Strategy Testing

Understanding Back testing vs Forward tests

Understanding this is quite important in our opinion, as there’s so much of misconception and the need to log forward tests via MyFxBook, FxBlue or other independent bodies just to prove the successful workings of a strategy.

In our strong opinion, there is a vast difference between Manual and Automated strategies.

For Manual strategies, where the orders are being strategically placed by the trader, it would be quite necessary to see forward test results in order to gauge the ability of the trader’s capabilities. A growing trade record with low drawdowns would show that the trader is in control of the trades and has a sound strategy.

Whereas for Automated trading, a forward test record of a few months is no way a representation of its soundness over the next following month or even the next week. Shattering events like the SNB announcement in Jan 2015 or Brexit voting in June 2016 and other fat finger flash crash incidences can result in account blow ups if the automated systems are not protected against these unpredictable occurrences. Some of these events which are well known ahead of time could be prevented by turning off the trading advisers and staying on the side lines till after the event, just to switch them back again to commence trading. However, fat finger flash crashes which are unpredictable can cause a wreck to the account.

With this understanding, we prefer to crash test our strategies over all such events that have occurred over the recent years to stress test the performance of our Expert Adviser strategies.

This elevates our confidence in trading Live with those stress tested strategies for the longer term with a peace of mind. Hence, we do away with the need to forward test these for months just to artificially boost confidence that such strategies are staged to perform under automation.

Back testing is another area which is not deeply understood by many traders who test their strategies for endurance and stress.

Most brokers do not have extended M1 data spanning years in MT4s history center. Hence, traders tend to use larger time frame data such as M30, H1 and above to test their automated strategies using the Tick by Tick method.

Firstly, this really becomes very time consuming to test strategies for over 3 to 5 years for durability checks. Also, the lack of understanding about what the Tick by Tick back test setting does in the MT4 Strategy Tester.

Running a Tick by Tick setting on a H1 back test time frame only breakdown the H1 data into tiny bits of 100 to 200 or so by MT4s own pattern to depict a tick by tick move, whereas in actual fact, there is no tick by tick data in MT4. The lowest time frame is M1 and not a tick by tick record.

But this Tick by Tick setting helps to break down a bigger time frame data into bits, just to depict some tick by tick movement, so that strategies can be back tested with those minute movements for better accuracy. It is in no way representative of the actual tick run. For instance, the price could have spiked 100 pips with a spread widening about 20 pips and then collapsing back to where it almost. This is not captured in the larger time frame Historical data.

So, there is no way that a Tick by Tick test could justify that a tested strategy is robust and could pass that stress conditions where prices experience wild moves.

The only closest and best way to back test is to use the lowest time frame, which is the M1 data.

Now, here is the set back. There is no way to obtain an M1 data for years back from the broker.

The only option is to download this externally from providers who provide such data. Most popularly used are Dukascopy’s Historical data.

This is what we use ourselves for our stringent strategy back tests.

We run our tests using the years of obtained M1 data under Control Points instead of Tick by Tick.

Let’s explain this.

When using the Control points setting, what MT4 does is to break down the data to about 10 to 20 moves.

Using it on an M1 data is more than sufficient for accuracy with this caveat….as long as strategies are not sensitive to lower than 10 pips moves. Meaning, if one is testing scalping strategies where take profits or stop losses are set lower than 10 or 15 pips, then a Tick by Tick setting with M1 data would be the way to go. Though this would take quite a lot of test time, there is no other way, as the strategy is set for such TP and SL moves.

In our case, where we do not run scalp like strategies, we are absolutely fine testing our strategies under Control Points.

To justify this ourselves, we have taken tested strategies under Control Points and have run them with the Tick by Tick setting. The results were remarkable. Just an insignificant difference in profits.

It just proved to us that for our kind of strategies, we do not require a Tick by Tick run with the M1 data, but just a quicker run under Control Points would be more than suffice to show the needed accuracy and stress test durability.

Armed with such information, we test out strategies for a few years in a stretch and all of our EAs (Expert Advisers) have in built features to record back test performances which record the Highest margin used, the Highest Drawdown and the number of times our preset limits have gotten triggered.

So, we have deep insight of our own strategy settings and are able to confidently put them into Live trading action.

Above this, we run our strategies under very conservative settings. This allows us to trade beyond the rattle of daily news events and shattering event like Fat Finger triggers, Brexit stunts and the SNB melt down of 2015 which had caused havoc and wrecked many accounts plus thrown many brokerages into liquidation.

With good brokers where we do not experience platform freezes and have ECN feeds, we are able to successfully trade our conservative settings with peace of mind.

For our Signals Provision, we cannot be providing these for small account sizes, as we would then need to open several small accounts to run these Live, which would serve no purpose for us except to please the eyes for some just to gain the confidence as Followers. This would naturally thin and be a waste of our resources. Certainly not be productive for us. No offense here but merely stating the reality in our case, as we would rather consolidate our resources like we do now into larger accounts to trade for our income generation.

Hence, we have chosen to run strategies with respectably sized accounts of $10K and above, also for Followers who wish to follow and copy our trades with their similar sized accounts.

For traders with smaller account sizes and who wish to trade with our automated EAs, we choose to provide these EA set files, which they can just attach to their MT4 charts and have them run. These are supported by our stringent backtests and evaluations. More understanding of this here……

The back test results of such strategies will be published along the downloadable set files.

If we chose to put forth any manual strategies, we will have our forward tests up for scrutiny, as this is the only way to understand the performance of manual strategies, since manual strategies cannot be back tested.

error: Alert: Content selection is disabled!!