruments.

All my testing are done with real tick; I long ago stoped using control points. Better to be confident 100% before launching the EA to live trading.
  • #298
  • Edited 12:39pm Sep 5, 2018 12:27pm | Edited 12:39pm
Quoting TheRealDeal
And to backup my opinion about GBPCAD that its best mean reversion pair from all, i created this video. The style is to be seen from the equity and stats. From 1k to 20M in 9 years: https://youtu.be/94fZeM9oTXI TRD
Have you used real spread or fixed?

I highly suggest you use real spread; as you will see big difference in results

when I test GBPCAD with real ticks I get negative curve all the way down; as opposed to GBPUSD for example which is in the positive
  • #299
  • Sep 5, 2018 12:40pm Sep 5, 2018 12:40pm
Quoting alphadude
{quote} dukascopy is quite good data actually. I use to test as well. It goes as far back as 2008 for some instruments. All my testing are done with real tick; I long ago stoped using control points. Better to be confident 100% before launching the EA to live trading.
Lately I am too paranoid about any commercial or free software. I am writing my own now
  • #300
  • Sep 5, 2018 12:48pm Sep 5, 2018 12:48pm
Quoting alphadude
{quote} Have you used real spread or fixed? I highly suggest you use real spread; as you will see big difference in results when I test GBPCAD with real ticks I get negative curve all the way down; as opposed to GBPUSD for example which is in the positive
I have an agreement with my broker about fixed spread, so I test only with fixed. Ofc in order to be able to participate also during rollover when the spread can go as high as 50 pips.
MT4 backtester cannot be trusted. I also noticed about it that depending on the strategy and type of test, results can deviate dramatically
Suggest you to move also to other software or create your strategy by manual testing and do only forward tests
The past doesn’t really matter much
2 traders viewing now