Are there pullback systems that backtest well?

Quote from Runningbear:

There are a quite a few good pullback systems discussed on the wealth-lab.com forum.

Runningbear
More...

I don't think so. I've looked closely at many of these, and when you look closely there are fatal flaws in most of them.

m
 
Quote from lindq:

Ken DTU: backtesting? no it's the Right edge of the chart that counts

But backtesting takes the fear out of falling off the edge. Without backtesting, it would have been virtually impossible for me to put some excellent pullback systems in place.

Ken DTU: pullbacks? no it's Breakouts that attract new trend buying/selling, not pullbacks, which is essentially fading the immediate past trend.

While breakouts may attract trend buyers (often at just the time the trend is ready to end!), pullbacks will attract the greater number of private/insitutional traders and investors who like to buy low and sell higher, and who know the value of getting into a clearly oversold stock.

IMO, a very big advantage of trading pullbacks vs trends is that a trend trading win/loss ratio will typically be nearer 50%, whereas a good pullback system will often run 75-80 %. For me, and for many others, watching half your trend trades crap out because you bought at a high is NOT a pleasant way to spend the week. And finally, a good pullback system can perform well in a flat to down market when some stocks are quickly oversold. But trading a trend system in overall market weakness is very problematic.
More...

Thanks for stating the case for pullback trading so well. I agree completely.

However, what this discussion shows (once agian) is that there are a million ways to skin this cat.

m
 
Quote from Walther:

Nihil novum sub soleil... or something like that .
Only approach might be innovative, like using 3d projection of price movement to predict future price turning points. Or better yet using unusual combination of indicators and or known methods.

Shees.. I thought you know that
More...

Heh?

Walther... what about teaching a monkey to trade and wiring his brain up to a computer?

Bwahahahahahahahahha
 
One purpose of an operational definition is to be able to partition cases so that there is no overlap between the sets then you can make reliable statistics because the basics of building a probability universe is to make sets disjoint so that probabilities add to one.
As my particular case I will be able to express definitions in term of simple binary logic and avoid fuzzy logic for the reason given above.

Quote from OddTrader:

Just 2 cents:

Yes, imo. Otherwise something equivalent. :confused:
More...
 
Quote from Walther:

Nihil novum sub soleil... or something like that .
Only approach might be innovative, like using 3d projection of price movement to predict future price turning points. Or better yet using unusual combination of indicators and or known methods.

Shees.. I thought you know that
More...
If I understand what you are saying, price projections, at least the ones based on eigenvalues/vectors and projecting price onto the orthogonal vectors, are a linear transform and will get you nowhere, or I should say, never got me anywhere.

nitro
 
Quote from harrytrader:

As my particular case I will be able to express definitions in term of simple binary logic and avoid fuzzy logic for the reason given above.
More...

Another 2 cents:

As long as the rules are clearly and operationally defined (i.e. objectively), probably whether they are based on binary logic, fuzzy logic or else would not be a critical issue, I just guess. :confused:
 
Actually, funny thing is I don't trust backtest purely...

If I wanted to make a "pure" "objectively for back-tested performance", I can make different kinds of 100% profitability( no breakeven trades, also doesn't matter what the style of the system is... scalp, pullback, swing, trending etc. etc.). Every software has it's limatations. Now, understanding the platform fully (or at least to a large extent) is equally important as the underlying concept of the software.

So... taking it to the extreme...

I have a platform that works great on long term trend following systems (uhh... realistically Wealth-Lab). But when it comes to intraday market... let's say intraday swings... others do better (let's say... Tradestation 2000i)... When it comes to scalping there are others (I haven't met a good retailed one but it can be programmed)...

But each system will have flaws of it's own based on the user's needs.

If you're doing EOD with different symbols, the WL is great but the signals are slow and doesn't give them out real-time especially if the orders are triggered within a bar... like a stop order has to wait till the bar ends.

If you're doing swings TS2000i is better but it can only test one symbol at a time. Single symbol money management can be tested but who the hell cares... or what kind of a dumbass thinks like that these day... ... ... ... oh sorry, you suck if you think like that... LOL

Well, let's say I get a scalping platform. One of the likely problems would be that it'll be very slow. And the data size required to test these will be massive. The database / network / hardware management would be critical and would requrie to have the software completely dedicated to the task, making it unefficient when you do the other kinds of tests....

Etc. etc....

Anyways, whether it's the platform, system or indicators, or the symbol, it's very important to know what the hell you're doing, as clearly as possible. If you don't have that... it's the same thing as a newbie discretionary trader trying to trade without a plan of any sort = disaster...
 
You must log in or register to reply here.
Back
Top