Gurus of Backtest ... How to construct backtest parameters to include context?

E

easymon1

Any Backtest Gurus, please chime in here.
Advice on backtesting in context would be most excellent.
Thanks

Fast charts 1minute, 2, 3, 5min futures, preferred.

trading double tops double bottoms
https://yippy.com/search?query=trading+double+tops+double+bottoms

How to construct backtest parameters?
https://yippy.com/search?query=how+to+construct+backtest+parameters

https://www.elitetrader.com/et/thre...-text-is-cheap-lol.348337/page-4#post-5171299
-
Would you say that this chart's double bottom could have been affected by that mainstay blood simple bearish hammer (gravestone doji) at resistance that preceeded it?
If so, how can we include pre-existing conditions, ie. context, in the backtest parameters?

delete dblbot9.png


 
Last edited:
easymon1 said:
Would you say that this chart's double bottom could have been affected by that mainstay blood simple bearish hammer (gravestone doji) at resistance that preceeded it?
More...

Again, you can add any parameter you like in the backtest code, including candlestick formations (they are easy to quantify).

However, telling the computer what a "resistance" or "support" line is that's an entirely different story.

You can spot dozens of "support/resistance" lines on the same chart, all equally valid, but only one of them is the right one.
 
Tradex said:
Again, you can add any parameter you like in the backtest code, including candlestick formations (they are easy to quantify).

However, telling the computer what a "resistance" or "support" line is that's an entirely different story.

You can spot dozens of "support/resistance" lines on the same chart, all equally valid, but only one of them is the right one.
More...
Sounds kinda hamfisted.
 
Easymon,

Just another perspective... I trade short term price swings and see that as a pretty nice trade.

Entering at the candle directly above "db" and staying in the trade until price got to the line marked 100% would be a solid trade in my systems.
 
Slope Trader said:
directly above "db" and staying in the trade until price got to the line marked 100%
More...

from the 'Double Top, Dbl Bottom. Show your Opinion, Use Charts - Especially Non-Workings!' dept.
https://elitetrader.com/et/threads/...cially-non-workings-text-is-cheap-lol.348337/

this is the chart that one is clipped from with the wrsi down there.

delete-dblbot-png.236602




i've sure taken worse trades and had a nice supper on the proceeds, don't get me wrong, lol.

looking for influencing factors in the neighborhood of the trade that would provide context to a backtest to prevent the hamfisted blunt force result that is actually leaving a Lot of information on the cutting room floor.

That in mind, the 100% business as usual target that dang near any dbot dtop instructions point to is not shabby, but a Lot of them go to 200% and a respectable portion go on from there offering a cascade of entries with the T-Line Trade that Bigalow promotes as a way to offer re-entry to a trend which helps prevent the "wouldn't i look stupid if i got out and it took off without me, so i hung on and it went to zero" syndrome. lol.

indeed, i wouldn't sneeze at that trade otherwise.
we want to find parameters that backtesting to be accurate would need to include, for example today the nq was bumping up against two previous swing highs which dam for sure had an effect that a backtester should consider to be accurate imho.

Makes a trader wanna be nice to the brain and the screen time it has survived, lol.
thanks for the feedback, if you have any dbl bot top "failures" that you would paste images of that'd be appreciated. I'm going for ten 'non-working' and a total of about 50 would be great. Slap an indicator on there if you would whether you use it or not so that much more will be there to eval.

The wrse failure swing seems to make a good heads up, not entry trigger necessarily.
There usually pops up a hammer or an engulfing at sup or res, or Bigalow's ubiquitous T-Line Trade which is good all-purpose entry / exit and makes a good floor polish too.
 
With speedy computer, backtest with many inputs can be done in a snap. How much input is good enough? Often times we backtest and the results are not so positive. We discard those with not so positive results. There might be a chance that those models are giving accurate results but they are not the results we want. So, we have any models until we have the desired results. Can this model predict future events? Will the model still be accurate once the Fed stop printing?
 
userque said:
What makes you say this? I mean, how did you come to this conclusion?
More...
I used TS for 5 yeas and did every kind of back-testing; after observing a week or months of trades in forward wallk real time - perfect match. In Ninja 8 over past 2 years, I get useless nonsense. Most of the 950 robot programs in Striker, 1895 in iCannon, 700+ in Futures Truth etc. are all running on TS.
 
Dazz said:
In Ninja 8 over past 2 years, I get useless nonsense
More...
Thanks,

Same strategies? I mean, are you comparing apples to apples? Are you comparing the same strategy you ran on TS vs. what you ran on NT?
 
You must log in or register to reply here.
Back
Top