Through several of my blog posts about the Metatrader platform and my recent posts about the Asirikuy Strategy Tester I have talked about some of the usual assumptions done in Forex trading simulations and how our new tester implementation seeks to eliminate all of them through the use of adequate strategy evaluation techniques which address all of these problems. On today’s post I want to talk about one of the most common assumptions made (and one of the least explored by people online) which relates to the ignoring of accurate historical swap rate calculations within trading systems. Through the following paragraphs I will share with you some of my findings about this and why swap rate calculations are fundamental and actually more important for trading systems that open positions for smaller periods of time.

What is a Forex swap rate ? In practical terms a swap rate is a debit or credit applied to your trading positions which is a function of the different interest rates to which the different currencies playing within the pair are subjected to. Since when you enter a Forex transaction you hold different currencies against each other (in pairs) the relative difference between the interest rate of the part of the pair you “bought” and the part you “sold” generates a net positive or negative charge merely through the holding of the position. The swap rate is larger when the central bank interest rate spread between two banks is larger while it is smaller when the gap is shorter. Another very interesting thing to notice is that when a positive swap rate exists for a position the opposite position has a larger and negative swap so in the longer term (if no consideration of swaps is taking into account) the swap will most likely be a negative contributing factor for a strategy. The only way in which swap can be a positive contributing factor is when it is explicitly taken into account into trading (as in carry trade strategies).

–

–

Some other interesting aspects about swap involve the way in which it is charged to your account. Most brokers charge or credit your trades for swap once each day at their daily close, regardless of the amount of time your positions has been opened so if you have a position opened at your broker’s daily close you will get swap debit/credit irrespective of your holding time (even if you opened two minutes before). Another important thing is that since swap is an interest derived component of Forex it is charged “everyday” so in order to compensate for weekends the credit/debits received from swap of Wednesday are triple of the regular amounts.

When we take into account the above it starts to become obvious that we’re neglecting an important part of trading if we do not take swap into the equation when carrying out Forex simulations. Generally people will consider that swap rates are only a small influence in trading and therefore will assume that simply considering the current swap level (as in MT4 simulations) is a reliable way to get past the issue. However a closer look at the problem reveals that this solution isn’t either adequate nor realistic as swap rates have varied hugely within the past 20 years and the effect of these variations is actually very important for the evaluation of most strategies.

The belief that only “carry trade” type strategies or long term trend following systems are affected by swap is also a myth as in fact the systems which are most dramatically affected by the influence of swap in the longer term are short term trading systems (especially those with low profit or loss targets). Imagine that a system trading the 5 minute time frame takes 100 trades out of 700 over a 10 year period during Wednesday before the day closing mark and exits them just after that (and each swap rate charge is just 1/3rd the average profit). If the system takes a long term 3 swap rate multiplied hit – because swap is always long term negative if not explicitly included inside a strategy – across such a large number of its positions it will – without a doubt – end up with negative results. Now this is quite an extreme case (to make my point) but in the end all strategies are considerably affected by swap and ignoring it is simply NOT the solution.

Perhaps the most important problem related with swap is the fact that it doesn’t have a “constant” behavior so it is quite easy to exploit back-testing problems based on the absence of accurate historical swaps through the use of short time frame trading systems. For example you’ll notice that on many high-swap yielding pairs there are predictable short term movements in Wednesday closes because people attempt to get in or out to avoid or take advantage of this trading moment (the triple swap charge). If you create a short term strategy to create a system around this time you’ll be able to achieve amazing profits but these profits completely evaporate when you consider the long term swap charges you must take to actually trade the strategy. It is also vital to consider that you need to consider accurate historical swap rates to adequately evaluate this as even taking into account “actual swaps” as constant will NOT be an even approximate solution.

–

–

Let me exemplify a little bit how much swap rates have varied through time with an example derived from our Asirikuy Strategy Tester (which will be released soon to Asirikuy) evaluating a trading system on the USD/JPY through the 2001-2011 period using accurate historical swap rates or ignoring their existence (this system is a daily trading system just used to exemplify the point). The pink lines show the relative magnitude of swap charges (doesn’t have anything to do with the Y axis scale but just plotted so that you can see their variation) and the blue lines represent the balance curves. As you can see through the whole trading period swap charges have varied a LOT and the current swap charges are just an extremely miserable amount of what they have been historically. When you see a graph like this it becomes obvious that if you assumed the current swap was the rate for the whole period you wouldn’t get accurate results.

Overall we can see how swap affected the strategy through time, primarily how it affects the ending balance of the system. When taking into account swap rates the ending absolute profit was 1.22% while without the swap rates it was 4.37%. We can see that while swap rates appear to have only a small influence in the short term their long term influence is quite significant since you can see that they inevitable diminish the edge of the trading system. Since swap rates are proportional to the lot sizes traded they will generally have a proportional influence with account size and lot size traded (just like the spread) and therefore they will cause a dent in the average compounded yearly profit to maximum draw down ratios of trading strategies.

After doing these experiments (carried out for the purpose of implementing this feature on the Asirikuy Strategy Tester) I can say with confidence that swap rates are an important aspect of trading system evaluation and that every trading system which ignores swaps needs to be re-evaluated in order to see what the effect of including them actually is. Definitely the new Asirikuy strategy tester will allow us to evaluate these aspects in Asirikuy and going forward we will definitely retest our systems to see what the effect and magnitude of our current assumptions about swaps actually are. Right now the only thing we know is that the ignoring of swaps causes an overall over-estimation of profitability and that including swaps may have an effect similar to that of spreads. However since swap rate charging is not uniform (triple swap on Wednesdays) some strategies may be more heavily burdened than others.

Of course if you would like to learn more about the evaluation of trading systems and how you too can tackle trading with knowledge and understanding please consider joining Asirikuy.com, a website filled with educational videos, trading systems, development and a sound, honest and transparent approach towards automated trading in general . I hope you enjoyed this article ! :o)

Looks like the swap rates follow very long term cycles. Thus, a conservative approach would be to develop/backtest systems with respect the worst swaps in the testing period.

It will be great to see your new tool running in linux. Personally, I’m unable to do anything serious without a good shell.

Best.

Hi McDuck,

Thank you for your post :o) Well the problem is that the “worst swap” is difficult to define since for different strategies the “worst swap” situation might be different. For example if you take the largest swap rate difference (the highest losing different between bid and ask swap) there might be some strategies which trade heavily on one side of the market during a period where the absolute negative swap against these positions is larger but the overall swap doesn’t look “bad”. Therefore the only way to evaluate swaps is to cover them historically through the use of accurate swap rates through the whole testing period. I hope this helps :o)

Best Regards,

Daniel

Hi Daniel,

How would you compare profitability of a system running in a regular account, vs the same system running in a commission (no swaps) account? The commissions are also charged during night rollover, however they are fixed per lot size being held. Maybe the strategy tester should allow to specify the commision value as an alternative to input swap rates.

Thanks!