Back Testing & Evaluation

Back Testing is the corner stone of your business. Without it, you have no idea of a system’s ability to perform. The more back testing you do the more confident you can be in the system’s ability to reproduce similar results in real time. Back testing only a handful of trades will not do this. Although this is just a guide, you are far better off back testing at least over a two year period, and with at least 50 trades.

If you have developed a discretionary trading system that especially involves multiple time frames, back testing is not going to be as easy at is for those who have a mechanical system. If this is you, then we suggest paper trading but disregarding any 'maximum trades open' rules while you are paper trading. For example, if your capital wouldn't allow you to have any more than 10 trades open at once, while paper trading you would ignore this rule completely because it allows you to gather much more data in a shorter amount of time. In other words, paper trade as many trades as humanly possible to get as much data as you can; the more the better.

After going through the previous modules you will want to have an idea of the following:

• Objectives (e.g. make a 50% return p.a.)
• Trading style (e.g. growth trader)
• Trading rules (i.e. the 4 T’s)
• Starting capital (e.g. $30,000)
• Maximum drawdown tolerance (e.g. 25%)
• Risk per trade tolerance (e.g. 2%)
• Markets to trade (e.g. equities, CFD’s, options)
• Costs of trading these markets (e.g. $40 per trade and 5% interest on CFD positions calculated daily)

Once you have the above or at least a good idea of the above, the next thing is to start back testing your trading rules. When you back test your trading rules you want to record the following:

• Entry date
• Entry price
• Stop loss
• Trade size
• Margin requirement
• Exit date
• Exit price

Once you have back tested enough times over long enough periods of historical data, you can then determine the following metrics:

• Average net profit per trade
• Win to loss ratio
• Average number of days in a trade
• Largest use of capital (margin) per trade
• Biggest drawdown
  
  

Tony has the following:

• Starting capital of $50,000
• Objective is to make $30,000 over a year or 60%
• His style is growth trading; he is not concerned with income just yet
• Maximum drawdown tolerance of 25%

After settling on his choice of trend, trigger, trade and termination rules, his choice of markets, costs and his risk per trade, he back tests his system over a 2 ½ year period which included 75 trades over a large cross section of the top 50 stocks.

His results were as follows:

• Average net profit per trade was $150
• Average number of days in a trade was 21
• The largest use of his capital per trade was $3000
• His biggest drawdown was $5000 which was an 8 trade losing streak
• His win to loss ratio was 31:44 which Tony didn’t mind

1. Tony first divides the goal, which is $30,000 by the average net profit per trade which is $150. This gives him the number 200 (30,000 / 150). What this number is telling him is that he will need to make 200 trades over the next 12 months to reach his goal. The reason is because Tony already knows his average net profit per trade.

2. Next he multiplies the number of trades needed which is 200 by the average length of each trade which in this case is 21 days, giving him a total of 4200. So if the average length of a trade is 21 days, and Tony takes 200 trades he will need a total of 4200 days of market exposure, but he only has 12 months. So we divide 4200 by 365 (# of days in a year) which gives him 11.5, rounded up to the next round number giving him 12. This tells him that in order to make 200 trades in only one year Tony will need to have the ability to have 12 trades open on any given day because in all likelihood he will. The only way to produce 200 trades in a year is to have multiple trades open at once.

3. Then Tony must multiply 12 by his largest position size (or use of capital) and in this case it’s $3,000, giving him $36,000. Because the starting capital is $50,000, he can safely have 12 trades open, in fact he can have even more, but 12 is all he needs so that is fine, plus $3000 is the largest use of his capital and not the average, so the figures are conservative.

4. Now comes the tricky part and that is the drawdown. Unless Tony is able to simulate the trading of 200 trades in a 12 month period, the drawdown he currently has is only based on a smallish amount of data, and so because of this it is wise to consider worst case scenarios. Tony’s largest drawdown is $5,000 over 8 losing trades, or an average of $625 per losing trade. Let’s assume that we have 12 losing trades in a row (I use 12 because this is the number we came up with in point 2 to determine number of open trades on any given day); this gives us a $7500 drawdown (5000 / 8 = 625 * 12 = 7500). This is just using the average loss per trade during that drawdown period; however you could determine the average loss per losing trade from your own performance results. This is well within this Tony’s limit, but what if he had 12 losing streaks in a row (based on trading 12 different markets or sectors), this would wipe him out. The other way to look at it is that Tony is only willing to risk 25% or $12,500, and based on the average I used of $625, he can tolerate a 20 trade losing streak (12,500 / 625). Tony will have to ensure he diversifies his market exposure so as not to rely too heavily on one or two sectors or asset classes to avoid such a losing streak.

It is always wise to consider worst case scenarios, because historical performance is no guarantee of future returns. As just mentioned, Tony has options available to him to reduce the possibility of this worst case scenario, i.e. diversifying his market exposure to multiple sectors or asset classes within his pool of chosen equities (i.e. top 50 etc).

Another option to help gauge a more accurate assessment of any possible worst case scenarios is to back test more. Instead of only 75 trades, try 200 trades and over a 5 year period.

Back testing may seem tedious and there are software packages available to help reduce the time involved. If you do decide to purchase software, it is still wise to do some manual back testing first. This will give you a feel for the system which you can not get from using software.

So it seems Tony can operate his system fine as long as he does not have any more than a 20 trade losing streak, as this will take him over his drawdown limit. This means Tony will need to know a bit about the various markets he plans on trading and how he is going to find 200 trading opportunities across a diverse market place (i.e. trading many sectors/ asset classes/ mixing longs with shorts and so on) in a 12 month period. Of course Tony can adjust his goals, he can adjust his drawdown limit, he can try and adjust his system to try and improve its overall performance both in average net profit per trade and a smaller drawdown.

Objective______________ / Average net profit /trade____________

 = _____________ (required # of opportunities)  X  Average length/trade_______

 = _____________ Total Market Exposure / 365 = _________ Multiple trades necessary

 X   Largest use of capital_________ = $____________ Is this smaller than your capital?

What you need to do

1. Know your objectives, resources, trading system ideas, and risk tolerance
2. Back test your trading system
3. Determine its performance metrics
4. Determine if the system will reach your objectives within your risk tolerance

If NO, go back to 1 and try adjusting some of the following:

• Increase your starting capital
• Reduce your objective
• Increase net profit per trade
• Decrease position size

Increase Starting Capital

This is a 'no brainer'. It's logical that the more capital you have the less work that capital needs to do to reach the same objectives as a smaller amount of capital, all else being equal.

Also, there is nothing wrong with using your trading account as a savings account. By that I mean, if you are a wise investor you would save habitually anyway, so why not top up your trading account regularly?

Reduce Your Objective

Before you start buzzing around trying to improve your system by tweaking indicators, etc, please do consider whether you are asking too much from your money, your experience level and you resources. You should look at some comparisons of the fund managers out there and then compare experience levels. Look at the long term results of the stock markets indexes themselves. George Soros is probably one of the richest 'traders' of the markets there is, yet when interviewed he shocked the interviewer when he told him that his objectives are between 30 and 40% per annum. He said "You'd be surprised how rich you can become making 40% a year compounded!"

I am not suggesting that you settle on 30 to 40% a year, heck, if you want to aim higher then do so, but just equip yourself with the necessary resources and more importantly experience to get there. This takes time. If you design a system that makes 100% a year then good for you; but for you if such a system is difficult to find, don't get discouraged, reduce your objectives now with the intention of increasing your objectives later on as your abilities and experience increases. Trading is a great business and it shouldn't be ruined because you set your goals far too high far too early.  

Increase Net Profit Per Trade

If you have settled on your starting capital and your objectives, then the next thing is to tweak the system. There are many ways to do this, and you shouldn't just look at the indicators.

For example, what if you found by using a different broker you were able to save a few dollars per trade. This may not seem like much but over many, many trades this could dramatically reduce overall costs which always increases average net profits per trade.

Another example, you may decide that entering your positions in the first hour of trading is costing you too much in spreads and poor fills, due to increased volatility. Try entering your trades after the first hour.

Another example again, if you trade CFD's, where the interest costs on long trades are quite high, you may need to look at other markets where the interest rates are lower; or increase your short exposure.

Of course you can always look at tweaking the indicators or trying different ones. What you may find is that if you settle on a certain trend and trigger set up, you'll usually know quite early on if it is going to provide trading opportunities or not, even from a short amount of back testing. So although trying different combinations and set ups will be necessary early on, as soon as you find a profitable combination (i.e. it provides a net profit over time), you should look at the other factors of costs to help improve its performance.

Decrease position sizes

You may have decided to use a very uniform position sizing model, i.e. 5%, 10%, 25% etc of your available capital per trade. If this is the case, then reducing your position size will allow you to make more trades. This wont actually help your average net profit per trade, in fact may even decrease it slightly due to increased trading costs, however; if it allows you to make more trades per period it may help you to reach your objectives.

If you don't use a uniform position sizing model and you find some positions are a lot bigger than others, you may look to reduce these to a more acceptable level or remove them altogether, i.e. use a filter that will not allow a trade if the position size or margin requirement is more than a certain % of your capital.

If YES Move on to Step 5 - Module 3 > Running The Business