Simple Mean Reversion | Page 11 | 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

  • #201
  • Jul 31, 2018 2:09am Jul 31, 2018 2:09am
Attaching the chart to show my manual trading of averaging. Stop loss line is used to average and not liquidate trades.
Attached Image (click to enlarge)
Click to Enlarge Name: Trading pad for moemntum gbpjpy.jpg Size: 887 KB
Practice makes a person perfect
  • #202
  • Edited 4:33pm Aug 3, 2018 2:20am | Edited 4:33pm
I just implemented the strategy described in Post 1 in MQL5 and did a backtest based on real ticks from 2017-01-01 until 2018-07-30 on GBPUSD,H1 starting with 2000 USD capital:
Attached Image (click to enlarge)
Click to Enlarge Name: Backtest_realticks_1.png Size: 38 KB

I observed that the big losses are mostly due to days that price moves in one direction. We may get better results if we can filter those days.
Trading is the hardest way to make easy money...
  • #203
  • Aug 3, 2018 7:53am Aug 3, 2018 7:53am
Quoting MathTrader7
I just implemented the strategy described in Post 1 in MQL5 and did a backtest based on real ticks from 2017-01-01 until 2018-07-30 on GBPUSD,H1 starting with 2000 USD capital: {image} I observed that the big losses are mostly due to days that price trends in one direction. We may get better results if we can filter those days.
So you took every trade (on every new high/low), no minimum ATR band level or something else?

I try to limit the losses with the hard stop at ATR band level 6 and only take trades in the H4 EMA direction with my 2nd method (magic nr. 9990) for example.

Last week (2018.07.30 - 2018.08.03) wasn't a good week for this method in general, we often had such situations where price trends in one direction the whole day. Therefore the original method from post 1 had its first red week. With my method I got still a good green week (because of Dow)!

9990 = my variant of SMR with 4H MA filter, BE stop, no fixed timed 25%/50% exits but "candlestick jumping stop loss" - see above for more details
9993 = original method from post 1 with only BE stop added.
_
EA trades the broker min lots per symbol (0.01 for forex symbols, 0.10 for dax/dow) with some small kind of martingale (0.01 for first 3 trades, 0.02 for the next two trades and then 0.03 lots until MaxTradesPerSymbolAllowed is reached). Both variants open the first trade only when min level of the ATR band of 3 is reached and also have a hard stop if ATR band 6 is reached! The broker used has fixed spreads per symbol and charges no commissions per trade.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: 2018-08-06 Simple Mean Reversion EA 9990 kw31.png Size: 9 KB
Click to Enlarge Name: 2018-08-06 Simple Mean Reversion EA 9993 kw31.png Size: 10 KB
  • #204
  • Aug 3, 2018 8:03am Aug 3, 2018 8:03am
Quoting MathTrader7
I just implemented the strategy described in Post 1 in MQL5 and did a backtest based on real ticks from 2017-01-01 until 2018-07-30 on GBPUSD,H1 starting with 2000 USD capital: {image} I observed that the big losses are mostly due to days that price trends in one direction. We may get better results if we can filter those days.
nice work.

can you show backtest for EURUSD on the same period?
  • #205
  • Aug 3, 2018 8:48am Aug 3, 2018 8:48am
Quoting alphadude
{quote} nice work. can you show backtest for EURUSD on the same period?
Sure! I will post the results here.
Trading is the hardest way to make easy money...
  • #206
  • Aug 3, 2018 8:54am Aug 3, 2018 8:54am
Quoting Jagg
{quote} So you took every trade (on every new high/low), no minimum ATR band level or something else? I try to limit the losses with the hard stop at ATR band level 6 and only take trades in the H4 EMA direction with my 2nd method (magic nr. 9990) for example. Last week (2018.07.30 - 2018.08.03) wasn't a good week for this method in general, we often had such situations where price trends in one direction the whole day. Therefore the original method from post 1 had its first red week. With my method I got still a good green week (because of Dow)!...
The EA opens a maximum number of trades based on fixed distance (100 points). I will test it with a percentage of the daily ATR value to see the results.
The idea of using ATR band levels plus a trend filter seems valid to me. Good work!
Trading is the hardest way to make easy money...
  • #207
  • Aug 5, 2018 2:06am Aug 5, 2018 2:06am
Some observations when testing the Simple Mean Reversion EA (MT4) from Post # 48

DD is the evil to try and overcome.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 40 KB

Shallow pullbacks need to be managed (somehow)
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot2.png Size: 19 KB

Missed opportunities when PA corrects
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot3.png Size: 28 KB

Small winning cycles relative to larger losing cycles
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot4.png Size: 26 KB

Closures less than optimal (in fairness some of these may be difficult to plan in advance of the fact)
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot5.png Size: 29 KB

Certainly I see a number of areas that need improving, much like the Quantum approach.

Over a small 5 month testing period using default settings, the majority of tested symbols fail.

Commonly it is a low strike rate mated to average losers being double that of the winners.

Masterrmind...............
Master your Mind then Master your Trades
  • #208
  • Aug 5, 2018 6:10am Aug 5, 2018 6:10am
EURUSD H1 TF (10K) from January 1, 2018 to the current date.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 21 KB

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot2.png Size: 21 KB

Not a bad looking curve.

7.7% gain on 4.6% DD over 7 months.

Think about the impact of commision (if applicable) over 1236 trades, it adds up !

EURUSD H1 TF (10K) from March 5, 2018 to the current date.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot3.png Size: 21 KB

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot4.png Size: 22 KB

Also not bad looking

3.6% gain on 4.7% DD over 5 months.

Think about the impact of commision (if applicable) over 875 trades, it adds up !

All settings were default.
Attached Image

Ofcourse by optimizing some interesting results can be achieved.

Nevertheless you need to keep in mind that optimization may only lead to improvements over a known historical data set.

What should be done, is filling in any obvious holes in the strategy so that it shows a higher level of robustness.

The higher the level of robustness, the more likely it is to meet with live data in a real market.

If you go for the big volatile pairs be careful .....
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot6.png Size: 19 KB

ps. that was GN through 2018.

Good trading,

Masterrmind..............
Master your Mind then Master your Trades
  • #209
  • Aug 5, 2018 6:25am Aug 5, 2018 6:25am
Just wondering if any trader here has found any advantage of using Alphaomegas Sqrt(t) formulas ?
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 23 KB

Masterrmind.............
Master your Mind then Master your Trades
  • #210
  • Aug 5, 2018 7:34am Aug 5, 2018 7:34am
Quoting Masterrmind
Just wondering if any trader here has found any advantage of using Alphaomegas Sqrt(t) formulas ? {image} Masterrmind.............
Don't think that there's any edge per se. However, the variance of a (pseudo) random walk increases with the square root of the time. That is what i shows ... where you can expect to see prize after x candles on average ...
  • #211
  • Aug 5, 2018 9:55am Aug 5, 2018 9:55am
Quoting alphadude
{quote} nice work. can you show backtest for EURUSD on the same period?
Here is the backtest results based on real ticks:
Attached Image (click to enlarge)
Click to Enlarge Name: Backtest_realticks_2.png Size: 37 KB
Trading is the hardest way to make easy money...
  • #212
  • Aug 5, 2018 10:19am Aug 5, 2018 10:19am
Quoting MathTrader7
{quote} Here is the backtest results based on real ticks: {image}
what is the date range?
  • #213
  • Edited 12:06pm Aug 5, 2018 10:46am | Edited 12:06pm
Quoting alphadude
{quote} what is the date range?
same as the previous backtest, from 2017-01-01 until 2018-07-30 on EURUSD,H1 starting with 2000 USD capital.
Trading is the hardest way to make easy money...
  • #214
  • Aug 5, 2018 2:01pm Aug 5, 2018 2:01pm
A backtest based on real ticks from 2017-01-01 until 2018-07-30 on EURGBP,H1 starting with 2000 USD capital:
Attached Image (click to enlarge)
Click to Enlarge Name: Backtest_realticks_EURGBP.png Size: 35 KB

As expected, the strategy seems to work a bit better on ranging currency pairs.
Trading is the hardest way to make easy money...
  • #215
  • Aug 5, 2018 9:11pm Aug 5, 2018 9:11pm
Quoting AntiCre
{quote} Don't think that there's any edge per se. However, the variance of a (pseudo) random walk increases with the square root of the time. That is what i shows ... where you can expect to see prize after x candles on average ...
Perhaps we'll have to wait and see what other tests get posted here and the commentary that follows.

Masterrmind.............
Master your Mind then Master your Trades
  • #216
  • Aug 6, 2018 4:25am Aug 6, 2018 4:25am
Quoting MathTrader7
A backtest based on real ticks from 2017-01-01 until 2018-07-30 on EURGBP,H1 starting with 2000 USD capital: {image} As expected, the strategy seems to work a bit better on ranging currency pairs.
Nice work; what is the exit criteria?
  • #217
  • Aug 6, 2018 5:51am Aug 6, 2018 5:51am
Quoting alphadude
{quote} Nice work; what is the exit criteria?
As suggested by alphaomega in Post 1, the EA closes all open positions at 25% and 50% retracement depending on the time of day (one cycle). I am going to make some modifications to the original strategy. In case of any improvement, I will post the results here.
Trading is the hardest way to make easy money...
  • #218
  • Aug 6, 2018 8:19am Aug 6, 2018 8:19am
Quoting RondaRousey
{quote} take your profits quickly and hold on to your losses longer=mean reversion Avg losses always greater than avg profit=mean reversion Novice traders like mean reversion because of high win rate 60-90% Why? because its not a mental beating ie for the weak minded
Perhaps some novice traders like mean reversion but they also might like martingale or trend trading or swing trading or any other type of trading.

Certainly do not recommend it as a standalone trading strategy.

Plenty of highly intelligent minds are parked in this space though.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 29 KB

You never know you just might learn something useful.

Masterrmind............
Master your Mind then Master your Trades
  • #219
  • Edited 8:47am Aug 6, 2018 8:31am | Edited 8:47am
Quoting RondaRousey
{quote} take your profits quickly and hold on to your losses longer=mean reversion Avg losses always greater than avg profit=mean reversion Novice traders like mean reversion because of high win rate 60-90% Why? because its not a mental beating ie for the weak minded
what matters is not the win rate, or the average win/loss. what matters is the PNL

I certainly beleive that novice traders like to use tight stop losses; and be the broker's bitc*es
  • #220
  • Aug 6, 2018 7:36pm Aug 6, 2018 7:36pm
I would rather learn a highly profitable trading system from the stupid than an unprofitable trading system from the highly intelligent.

Never be fooled by intelligence.

My 2 bob worth

You came here for the truth and let me unveil it for you
0 traders viewing now