When is Strategy a good one..?

[lovethetrade]

Strategies don't "behave;" markets behave.

Strategies are a static key trying to fit a dynamic market lock. Some fit better than others. Some fit better one day than the next. But they don't behave. The *market* behaves.

A *fixed* key, and a *dynamic* lock. Stay on your toes, always.
Understand that you are the tail, not the dog, and never forget who wags whom.

On ignore, troll.

You never know, one day you might actually post something of value rather than being an effective troll. Can't say I'm holding my breath though. You're simply an unfortunate beneficiary of other people's goodwill.

 

[traider]

You can never assume a trading strategy is solid by looking at just one number like profit factor.

With enough trades, metrics like sharpe ratio or profit factor is good. Problem is most strategies will never reach that stage.

 

[FTDK]

With enough trades, metrics like sharpe ratio or profit factor is good. Problem is most strategies will never reach that stage.

One number on its own is meaningless, you just confirmed that. You need other variables to make sure the numbers have reliable meaning, like number of trades.

Even then profit factor on its own doesn't say anything about draw down or other numbers that can give you an indication about past risk.

 

[Wippi1817]

Profit factor is good enough if it's high

I would not agree on that one - profit factor is not enough

 

[Wippi1817]

A strategy is a good one when it has positive expectancy. Period.
Do you *expect* to make money?

Your question poses "strategy"; your comments juxtapose trading/performance. They ain't the same thing, and you conflate them only with a confused comprehension.

At any event, having positive expectancy is only halfway there: *real*life* dictates that we monitor relative risk, as well. "Relative?" Neither strategy nor trading can be evaluated without having an arbitrary reference -- so, for good or ill, you have to have a "Compared to what?" metric.

So, two questions result: Do you make money? as the first hurdle, and then, At what risk? compared between them.

Assessing market performance outside of these two questions is a fool's errand.

Yes I agree - I must be more precise - I want to measure a strategy on a selftrading system (bot) - for the first step I want to compare it to buy and hold - the interesting part is that I make money also in long term (e.g. 2 years data) but is this enough? no I dont think so - as you wrote I think I have to monitor risk as well....

 

[Wippi1817]

Rather go by Closed profit equity curve's drawdown percentages and how long to new equity highs.

That means netprofit in % max drawdown and a growing ROR curve or?

 

[Wippi1817]

Attaches some values:

Strategy 1: (1 year data - 30min timeframes)

NetProfit: 91.66 %
GrossProfit: 155.52 %
GrossLoss: -63.86 %
Profit Faktor: 2.4352
Gain2Pain Ratio: 1.4352
Number of Trades: 69
Number of Winning Trades: 28
Number of Loosing Trades: 41
Max Drawdown: currently unknown (not calc. yet)


Strategy 2: (1 year data several timeframe)
NetProfit: 299.55 %
GrossProfit: 732.86 %
GrossLoss: -433.31 %
Profit Faktor: 1.6913
Gain2Pain Ratio: 0.6913
Number of Trades: 557
Number of Winning Trades: 186
Number of Loosing Trades: 371
Max Drawdown: -30%

I think I have to much lossing trades...

 

[taowave]

Cmon.....

You seriously believe that your pitch gets you anything more than coffee and a donut at a hedge fund capital raise???

A strategy is good when you can tell anybody in a few sentences why it works, why it makes money and who pays your profits.

If you have such a strategy it will inevitably make money. All the other statistics like Profit Factor, Sharpe, GtPR mean jack shit

A strategy is good when you can tell anybody in a few sentences why it works, why it makes money and who pays your profits.

If you have such a strategy it will inevitably make money. All the other statistics like Profit Factor, Sharpe, GtPR mean jack shit

 

[ionone]

I think a strategy is good when it makes money, with low DD (<20%)
over 28 pairs for 20 years on 9 timeframes with zero spread (just to know the edge)

AND

when it has only a few parameters to avoid curve fitting

good luck with that

Jeff

 

[IamaMars]

In fact, there are some other important points, a strategy must be clear to you so that you can apply it on the market not only mechanically (well, there because the level is strong), but also with an understanding of its features and construction in general. You should understand that one strategy cannot work on all types of assets, so most likely you should have more than one. If we talk about stability and profit indicators - you have identified the right figures for yourself, of course they are approximate, but if you reach such a yield, then you will have nothing to worry about, especially if you trade regularly and generally well navigate the market, so as not to miss the most profitable price bursts.