Topic: Strategy Builder - Sharpe Ratio

This request may be inappropriate... but here goes....

Because the Sharpe Ratio can be most helpful in determining the viability of a strategy...... is it possible to have it as a filter for processing a strategy...... ie, the operator could set a standard and the strategy would have to make that standard as it goes along..... there could be a choice as to which number the strategy was to meet....

The software already computes this amount, I wonder if it could be incorporated in the building of a strategy..

Thanks


daveM

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

Re: Strategy Builder - Sharpe Ratio

It sounds reasonably.

Re: Strategy Builder - Sharpe Ratio

Hello Blaiserboy and Popov
Yes it is a good idea , and lets add the "conscutive losing trades as a filter" so it can used with max drawdown to select the safest strategies

Regards

Re: Strategy Builder - Sharpe Ratio

Somehow, without overloading the generator process, there might be a way to ensure that we have results that will be relatively successful.

Due to the nature of the generator algorithm, this software offers tremendous possibilities, especially if a person can allow the computation period to be lengthy in order to find good solutions.

Because the developer is willing to provide  a software which will incorporate unique features, we all have to be aware of things that can be added to produce the strategy which will trade effectively.

I have seen that there is not a lot of interest by the forum members in Sharpe Ratio, yet it would be a primary evaluation criteria.

It is essential, I feel that we provide ideas to the developer.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

Re: Strategy Builder - Sharpe Ratio

Trading System Metrics  article by Dave Jenyns



1. Win-to-loss ratio When assessing the performance of a trading system, one of the first statistics that gives you a good indication of tradability is the win-to-loss ratio. Quite simply, this is the ratio of the average winning trades taken against the average losing trades taken. If this ratio indicates you are, on average, winning more than you are losing, you are on the right track.

But don't get caught up in this statistic on its own, because it doesn't tell the whole story. It doesn't consider the size of your winning trades versus the size of your losing trades. Remember the Turtles? Their win-to-loss ratio was 40:60, but they were still hugely profitable.

2. Average wins and losses In addition to the win-to-loss ratio, you will want to make sure that the average value of your winning trades is greater than the average value of your losing ones. Say your back testing consisted of 200 trades. If 150 are losing trades and only 50 are winning trades, obviously your win-to-loss ratio is 25:75. But that on its own isn't enough to determine if a system is good or bad.

Understand that, if the average of your wins were, for example, $2000 and the average of your losses were $500, you are still coming out on top ((50x2000)-(150x500)=$25,000).

3. Expectancy A trading system's expectancy is perhaps one of the most powerful statistics you can have because it is a way of quantifying the performance of a system that is independent of the size of the trading float.

In short, it produces the expected dollar return for each dollar risked by the trading system. This is different to the reward-to-risk ratio and average wins to losses that we described above, in that it defines a return in dollar terms for every dollar that you risk. If your system has an expectancy of +0.75, on average, you would expect to make 0.75 times the amount you risked in the trade. If you risk $1, then you would expect to make, on average, $0.75 for every trade you take.

As a guide, if you can achieve expectancy of $0.60, you're heading in the right direction.

4. Maximum consecutive losses Look back through your testing results to see, statistically, how many losses in a row your system sustained while still being profitable. This is important to know upfront, since this statistic will give you confidence during those low times when it feels like you should throw in the towel.

For example, imagine you have been hit with five or six losses in a row. Without knowing your maximum consecutive losses, you might think your system isn't working. This is where most naive traders go wrong. The truth be known, based on the historical data, your system may have actually sustained 10 losses and still been profitable.

5. Maximum drawdown The maximum drawdown is the worst period of 'peak to valley' performance of your system, regardless of whether or not the drawdown consisted of consecutive months of negative performance.

This statistic is automatically calculated, so it's just a matter of asking yourself: am I comfortable with that size loss? If not, you will need to do more system tweaking to get it to a level that you can live with.

Again, it all comes back to the risk-to-reward ratio. Typically the more risk you take, the greater the reward. I have traded a system in the past that returned 140% p.a. Now that sounds great, but that particular system had a maximum drawdown of 80%. Could you trade a system where it's likely you'd lose 80% of all your capital at least once while trading it? Could you stomach that?

It's important you trade a system you're comfortable with.

6. Number of trades Then there's the number of trades a system gives over the course of a year. I find this an invaluable, yet rarely talked about, statistic.

Your trading system should not give too many or too few trades. The number of trades that a trading system gives should be approximately the same as that which can realistically be taken.

The two sides of the coin are equally dangerous. If a system gives too many trades, you will be forced to choose between signals, therefore adding ambiguity to the system. With ambiguity comes human discretion and this often has a detrimental effect on the performance of the trading system.

On the other hand, if a system gives too few trades, your trading capital will not be fully utilised and you may not be taking full advantage of the available trading opportunities.

So how do you calculate the optimal number of trades for a trading system?

This is done with the calculation called 'opportunity'. Opportunity helps determine your optimal opportunity for a trading system.

7. Profitability Profitability is simply the return on investment over a yearly term.

Let's be blunt. We're all in this to make money. At the end of the day, profitability really is the most important metric by which to measure your trade system. But, while it's important, it needs to be balanced with the other six measures I've just discussed.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

Re: Strategy Builder - Sharpe Ratio

I have seen that there is not a lot of interest by the forum members in Sharpe Ratio, yet it would be a primary evaluation criteria.

We are interested in developing an algorithm for evaluating the generated and manually created strategies. This algorithm will include several criteria and will execute some tests on the strategies. Our idea is this to be a separated module to the program. Actually we are only in the beginning and have very basic idea about it.

If you have can share some experience with such strategy evaluation method, we'll be happy to include it in the future releases of the backtesting program. Probably we'll start an separate topic about such module, but we'll take into account all feedback we are receiving. However, it will be nice if you provide some theoretical or statistical information about different criteria.

It looks like Sharpe Ratio, together with some of the other statistical parameters included in the program we'll be used, but we have to decide how exactly they will contribute to the evaluation result.

Any further feedback is highly welcome.

Re: Strategy Builder - Sharpe Ratio

There are many measurements of a system, and combinations...... I think you have the important ones already incorporated into the strategy overview.

The difficulty will be, I think, to present the information in a manner that will satisfy the system developer as he/she is generating a system... during and also after.

One person will want data as he is generating. while another will want the information later, and another will want to eliminate generating a system that does not meet the metrics that he has defined prior to starting the generator.

Hopefully many people will contribute some ideas as to what they feel to be important and how they would like to see the information displayed.

Myself, I would like to be able to predefine some metrics and let the generator work to develop things that meet those standards.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

Re: Strategy Builder - Sharpe Ratio

some ideas for metrics.......... perhaps this will stimulate ideas from others


http://www.ourhonestreviews.com/stealpips/7-metrics-to-analyze-your-trading-system

Here is another approach which may or may not be useful to people

http://www.automated-trading-system.com/e-ratio-trading-edge/

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

Re: Strategy Builder - Sharpe Ratio

I would be really pleased if I could set criteria for and also observe during the process the following

Profit Factor
Sharpe Ratio
Win Loss Ratio
Expectancy
Profit per Day


Being able to observe those during the generation phase would allow the work to be stopped or continued or modified.

A tremendous time saver.!!

Thanks for considering this

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

Re: Strategy Builder - Sharpe Ratio

As you wrote above, there are two approaches:
  - a super fast generator with some basic limitations (like the current one) that generates strategies and stores them in a buffer and after that an evaluation module to filter them according to more sophisticated criteria.
- all requirements to be predefined, so the generator to generate - evaluate each strategy and to reports only these that pass the test.

My plans are to release first final versions of FSB / FST with the current features and after that to start work on that evaluation module. It will greatly reduce the number of strategies that have to go to the next stage - trading on a demo account.