Analysing trades: Importance of using enough data

Author Topic: Analysing trades: Importance of using enough data  (Read 143253 times)

Offline donnaforex

  • Administrator
  • Hero Member
  • ******
  • Posts: 4218
Analysing trades: Importance of using enough data
« on: August 30, 2017, 02:02:59 PM »
Often times, we end up starting an EA at a 'bad' time and experience an immediate drawdown or a run of disappointing trades, maybe even 10 losing trades in a row. An inexperienced trader might conclude that this EA doesn't work or they have set it up wrong and simply move on to the next one. This scenario keeps repeating and suddenly you find yourself thinking you are cursed and this trading game just isn't for you.

The problem we have here is that the data you are looking at is not representative.

Statistically, the more trades we have to look at the more representative that data will be of the overall performance of the EA (assuming future similar market conditions, which might or might not happen).

As a simple example to illustrate my point, you might have an EA that wins 50% of the time (and on those wins, it might gain 30 pips, versus a loss of 20 pips when it loses, so a net positive EA).

If this EA makes 10,000 trades, then we might expect 5,000 of them to be winners, and 5,000 to be losers (or very close to that).

If we zoom in on this EA's trades and pick blocks of 20 trades, we might find a very different picture. Certain trading conditions or news events might occur where the EA makes 20 losing trades in a row. Alternatively, it might make 20 wins in a row. So, picking any one 20 trade 'block' is not going to be much use for our purposes of finding out if the EA is worth trading or not.

This is exactly what is happening to you when you load up your EA for the first time and hit an immediate run of losses. You quite likely happened up a losing streak and streaks are very common, even if you assume that the markets are completely random (which i don't believe they are, but that's a long discussion for another time).

Checklist for success:

1. Make sure you know the character of the EA you are trading. Does it make lots of small gains and then have a few big losses, or does it have stoplosses of a similar size to it's take profit points? How does the equity curve look? (research other people's stats). Knowing what to expect and what is normal for an EA will help give you confidence to trade through the low points.

2. Make sure your EA is set up correctly. If you purchase an EA then make use of their support or post here in the forum for help - it's always worth double checking if you are unsure about trades your EA is making. Most importantly, ensure you fully understand the risk settings and that you have selected a level that you can handle.

3. Look at representative amounts of data. Sometimes we don't have the luxury of having huge volumes of trades to look at and have to make do with what we can find, but bear in mind all my points above and note that the more trades an EA has, the more likely we are looking at representative data and not simply a series of flukes or streaks.

4. Don't panic if you start off with a losing trade or two and develop a longer term mindset. If you research an EA before you buy it then you should be able to proceed with enough confidence to trade it long term and pull through any short term dips. Whilst refund options are always nice to have, 30 or 60 days just doesn't give you enough time in most cases to come to any conclusions, you are going to need to look further afield than just the data you can collect yourself in one month.

5. Know when to quit. Obtaining a large dataset of trades doesn't mean your EA will automatically churn out stable wins. If your EA has made 300+ trades and isn't
going anywhere, this is starting to look like a loser.  The more trades you have without any positive direction showing, the stronger signal this is giving you to quit.

I have attached here an example to look at from my own stats. It shows an EA which has occasional large losses punctuated by periods of steady gains. I know a number of people jumped into this EA right where you can see my drawdown and subsequently bailed out. For me, it wasn't so bad to watch because of the prior gains, but if i had started the EA at this point it would have been hard to watch. It has now built up 350 trades which is a nice set of data but still has a way to go to prove itself fully.
« Last Edit: September 02, 2017, 10:41:52 AM by donnaforex »
Follow DonnaForex on Facebook and Twitter.
If you need to contact me, email or Skype 'donnaforex' or PM via the forum.


browse forum