Goblin "BiPolar" Edition - page 56

New comment
 

Mikl, try this....

 
Originally Posted by xxDavidxSxx

you can set equity target to 0. Its not crucial.

200:1 leverage.

.01 starting lot size per 1k in account. (.05 risk) recommended on initial deposit

.02 starting lot size per 1k in account (1% risk) forward tested well, but not recommended on initial deposit. Wait till trading off of winnings.

pips=no less than 30

5 max trades

18-23 target

Those are crucial

No more than 3 pairs. If your worried and a nervous type, then run it on less pairs.

Dave, do you seggest to use 0.5% risk because of the rest available margin when positions are open, or just to withstand drawdowns wich occcur when MaxTrades hit? Don't you think it is risky to start 0.2 lot, instead, isn't it better when you have let's say 2k account, to split it into two separate accounts of 1k and continue trading with 0.5 risk? BTW, FiboLotSizeProgression can also be used...

 
xxDavidxSxx:
there is not much modification I personally would do to bipoler. what ever our differences, Bluto did a fantastic job especially considering the how quick he wrote it.

The ATR thing: if a pair's daily ATR became more than (X #) (for me it would be 90) then bipoler would not start any new cycles. And let the current cycles run till they close and discontinue trading untill daily ATR dropped back below (X #).

And...

Equity target: If it could only consider the equity of the pair its placed on. When trading multiple pairs equity target becomes much less affective. Rarely gets used.

I really don't see a need for exacting entrys. If we could use indicators to exact entrys allot more ea's would make money. But having more than one exit method in play increases chances of a profitable exit.

And some one showed me some results on martingale lot sizing 1,1.5,2,3,4,6,12 ect instead of doubling down each time. Kinda like half steppin.

I was trying to make that change my self but cann't figure how.

I dunno if any one can or would consider these mods but there they are for the taking.

The usage of ATR as a filter is probably a good idea.

The builtin money management is insufficient for me, I fix lots per trade and scale them by the current equity, which allows to collect statistics and use the Kelly formula (basically determine percentage of capital to risk and calculate lot size relative to stops). If multiple pairs share an account, its easy to distribute capital usage if you can work with a simple percentage.

I played a lot with different entry signals and filters. It turns out that the simple method of Bipolar is sufficient and the trend filter helps a lot.

 

Example with s/l

Here is a backtest example from 1999-2007 using t/p 10, s/l 10, pips 5 and max trades 5 with only the JMA entry signal on a M5 chart. Note that this example keeps the lot sizes constant at 0.2, 0.2, 0.1, 0.3, 0.4 lots for an 5k account. The result is of course much better if compounding is used.

This example illustrates the power of Bipolar, even if stops are used!

 
alassio:
Here is a backtest example from 1999-2007 using t/p 10, s/l 10, pips 5 and max trades 5 with only the JMA entry signal on a M5 chart. Note that this example keeps the lot sizes constant at 0.2, 0.2, 0.1, 0.3, 0.4 lots for an 5k account. The result is of course much better if compounding is used. This example illustrates the power of Bipolar, even if stops are used!

hi, alassio

i am surprise that you use s/l, could you explain to me how to use it?

when i open a 3 order of serie the first order close to s/l, then goblin closes other orders(continue_opening=false).

do you use to enter signal this part of code ?:

if (MathAbs(jma1 - jma2) / Point > 2.0)

{

if (jma1 < jma2) {SignalOrderType=1;}

if (jma1 > jma2) {SignalOrderType=2;}

}

now , i am testing goblin with variable pips using ATR. ( i've used it on Firebird).

regards

giapel

 

backtest goblin grid v4 on eurusd

good result

giapel

 
giapel:
hi, alassio

i am surprise that you use s/l, could you explain to me how to use it?

when i open a 3 order of serie the first order close to s/l, then goblin closes other orders(continue_opening=false).

do you use to enter signal this part of code ?:

if (MathAbs(jma1 - jma2) / Point > 2.0)

{

if (jma1 < jma2) {SignalOrderType=1;}

if (jma1 > jma2) {SignalOrderType=2;}

}

now , i am testing goblin with variable pips using ATR. ( i've used it on Firebird).

regards

giapel

Using stops is very simple, set LongInitialStop=10,ShortInitialStop=10. The real stops of individual orders will all be set at the same place, which is the price of the last order + Pips+InitialStop. This way, if one order is stopped out, all will eventually be stopped out and closing the trade is what Bipolar does.

Using stops has the advantage that your risk is clearly defined and you can start improving the expected value of a trade. This is impossible if you have to judge the risk of blowing the account and the probability of this event. I prefer having a consistent win/loss rate that allows you to do proper money management and risk of ruin calculations etc.

Yes, I used only the above mentioned signal, you have to experiment a bit with timeframe and lag, but experiments with other signals didn't really improve the result. Only the TrendFilter of Bipolar is useful. However, if there is a consistent expected value of a trade, you want to have as many trades as possible and to compound the gains. That's the way to optimal profit. The TrendFilter improved the results but also considerably cut down the number of trades. So the total result is less profit, but maybe with less risk (drawdown).

I think using ATR will help, whether it is used as a filter or to vary the pips and t/p and s/l values. You only have to properly scale everything, including lot size, in order to still have the same max risk.

 

I forgot to mention that everybody can see the effect of stops to his settings by using the Excel sheet for simulating probabilities and expected value I have attached a few posts before. It also shows the effect of reallocating lot sizes among individual orders in a trade. The doubling up is often not optimal, because it increases the size of a loser quite strongly, but win probability is only improved marginally.

My experiments finally all have been done at lot allocation and money management. I tried to improve entries, but without significant improvement. Maybe if someone comes up with good trade filters, but the basic entry method seems just fine.

 
Trader83:
Dave, do you seggest to use 0.5% risk because of the rest available margin when positions are open, or just to withstand drawdowns wich occcur when MaxTrades hit? Don't you think it is risky to start 0.2 lot, instead, isn't it better when you have let's say 2k account, to split it into two separate accounts of 1k and continue trading with 0.5 risk? BTW, FiboLotSizeProgression can also be used...

I go with .5% to give it plenty of equity on maxing out orders. Because I am running 3 pairs. Max draw down should be what ever the loss is on 5 open orders. (not much)

You can run 1% on 5k using 1 pair. I did with 3 pairs on demo to push the limmits. Worked out well for 3 months.

Might be good split between accounts.

Dave

 
Maji:
I think LotExponent = 1.5 should do that.

I tried changing the 2's here in these lines to 1.5 but didn't do it. I did search for lotExponant and its not there.

Any ideas?

thanks

Dave

1...495051525354555657585960616263...88
New comment