View Full Version : 90% success!!
09-08-2008, 10:56 AM
First let me say I'm thrilled with Ninja Trader!! And all of the terrific contributors, and developers!
Now onto the question... I've developed a fun strategy that seeks a few points on the YM by jumping onto a price wave and hoping back out of the market with a few dollars.
The backtest looks terrfic across all of the future assest classes as well as DIA and SPY.
The question is why is the strategy often showing marginally negative, returns, or more important, DOES ONE WEIGHT THE SUCCESS RATE OF THE STRATEGY IN A PERCENTAGE BASIS HIGHER THAN THE DOLLAR RETURNS?
As an aside, my intention is to manually trade the strategy across several assets during the course of the trading day, as signals manifest, and current market condition warrent.
09-08-2008, 11:01 AM
I am not sure what you mean the strategy is showing negative returns. If it is showing negative returns in your Strategy Analyzer that is because there were some losing trades?
09-08-2008, 11:05 AM
Can you supply screen shots of what you are looking at and describe in detail what you are talking in terms of the screen shots.
However, generally, how you determine what a "success" is can be quite subjective and can only be determined by the trader.
09-08-2008, 11:18 AM
Hey Josh thanks for the reply. I'm at the office now so I can't post a screen shot but will later. For now:
For example out of 100 trades 90% (5 tick target) were winners and 10% were losers (20 stop loss), of that several were closed at the end of the day. Marginal loss means like 100-200$ with gross of say 13000$.
The question is; does the percentage of winning trades in comparison to loosing trades mean more than the dollars won or lost? I could manually close a trade or manage a trade to make it a winner, I'm interested in high probability entry points...
I have fiddled with the Target Ticks and the Stop Loss ticks and it accounts for 5-10% change and that's still like 80-85% winners at worst, pretty terrific.
09-08-2008, 11:25 AM
Thank you for your post and I'm glad everything is working for you, but unfortunately we cannot tell you how to interpret your results.
Enjoy your week and happy trading.
09-08-2008, 11:29 AM
This is tough without seeing all of the numbers, and I am not able to provide you with "advice"; however I can help you interperate your results from an application standpoint.
You can have a very high precentage of winners to losers yet still have a losing strategy from a PnL sandpoint.
If you have a stop loss of 100 ticks and a profit target of 1 tick and you have:
Results could be:
99 x 1 tick profit = 99 ticks profit total
1 x 100 ticks loss = 100 ticks loss total
This means you have a losing strategy by one tick, for example. The opposite is true as well: low number of winners yet larger winners than losers can yeild a profitable strategy.
I suggest searching on the internet for win loss percentage, risk management, etc.
09-08-2008, 08:59 PM
Did Ok Today. Although I can see how that counts for much over time.
Any feedback would be great.
10-04-2008, 07:13 PM
a few things re your results.
you are not taking commission into consideration. that's a must.
% profitable doesn't show the entire story. your losses to profits are 4 to 1. meaning a good run of losses can easily put you upside down.
the max DD is almost 75% of your net gain. in my opinion a little high.
you are also seeing all of your profits from long trades. is this from general market conditions? how did it do in Sept? or is this a long specific system?
10-05-2008, 12:01 PM
Well it works if you have a really big account and trade one contract, it's not piratical in the short run.
I'm working on tick data models now with super tight stops, on the ES. Any thoughts?
10-05-2008, 12:54 PM
i assume you will be trading with great frequency. all the more reason to include commission in any testing.
when working with tick data systems it is very important to make the order placement logic bullet proof. have you ever seen a streaming flow of one lot prints on time and sales? that's someone's screwed up code and subsequent zeroed out account.
high frequency trading is very profitable for large (bc of their size they pay virtually nothing in commission) and very well connected (to the exchange's servers) algo firms.
with super tight stops your entries have to be good. less space for the trade to right itself.
10-05-2008, 03:49 PM
That came to mind.
So I was actually thinking of generating a lot of signals and using subjectivity as to whether or not to enter a trade. Like if the market internals are in agreement (trin, tick, ad) and the signal fires then take the trade with 2 contracts, small target, and a runner. Couple this with a daily goal of three net points. Run it for a while then up the contract and continue.
11-12-2008, 07:41 PM
You need to read Optimize Your Trading Edge by Bo Yoder. It will open your eyes to win/loss, risk/reward ratios and help you fine tune your "edge".
Best of luck!