Lot sizing strategies and scaling out.

New comment
 

Yesterday I posted about how to prevent whipsaw and got some excellent advice about using Standard Deviation and one of his indicators. ('Methods for preventing Whipsaw...')

Today, I am curious about lot sizing and how to keep otherwise profitable trades from turning into unprofitable trades due to indicator lag. (Right... something everyone wants to know... :))

Say I am trading end to end using a moving average with a fixed lot size and when the market is trending I make money, but when the market is ranging I lose money, because the indicator is always one or two steps behind.

Question is this (2 parter): 1.) Using a fixed lot size (say of 0.1) I lose money on trading end to end with a moving average. Would it make sense to increase the lot size and close all but (say 0.1) at the first sign of profit and thus hopefully buffer the losses I incur during ranging markets, or is this a dumb idea? 2.) What are some common strategies for scaling out of positions, or is scaling out a dumb idea as well?

Thanks all!

Mike

 

In my opinion, scaling is for the really big accounts when you are maxed out on the number of lots you can make in a single trade. But then the flip side to that notion is you need know the stochastics of your scaling algorithm long before you get maxed out on number of lots for a single trade.


Yesterday I provided some interesting statistical info regarding a scaling EA written by reubencouto. See 'EA Kanguru 4.6 - Promissing results!!'


If you treat the scaled trades as a single trade with respect to the size of your account (eg the aggregate value of 10 scaled trades represents 2% of your account). Rosh's article 'New article: On the Long Way to Be a Successful Trader - The Two Very First Steps' suggests 2% (ie 1/50 of your account) which I call "Trade Size = 50" provides a very stable EA if the winning percentage and Reward-Risk ratios are properly proportioned. The Trade Size you use is really a function of your personal taste for drawdown. If the drawdown you are currently experiencing, is too risky for your personal taste simply increase the Trade Size [ie 1/(Trade Size) ] Again, Trade Size is the percent of your account that each trade represents.


From a purely statistical perspective, in a Trade Simulator that I created in Excel, it indicates that if the Preliminary Profit Factor is over 3.0 you can obtain a pretty stable EA with Trades Sizes as low as 25.

The formula that I use for calculating Preliminary Profit Factor is: PPF = Winning Pct/Losing Pct x Reward/Risk


One thing that really impacts the stability of an EA is having a Winning Percent over 65% before attempting to reduce the Trade Size below 50.

 
FXtrader2008 wrote >>

In my opinion, scaling is for the really big accounts when you are maxed out on the number of lots you can make in a single trade. But then the flip side to that notion is you need know the stochastics of your scaling algorithm long before you get maxed out on number of lots for a single trade.

Yesterday I provided some interesting statistical info regarding a scaling EA written by reubencouto. See 'EA Kanguru 4.6 - Promissing results!!'

If you treat the scaled trades as a single trade with respect to the size of your account (eg the aggregate value of 10 scaled trades represents 2% of your account). Rosh's article 'New article: On the Long Way to Be a Successful Trader - The Two Very First Steps' suggests 2% (ie 1/50 of your account) which I call "Trade Size = 50" provides a very stable EA if the winning percentage and Reward-Risk ratios are properly proportioned. The Trade Size you use is really a function of your personal taste for drawdown. If the drawdown you are currently experiencing, is too risky for your personal taste simply increase the Trade Size [ie 1/(Trade Size) ] Again, Trade Size is the percent of your account that each trade represents.

From a purely statistical perspective, in a Trade Simulator that I created in Excel, it indicates that if the Preliminary Profit Factor is over 3.0 you can obtain a pretty stable EA with Trades Sizes as low as 25.

The formula that I use for calculating Preliminary Profit Factor is: PPF = Winning Pct/Losing Pct x Reward/Risk

One thing that really impacts the stability of an EA is having a Winning Percent over 65% before attempting to reduce the Trade Size below 50.

FXtrader2008 -

The forum (and I) would be wise to pay attention to what you have to say. Thank you.

-Mike

 

Mike,


Thank you for the compliment.


An additional thought just came to mind regarding "scaling in" that had never occurred to me. I've never given "scaling in" much thought until I evaluated the results of Ruben's scaling EA ('EA Kanguru 4.6 - Promissing results!!') and recognized the beauty of his EA. As a result of scaling in, he was actually able to eliminate 2 losses that would have occurred if he had entered the trade with one single large trade on the trend signal that he utilizes.


On the premise just mentioned, scaling in offers a significant benefit, if you add another trade to an existing one ONLY if the existing trade is showing a profit, otherwise do not add to a bad position. Let the little first trade run its losing course and wait for a new trade signal before starting the process over again. That way you really diminish the size of your losses and enhance the size of your wins.


I'm glad you started this thread, it has been helpful for me as well.


Cheers

 

You are very welcome.

Here's a screenshot of something I'm working with.

The indicator is a the thin white line, the others are there for effect. The period for the indicator is '1'.

The three middle trades would've been losers, but using a lot size of '1' I close '0.1' when the indicator is flat until '0.1' remains in the order. (Indicator works with High/Low and StdDev)

- Mike

 

Exotic algorithm. Took me a while to actually see the "thin" white line because the thick one was the only thing catching my attention. I'm curious: do you run this algorithm on a 1H or 4H chart? To me it seems there would not be enough room for a profit on a chart under 1H.


As stated in my previous post I've never given scaling "in" or "out" much though. However, in the last few days the wheels have been turning. In your opinion, from a trade management or perhaps even a money management perspective, do you think scaling "in" or "out" is a technique that enhances the bottom line? If so how much? 25%, 50%, 75% or more?


You've got me hooked. I toyed around with a scaling "in" concept in the mental playground most of yesterday. My guestimation calcs suggest the potential enhancement to the bottom line is well over 75%.


Today I'm going to code up a routine and tie it into my EA and see what happens.


Cheers

 

I started trading this algorithm on the 15 minute charts, 1H and 4H charts and you are right, under 1H the algorithm gives back most of what it earns. The 1H and 4H do not.

I'm not confident enough to say how scaling in/out affects the bottom line w/o running some numbers. I had tried scaling in/out when I was a newbie a year ago and found myself buying low and high and selling low and high, meaning, not knowing when to stop scaling. I moved onto other strategies at the time, as I recall.

Late Friday night, while working on an indicator that automatically optimizes period-based indicators before and after trades, and after reading your response I posted this post on my blog: http://arachnode.net/blogs/arachnode_net/archive/2009/06/13/automatic-optimization-of-forex-indicators.aspx (Crazy idea, I know, but the indicator will tell you which period is the best for any indicator you select, and will display the profit and profit factor... why spend time trying to code a SMA on H4 with a period of '4' when the best period is actually '84'? :))

The answer to how/when to scale in/out likely lies in a bipolar indicator, like RSI, CCI, etc. In my blog post above, the RSI indicator for EURUSD H1 works best at period '10', but also fits nicely trading at the '30' and '70' levels, which would provide, perhaps, a scale in/out point at '50'. Meaning, whether scaling in/out, your scaling should conclude at '50'.

I'm currently focusing on finishing up the automatic optimization code, and then will likely revisit scaling.

Did you try scaling in/out in any of your EA's?

New comment
Reason: