Creating a Powerful Portfolio Blend from Scratch – Part 3
Fine Tuning the Portfolio and Adding Realism to the Back-test Analytics
In Part 2 of our 3 Part article titled ‘Creating a Powerful Blend from Scratch”, we looked at the blending phase of portfolio creation where through the selection of equity curves that offered correlation benefits, we progressively built our portfolio from ground up to stand the test of time in a wide range of different market conditions.
We were left with a back-test result that just seemed too good to be true.
Test Period – 1st January 2010 to 16th August 2017
Trade sample size: 1,176
Win Ratio: 43.8%
Maximum Drawdown: 11.6%
MAR ratio: 5.28
Well…you are quite right. It is too good to be true.
Before we can pluck the courage up to launch into our live trading following our exciting back-test results, now comes the time to remove any bullish euphoria and focus on the reality of live trading. We do this to dampen our expectations and ensure our risk weightings are correctly applied for live conditions. This is where the brutal reality of the frictional costs of trading need to take center stage….as we inevitably incur additional costs to that anticipated from our back-tests attributed to the realities of live trading.
This process makes us exceedingly aware of the very thin slice of alpha that we actually possess, and how the frictional costs of trading inevitably add up to be one of the biggest banes of our existence. What previously was a glorious compounding equity curve gets hammered to death by the thieving middle fingers of the brokers who place our trades.
Ok….so fortunately spread from Dukascopy has already been incorporated into the results, but if you use a different broker then you need to ensure that the true costs of spread and commission are reflected in your results. For the purposes of this backtest and going forward, we will be using Dukascopy so we have already taken this into account.
In a nutshell the cost of spread for Dukascopy across our universe is approximately $40 per round trip risking $500 on a $200K account or 0.25% trade risk%….This therefore represents 8% of your trade risk %.
On a total of 1176 trades on a 0.25% trade risk we chew up approximately $47K in commissions which represents approximately 26% of our gross profit before commissions……that’s right…..read it and weep. That is the privilege of paying those kind generous brokers we love.
SWAP (Holding Costs)
….but it doesn’t end there….now we need to compound our misery further by adding the costs of SWAP to our returns to pay the holding costs associated with trading leveraged instruments. Your brokers will declare that sometimes you can make money on your SWAP dependent on inter-bank lending rates…..however that is the exception as opposed to the rule.
We have a guide to use to load up our return distributions with an estimated total SWAP cost for trades with an average duration of 12 days hold…….are you ready for it?…….be prepared to take an additional $20 per round trip of your equity curve or half of your spread costs again……and you are lucky because if your average hold extends out to the months….then SWAPs of up to 3 x your total commission costs are fairly standard.
….but wait…..you want more don’t you? Ok here goes. What about that dreaded word we love to hate called slippage. For breakout traders, this is an inevitable issue you just have to accept. Unfortunately we enter trades at the very worst times under high momentum where brokers love to launch the excuse of trade slippage to state why your pending orders were not triggered at the right price…..so how much should we assume?
Well given that sometimes you don’t slip but at other times your slippage can be quite extreme the best we can do is apply an average slippage per trade. For the purposes of this exercise and from past live experience I am applying a fairly realistic 2% slippage for every trade which eats into your total gross profit by about 4% for the privilege.
Other Frictional costs
For the purposes of this exercise we are ignoring trader error which in the world of discretionary trading can be materially significant…however let’s just say that the assumptions so far could positively or negatively materially differ from what’s stated…however at least for this exercise we have significantly diluted our expectations by at least incorporating more realistic assumptions into the results.
So what are the material impacts to our equity curve?…………for those who have anxiety, it is probably best to look away now.
Take some time to familiarize yourselves with the performance metrics above as it demonstrates how the majority or the gross profits generated by your system are distributed to your brokers and other intermediaries and how you only have a very slender thin slice of alpha to actually play with.
The impact is significant on compounded accounts. Just to drum it in, what previously was a 61.4% CAGR with a Max Drawdown of 11.63% is now a CAGR of 43.3% with a max Drawdown of 15.91% and we still have some work to do in reducing our expectations further by scaling our portfolio to our Max Drawdown tolerance of 10%.
Defining your Maximum Drawdown Tolerance
Before we start assessing our risk of ruin, we first need to define our Maximum Drawdown tolerance. Now for retail traders risking small amounts of capital that they can afford to lose, not much attention is usually placed on maximum drawdown tolerances….however we are in this game for the long haul and want to trade size….and as a result, our drawdown tolerance needs to be carefully defined as it represents the maximum capital loss we are prepared to digest before we call it a day with our system.
We need to recognise that our max drawdown is an inevitable consequence of a biased string of unfavourable losses and with the Law of Large numbers our worst drawdown is inevitably a future event waiting to happen.
My personal drawdown tolerance is around the 20% level of available trade capital, so I want to ensure that at least over the history of the back-test that encompasses very unfavourable market conditions for trend following in general, that I scale my position size weighting to a level that avoids this painful statistic. Given that our worst drawdown is inevitably somewhere in the future I therefore halve my drawdown expectations with regards to the risk weighting of our portfolio. So for the period of our back-tests, we are looking for a max drawdown of about 10%.
Here is the result after scaling our position sizing to a 1% trade risk per trade.
So now we are at realistic levels with an appropriately scaled portfolio weighting in accordance with my risk appetite….where to now? Well now we do a cross check and test that our risk of ruin is 0%.
Risk of ruin
In understanding your risk of ruin you need the following information from your backtest.
risk of ruin = ((1 – Edge)/(1 + Edge)) ^ Capital_Units
Now don’t worry too much about this formula as there are some good risk of ruin calculators on the web. Just look them up.
Here are the results of the risk of ruin calculation applied to our back-tests and some scenario tests.
Looking good so far. My max drawdown tolerance of 20% is unlikely to be exceeded….but what happens with a max drawdown tolerance of 10.7% that we have already experienced in our back-tests?
Well on this particular generation it demonstrates that there is a 78% chance of risk of ruin. Clearly it has been a wise move to dilute the risk weightings well below my maximum drawdown tolerance.
Now there are no guarantees in trading…however at least within expectations of a defined confidence interval, my risk of ruin is unlikely.
Well now that all the hard yards are done and we have conservatively pitched the return distributions of our back-test results to reflect realism and serve as a performance baseline, it is now time to study the statistics in more detail to answer more questions we may have down the track in the thick of live trading.
Preparing Your Psyche for Live Trading
Now that we have completed our back-testing and have scaled our portfolio in response to our desired risk appetite, it is time to prepare ourselves for the inevitable questions we will face when we are live trading and are under the pump. The back-testing process arms ourselves with critical benchmark statistics that we can refer to when the inevitable questions inside our head arise when enduring the tough times. There is nothing to say about the good times….apart from the general feeling of euphoria and being on top of the world.
So what will the bad times bring? These are the times you are tempted to second guess your process, revenge trade, go suicidal with your risk management or strategy switch….and without a road map to guide you, you will be listing like a rudderless ship in a stormy sea. The greatest strength of a back-test is not the assessment of the robustness of your strategy (but this does help), but rather in that it provides a reference tool for you to use to respond to those dancing queries that will raise in your head from time to time.
Have a look at the analytics above and let’s step through it so we have a firm understanding of the portfolio’s performance and some of the important metrics you need to understand for your live trading experience. I have converted $ references to R values for ease of applying this system to your desired capital allocation. R represents your trade risk. For example with a total capital allocation of $200K and a trade risk of 1% or $2,000, then R = $2,000. For a capital allocation of $10,000 where trade risk is 1% or $100, then R = $100.
On average this strategy (applied to 24 liquid instruments) on the M30 timeframe has experienced 1,176 trades between 4/1/2010 and 16/8/2017 (2,783 calendar days). The average trades taken per Calendar Day are 0.42 trades per day. In trading days, this represents approximately 1,935 days or 0.61 trades per trading day. Though this appears to be a slow trader, looks can be deceptive as some days are far more active than others. The maximum number of open trades taken on any given day over the trading period was 16 trades. Now remember that we always open two trades at once (a short term trade and a longer term trade) so this in reality can be halved to a maximum of 8 open events in any given day over the period. This is a rare event in that there were only 5 days over the entire period where more than 5 open events occurred over the period…..however sometimes you need to have your head about you. Always be prepared……..
The maximum number of trade exits on any given day was 7 trade exit events…..so once again….you need to be on your guard for those explosive periods when many of your instruments will pick up steam due to their correlated nature.
Profitable and Unprofitable Periods and Instruments
Don’t be deceived in thinking that all your instruments will be profitable or that you will be profitable every month. That is just not reality for any trend following or momentum based trading system.
We can already see in the statistics above that 5 of the 24 instruments were unprofitable over the entire period…..but it gets worse than that.
The summary statistics below are deceptive as it appears this strategy is a breeze with every year being profitable and an equity curve that appears attractively alluring.
To achieve that result, this is what we had to endure during our journey:
- Only 42% of your trades are going to be winners…so right from the get go, your default position is that you are going to lose on entry.
- The maximum number of consecutive losses so far over the 7.62 year period has been 14 losses in a row……..so when you are at your 9th loss in a row it is up to you to keep to the process and don’t try and second guess or over-ride your system. This will test anyone whether they have a back-test to refer to or not…..but your focus on the process is the only thing that is going to save you in this event.
- The vast majority of your trades are just to keep your head above water. Look at the histogram below. Of your total 1,176 trades over the period, 90% of them simply managed to allow you to break-even, 9.5% of them allowed you to exceed beta and two extreme outliers delivered icing on the cake. Let this simple fact of positive skew sink in. This is what you are dealing with if you want to win this game. Do you have the stomach to take it?
Here is a breakdown of what you need to face on a monthly basis. Look at that sea of red in the monthly results. Some periods of negative returns extend for 5 consecutive months in a row.
Okay, so after the appreciation of the anticipated pain that will be involved…we are now getting close to dipping our toes in the water. The concept of parallel back-testing when live trading is all about monitoring your processes and ensuring that what was produced in your less anxious testing environment is mirrored in your live trading activities. This is where you launch your trade journal to annotate how you are performing in relation to your testing environment. You want to ensure that the rhythm you maintained to produce your back-test results continue on in your live trading and under a discretionary system, a way to achieve this is through the continuance of your back-test (in live walk forward mode) in tandem with your live trades.
Under a parallel environment you will be better positioned to assess whether you poor performance in any given period can be attributed to market conditions as opposed to trading error (or a departure from your back-testing process. This is where you need to get hold of some good back-testing software that allows you to continuously update your tests with the addition of new future data.
So there we have it. This three part series has hopefully demonstrated that while the strategies deployed may be simple to capture very slim slices of alpha, the process of portfolio creation is actually quite extensive and exhausting……….but well worth it.
Like engineers, we have constructed a robust portfolio from ground up to survive the turbulent markets and harvest short term momentum wherever it may be lurking. As you step through this process you will become very aware how I rarely mentioned the word “profit” in our entire 3 part series. The entire emphasis of this story was on “risk management”.
Provided your portfolio is robust and can protect capital during the tough times, profits will simply become a natural outcome of this process.
I hope you enjoyed this 3 part series and perhaps it may inspire you to have a crack at building your very own robust portfolio. As a long time trader, I have dabbled with many different systems and techniques. It was not until I invested my efforts away from a focus on single systems towards the concept of systematic diversification that my fortunes changed and allowed me to focus on simpler systems under diversification that could be applied to any liquid instrument.
Through this approach to portfolio creation you avoid the tendency to select curve fit solutions and use the power of correlation and position sizing to manage your risk weighted returns.
Trade well and prosper