July 31

By Kevin Davey

July 31, 2017

Many traders have a favorite method for trading, and they rarely deviate from that plan, regardless of what they are trading. One method many people like is what I call the “peel off” method. In its simplest form, the trader starts out with two contracts, exits the first at a small profit (the “peel off”), and then holds the second as a “runner,” going for big profits. In some cases when the first contract is exited, the second contract is then modified to have a breakeven stop.

Of course, for every person who uses this method, another person thinks the approach is complete hogwash. Personal feelings aside, I always like to do what the data tells me; if the performance improves with a certain method, then I’ll use it. If it makes it worse, I’ll discard the idea. With that approach in mind, let’s examine the “peel off” method for a few actual trading systems.

### Analysis Approach

For a first cut in the evaluation, we’ll keep things simple. First, we will run our baseline case, two contracts traded for each trade. In the second case (the “peel off” case), we will enter each trade with two contracts, and exit the first with a profit of ProfSTS dollars.

In Tradestation, assuming Condition 1 is our entry criteria, and Condition 2 is our exit criteria, the code for long entries would look like this:

Baseline Code

``If Condition1 then buy 2 contracts next bar at market; ``If Condition2 then sell 2 contracts next bar at market;``

For the “peel off” case, the code would look like this:

“Peel Off” Code

``Input: ProfSTS(500); ``//profit threshold for exiting the first contract ``If Condition1 then buy 2 contracts next bar at market; ``If CurrentContracts > 0 then``begin ``   //prevents division by zero errors ``   If OpenPositionProfit/CurrentContracts > ProfSTS and CurrentContracts=2 then ``      begin ``      Sell 1 contract next bar at market; ``      End; ``   End;``If Condition2 then sell all contracts next bar at market;``

Since we are defining ProfSTS as an input, we will be able to optimize it. This might give us some insight into when the “peel off” method is superior.

### Evaluation Approach

Since each new entry always starts with two contracts, we can compare the Return on Account of the two cases. Return on Account includes Net Profit and Maximum Drawdown in its calculation, so we have risk and reward covered in one number.

### Test Strategies

For this evaluation, we will test three unique strategies. These are all strategies that I personally trade and were developed using my typical walk-forward analysis approach. All of the strategies are different:

1. An intraday 1-minute bar ES strategy that has no profit target, and a large stop loss
2. A 30-minute bar Gold swing system, with both a profit target and a stop loss
3. A 360-minute Japanese Yen swing system, with just a stop loss

### Baseline Results

The overall walk-forward performance for each of the strategies, trading two contracts each, is shown below for the period of January 1, 2007 to January 1, 2015.

### In-Sample Optimization Results

To test how the “peel off” scheme works, the data from 1/1/2007 to 12/31/2011 (5 years) will be used to optimize the profit target exit level (input ProfSTS) and then validated on the out-of-sample data from 1/1/2012 to 1/1/2015 (3 years). The optimization will use the maximum Return on Account as its criteria.

Results of the in-sample optimization are shown in the charts below.

For the ES strategy, the “peel off” method always produces results as good as, or better than, the baseline case of holding two contracts throughout the trade. The maximum Return on Account corresponds to a profit target of \$1,500 which produces a Return on Account more than double of the baseline case.

For the GC strategy, sometimes the “peel off” method is better, and sometimes it is not. Based on the optimization, we will use a profit target of \$250 which produces nearly double the Return on Account, compared to the baseline.

For JY, the “peel off” method is almost always worse than the baseline case. Its optimum is at \$3,250 profit level.

### Verification With Out-Of-Sample Results

Now that we have established the optimum profit target to “peel off” the first contract, we can run the results on out-of-sample data from 2012-2015. These results are shown in the table below.

Strategy

Baseline Return on Account

Optimum "Peel Off" Return on Account

Is "Peel Off" Better Or Worse

ES

207%

206%

Same

GC

25%

43%

Better

JY

416%

358%

Worse

Results of the out-of-sample analysis are mixed. The ES is the same, the GC strategy is better with the “peel off” method, and the JY strategy is better with the baseline (always trading two contracts).

### Major Problem With This Analysis

While this analysis was inconclusive – sometimes the “peel off” method was better, sometimes it was worse, it should be noted that implementing a major strategy change like this, in general, should not be done once the strategy is complete. In this study, I took existing strategies, developed, and optimized without the “peel off” method in mind – and then added on the “peel off” approach. Clearly, this is not the best way to do it. I did it for this study because I primarily wanted to show the “peel off” approach on actual live trading strategies.

The correct way to build a strategy with “peel off” is to include the rule at the start of development. That way, the whole strategy is optimized with “peel off” in mind. If it improves the performance, then it will be in the strategy from the beginning, rather than bolted on at the end. This is a much more preferable way of developing – since adding on rules, etc. after development can easily lead to curve fitting (the old problem of continuously trying to improve backtest results).

### Conclusion

This article has demonstrated how to analyze the performance of the “peel off” method of trading, where one contract is exited at a smaller profit, and a second contract is allowed to run. Not surprisingly, different strategies yield different conclusions about “peel off” trading, although some of that may be due to the analysis method itself (evaluating the method on an existing strategy).

It is recommended that the reader, if he possibly wants to include the “peel off” method, incorporate the approach at the very beginning of strategy development and analysis, and utilize a robust walk-forward test process to analyze the strategy.

— Kevin J. Davey of KJ Trading Systems

Kevin Davey

Kevin Davey is a professional trader and a top performing systems developer. Kevin is the author of “Building Algorithmic Trading Systems: A Trader's Journey From Data Mining to Monte Carlo Simulation to Live Trading” (Wiley Trading, 2014.) . He generated triple digit annual returns 148 percent, 107 percent, and 112 percent in three consecutive World Cup of Futures Trading Championships® using algorithmic trading systems.

His web site, www.kjtradingsystems.com, provides trading mentoring, trading signals, and free trading videos and articles. He writes extensively in industry publications such as Futures Magazine and Active Trader and was featured as a “Market Master” in the book The Universal Principles of Successful Trading by Brent Penfold (Wiley, 2010).
Active in social media, Kevin has over 15,000 Twitter followers. An aerospace engineer and MBA by background, he has been an independent trader for over 20 years. Kevin continues to trade full time and develop algorithmic trading strategies.

## Jim Simons – The Man Who Solved the Markets

• This was an interesting post, thanks! I’ve recently read a few books circa 2005 on day trading, and there was much discussion of this peeling off. I discounted it as a pre-backtest era system because it didn’t seem to make much sense. There is not any inherent likelihood that the second leg of the trade is going to be successful based on the first leg…unless there is a strong trend, in which case why give up part of your profits by selling early? Your suggestion that one should build a system sith peeling off in mind though is problematic, isn’t it? How does one compare it to other systems if there is more than the “peeling off” variable involved? I’m sure one could build a profitable system using peeling, but is it better or worse? I think you may have answered the question in this post.

• Kevin says:

Hi Matt – Thanks for your comments. You can test peel off, but as you state – you need to make sure you compare it properly to non-peel off systems. So, if I designed a peel off system with 2 contracts, I’d need to compare it to a system trading 2 contracts, not a system trading 1 contract, since the capital requirements for the systems would be different.

In the end, I think the key is to do proper test and evaluation, and wherever that leads you is good. I think too many people assume that peel off is better, when that may not be the case at all.

• Was the point of the article an analysis of the peel off strategy or an analysis of the development method? Or is it an analysis of the relative market types that the strategy is deployed to?

I have a similar strategy that uses a double Bollinger, where after a sharp break in price above/below the outer Bollinger band, a double-sized position is taken, then as price reverts back to the mean, 1/2 the position is removed when crossing the inner Bollinger, the remainder is removed when price crosses the moving average.

The success of this strategy seems to be largely dependent on the asset and state of the market being traded. Trending markets with frequent breakouts work best, range bound markets perform poorly. FANG seem to perform well, while less trendy and volatile stocks perform poorly. And for some reason NASDAQ futures which can be associated with, or act as proxy to FANG stocks does not perform well.

• CY says:

The best part of reading your articles is like finding gems in a coal mine. Even harder to get contents from traders like andrea, tim rea and michael cook.
The only few people I found other than them are Derek Wong (chat with traders) and Perry Kaufman (so far).

Just a question:
Do you have any good contents of recommendations that has similar process thinking ?
Thanks
from CY
“Diversification is the only free lunch in the world”

• Kevin says:

Thanks for the comment Ernie. You bring up some good points. I think my results/conclusions are a result of how I tested, and what I tested, as you mention. I like hearing your experience where this type of contract sizing works. Honestly, I never looked at being selective with the approach, as far as certain markets and certain regimes. Looks like you have given everyone some more (hopefully fruitful) research work to do – thanks!!

• Kevin says:

Thanks for the comment. I have never spoken with Perry, but he has good stuff. I have talked with all the others, and in fact Tim, Andrea, Michael and I talk at least monthly via Skype. I find talking to traders really opens up my thinking (trading is typically a lonely business). I also like Art Collins, and of course Bob Pardo’s book. You should check out both of them.

• dcurrents says:

I use the peel-off approach for some models. In my case, it is more for psychological reasons then which produces best returns. My models are quantitative, but it is not automated trading. So psychological comfort is overriding factor.

• Juan I Marquez says:

Another awesome article by Kevin. Thanks!