Say I have a good or several good strategies that I have been running for a couple of years. Now I have never optimized them ( more then for Robustness check reason) but did some heavy simple optimization on last 3 year data only and I got about 15 different versions of the same strategies that actually seems to have worked excellent looking back on 10 years intraday OS data as well ( of course I got 1000's of them but 15 really good ones). To my surprise many of these new versions of the same strategy has different entry and exit points that actually make them surprisingly diversified and combined it looks really good. Is this something You have ever tried or practiced in Your trading I mean trading several version of the same strategy simultaneously to create diversification.
I can see a benefit of this since if You only had 10 strategies you could create 100 of them thru optimization and use some genetic portfolio builder to pick the best set of strategies to minimize drawdown or maximize profit or both. What do You think of that approach ?
No I use it alot for 2 years. It just looks at the list of trades with time stamp and try to pair strategies together per Your ranking criteria in regards to correlation that might be netprofit,drawdown,montecarlo performance of portfolio based on opentrades MAE, profit/loss per hour,day,week or months or anything else You might think of and can program.If to many parameters You need Genetic Evolution for this otherwise it would take forever.
The following user says Thank You to Mabi for this post:
Thanks for your generosity on this forum and your website. I have a few elementary questions, as I am only at the early preliminary stages of this process:
1. You have been a proponent of Tradestation, and use them as your primary platform and broker. Have you looked at using Multicharts with another broker? How did those tradeoffs balance?
2. What are the specs of your main testing computer? Which step in your process takes the most computer time? What do you think the minimum should be as to avoid the long waits?
3. When you are walking forward a currently live strategy, would you put it in the penalty box if the WF parameters jumped significantly relative to expectations, but the strategy is otherwise performing as expected?
The following 2 users say Thank You to drm7 for this post:
Thanks for the questions. It is nice to know there are people at futures.io still interested in algo trading. Many people do not seem to be...
1. I have tried Multicharts, do not own it, but in the event of Tradestation platform ever becoming unavailable (doubtful, but you never know), MC would be my backup plan. Most strategies would perform about the same, AFTER you account for different rollover dates in the data - BIG point). With MC, you'd have access to many brokers, but you'd have to pay a 3rd party data vendor too.
2. I have a dinosaur by today's standards, almost 5 years old. Intel i7-4790 @ 4 Mhz, 16 Gb ram, SSD hard drive, Win 7. According to https://www.cpubenchmark.net/cpu_list.php it is the 203rd best CPU.
I run approx 80-90 charts continuously, and do optimization, all on this PC. It is just fine
Longest step: walkforward optimization Avoiding long waits: simplify your strategy. If you find yourself running optimizations that take days, perhaps your strategy is too complicated.
3. I have never been able to correlate jumping WF parameters with future performance. But realize that this is highly dependent on your parameter range choice and resolution. Bouncing performance with a stop loss increment of 1 dollar vs 1000 dollars might matter. Right or wrong, I tend to ignore this, since I have strategies that both do good, and do bad, with and without jumping parameters.
Thanks for the questions, hope this helps!
Kevin
2.
The following 4 users say Thank You to kevinkdog for this post:
Given how much importance a discretionary trader places in daily preparation and maintaining a consistent routine, I was curious to know what a day/week in the life of Kevin looks like as a full time algo trader?
Cheers!
The following 2 users say Thank You to amoeba for this post:
Thanks for the question. Not sure if my algo trading day is typical of what others do, but here it is:
1. Make sure all strategies are running, and are in sync - live accounts versus strategy positions. I usually check this multiple time a day, and evening.
2. Work on new strategies.
3. Work on other trading topics, for instance diversification, or positions sizing, or money management.
4. Read trading books, magazines, websites, etc.
5. Talk to other traders via Skype, or mainly e-mail.
6. Thinking of better ways to do what I am doing. May involve researching completely new trading styles.
On any given day, it is hard to say how my time breaks out, other than #1 always gets addressed. Sometimes I'll step away from strategy building for a while, to recharge my mental battery.
Hope this helps!
The following 8 users say Thank You to kevinkdog for this post:
This came in via PM, but I felt it general enough to share with everyone... Thanks for the questions!
As far as you can estimate, to create genuinely profitable stragies using the in built Strategy Builder on NinjaTrader8?
I have never used it, so I cannot say. But in general, the built in features of any platform allow you to create strategies. And create nice backtests. But are they any good - will they work going forward? That is the million dollar question. My feeling is that most trading software actually lures you into some bad habits - just look how easy it is to optimize a variable, add a new rule, tweak that backtest, etc. And that leads to misleading backtests.
With that in mind, I do think you can build good strategies, but you really need a solid development process that helps you avoid the pitfalls of backtesting. That is not easy.
Do you find successful intra day strategies almost always require higher timeframe analysis of some kind to generate profit?
My experience is that intraday strategies are hard to develop, period. Maybe having higher timeframes help, I have not thoroughly researched this.
Can you put your finger on an average number of indicators/conditions upon which your strategies are built, with emphasis on the intra day strategies?
My experience with any strategy is that the simpler the better. Most people fall into a trap where they keep adding rules to make the backtest better. That rarely works. Keep it simple. Try only a few conditions, maybe 1 or 2 for entry, and same for exit.
I was hoping you might be able to shed some light on the correct way to initially test a strategy in your view.
Would it make sense to take a 4 year historical period as in sample data, optimize to seek out the best parameters during this period, without over-optimizing of course, and then move forward to Walk Forward Optimization?
How do you ensure you aren't over optimizing, is it more of a feel thing rather than exact science?
How would you go about the Walk Foward Optimization in terms of size of data set it's being tested on?
Any fedback highly appreciated as always!
Thanks.
The following user says Thank You to WhatIfGod for this post:
Thanks for the questions. These are complicated questions, way too much to put in a thread post.
A few years ago, I did an Elite member webinar where I discussed a lot of this.
Also, my first book "Building Winning Algorithmic Trading Systems" also covers your questions in much more detail. You probably can find a copy at a library, or maybe online with a free trial at overdrive.com
Sorry I can't give short and sweet answers here. You have asked some great questions!
The following user says Thank You to kevinkdog for this post:
Strategies can and do go out of sync, so you have to be prepared to fix and adjust. Sometimes it is a code issue, sometimes a data issue, sometimes exchange issue, sometimes internet issue, etc. The list is long. One thing algo trading is NOT: it is NOT unattended trading. Even hedge funds, etc have people monitoring positions.
The following 3 users say Thank You to kevinkdog for this post:
I had a quick question from your book; for the Euro strategy example, you had allowed yourself one month to develop a strategy based on the defined goals.
I was wondering how much time it took you to come to the final strategy and if you went through other ideas before the final published version?
Thanks for the question, and thanks for reading the book. I developed that strategy more than 5 years ago, so unfortunately I do not have exact numbers for that. I am pretty sure I went through multiple ideas and variations during preliminary testing, but I don't recall what or how much. Trying multiple ideas etc. can be dangerous if done incorrectly, as I describe in the book.
Thanks!
The following user says Thank You to kevinkdog for this post:
I suppose what I was looking to learn was regarding the comment about maybe taking a 100 ideas to get a single tradable strategy (understand this is approximation), I assume this number includes iterations of a single goal?
So in a year would you be working with 10-15 goals?
Many who I work with aim for one new strategy per month. That is a good goal to shoot for. That might not seem like a lot, but after a year you'd have 12 strategies. And if you collaborate and share strategies with others, you could have a ton more than that.
Hope this helps!
Kevin
The following 2 users say Thank You to kevinkdog for this post:
You are welcome. I always appreciate those who have an honest approach to risk, and the challenge of trading. You definitely one of my colleagues in the industry who possess that.
We truly enjoyed working with you on this Podcast, and I hope to have you again in the future!
Matt Z
Optimus Futures
There is a substantial risk of loss in futures trading. Past performance is not indicative of future results.
Trading futures and options involves substantial risk of loss and is not suitable for all investors. Past performance is not necessarily indicative of future results. You may lose more than your initial investment. All posts are opinions and do not claim to be facts. Please conduct your own due diligence. Use only Risk capital when trading Futures.
1 800 771 6748 local 561 367 8686 email support@OptimusFutures.com
The following 2 users say Thank You to mattz for this post:
1) Do you have a specific way to kind of statistically determine the correlation of a portfolio? In your book I think you said that you would run a simple correlation of daily returns. Is that a function that can be done in Excel?
2) You made one comment in this thread that you've never been able to successfully trade multiple strategies on one given chart in TS. I've noticed that same problem, and it has a big effect on me since most of my trades on on the daily ES chart. Do you know if Ninja Trader has the same problem?
Thanks so much for your time!
The following user says Thank You to eferggbd for this post:
1) There are a ton of ways to do the correlation. You could use daily, weekly, monthly returns for example. And you could examine correlation over multiple look back lengths. Then, you have to determine what is an allowable correlation level.
As you can imagine, this makes the analysis complicated.
Then there is another issue. Say you have a crude oil strategy and a gold strategy, and both have been steadily making money over the past month. If you compare the daily returns over the past month, the correlation might be close to 1. Is that significant? Maybe both strategies have had good runs, or maybe the strategies are too correlated? So some people look at just downside correlation - they do not want strategies to all perform badly at the same time. But upside performance may be OK.
So, back to the original question, I do have a specific way, that addresses many of these issues (not all of them). I do use Excel correlation function to do it, and I create a table of correlation results that I use. Here is a small example:
2) I have never used Ninja to trade like I do with Tradestation. I do have an approach that sends Tradestation signals to Ninja, and then on to a broker. The broker only gets signals, so I can trade as many ES strategies as I want in one account, and there are no problems. Of course, adding Ninja in the mix complicates things, but it works well.
I have looked at the issue you mention quite a bit, and here are some articles I wrote on the subject.:
Dude thanks Kevin! That helped me a lot. Man you give so much of your time to help out your homies on this forum. It's greatly appreciated!
I thought of another potential idea by the way for how to run two strategies on one instrument in TS. If it's, say, the ES and my two strategies are both based on daily chart, I could have one run on the daily chart, and then I could open, say, a 60-min ES chart, and then in the code call on data2 (which would be daily ES data) and then the code would just strictly utilize the data2 for all its conditions even though its being run on the 60-min chart.
Depending on your strategy code, that approach may or may not help you. I've had issues running 2 daily charts, 1 daily chart with an XX minute chart, and 2 XX minute charts (same time or different for each).
The best approach is to live test and see if behavior is as expected.
Kevin, in one of your webinars you mentioned TASC magazine as a good source of trading ideas. In your opinion, does it hold for fx, too or would you call another source a "#1"? I am not expecting getting good strategies served on a silver plate, just ideas with more fx than futures in focus.
Sure, it is a good source of ideas for all types of markets. You may have to make some adjustments for your forex markets, but I'm sure you'll find plenty to build off of.
Thanks for the question.
Kevin
The following 2 users say Thank You to kevinkdog for this post:
Trading: The one I'm creating in the present....Index Futures mini/micro, ZF
Posts: 2,285 since Nov 2011
Thanks: 7,247 given,
4,433
received
In the spirit of encouragement.... I would suggest getting more involved here in FIO’s community and many ideas, edge related and otherwise, can be cultivated, gleaned and grown from your interactions with other members.
Try for 5 posts a day, for example. In about a week you will have doubles your post count. Just an idea.
Broker: Primary Advantage Futures. Also ED&F and Tradestation
Trading: Primarily Energy but also a little GE, GC, SI & Bitcoin
Posts: 4,034 since Dec 2013
Thanks: 3,340 given,
7,954
received
Since Kevin probably won't self-promote himself, interested readers may like to know that he just did an interview/podcast over at Better System Trader focused on his latest book on entries and exits.
Bars on Kagi, Point & Figure, Renko, and Line Break charts, bars do not form until prices have moved a certain amount. For this reason, when back-testing, it is possible to know the direction of the bar, and to guarantee that the bar is of a certain size, before initiating a trade. To illustrate this effect, the following strategy can be placed on a P&F, Renko or Line Break Chart and will show 100% profitability in back-testing: