Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
KJ Trading Systems Kevin Davey - Ask Me Anything (AMA)
Apologies if this has been addressed already. How do you keep track of your performance and stats across your entire portfolio?
If im not mistaken you trade dozens of strategies simultaneously, which means they've all got their own risk and performance metrics each of which combine to create your portfolio as a whole.
Why I ask is because I like keeping very detailed stats on my performance including all the usual suspects like MAE, MFE, Sharpe, R:R, expectancy, and many more. However this has up to date just been for one strategy. Im thinking it gets a little more complex when you're trading dozens of strategies.
So I guess what im asking is, do you keep stats for each individual strategy. And then also keep stats for your portfolio as a whole.
Then in practical terms, how do you actually do that? I presume you don't do all of this using spreadsheets given the sheer number of strategies. Do you rely on your trading platform for all your stats?
Thanks.
Diversification is the only free lunch
Can you help answer these questions from other members on NexusFi?
For most strategies, I use Tradestation reports for reporting, when I need to look at detail. In addition, I track all strategies with just net profit each month, and compare that to the profit I expect. That is a "high level" view that is sufficient for most purposes. I do that and can quickly see which strategies are not doing well, and then I can dive deeper into them if I need to. Most months, that might be a handful of strategies at most.
Of course, I give up a lot of detail this way, but I am more concerned about the bigger picture - how everything is doing overall. For example, the MAE of Strategy #15 doesn't matter to me at all. And if it ever did, it was when I was evaluating the strategy. One it passes that evaluation, I give up seriously watching most of the detail.
I can certainly see the benefits of having in and out of sample data when developing a system. You don't want to curve fit your system to a certain set of data, or a certain market cycle.
However my question is what about developing swing trading systems (where trades last roughly 3 to 20 days), and backtesting the system from the year 2000 till now. Over the last 15 years we've had bear markets, bull markets, sideways markets, crashes. I think we've had a wide variety of market cycles/types over the last 15 years.
Do you think one can include the entire 15 year period during their swing trading system development, rather than worrying about the whole in/out of sample issue, simply because you wouldn't be curve fitting your system to a certain type of market.
Im not sure I see the benefit of backtesting for example 2000 - 2004. Then once you're happy with that you test 2004 - 2008 etc.
I think it would be a mistake to backtest 15 years worth of data if the market was in a constant bull the entire time, but that has not been the case since 2000 (although admittedly the 2008 crash didn't last very long in the broad scheme of things I guess?).
I normally use walkforward testing, so I end up using all the data, and most of it becomes out of sample.
If you are saying test and optimize the whole 15 year period, and then using the best results from that, because of the many market types, I would personally never do it, but I see your point. I think regardless of how many years you test, the more out of sample data you have, the better off you are.
I'm not sure I addressed your real concern. If not, just let me know.
Thanks for taking the time to answer my last few questions, I appreciate it. I have another for you,...
You trade several strategies which combine to form an overall portfolio. This makes a lot of sense. Do each of your strategies focus on different markets (or at least different timeframes - intraday vs swing). The reason I ask is, if you have more than one strategy trading the same market is there a risk of getting conflicting signals? Or perhaps getting confirming signals for the same market/instrument but from different strategies?
Im just wondering how you keep everything separate. Is it just a case of sticking to one strategy per market/instrument, or do you have multiple brokerage accounts with one strategy per account (thus allowing you to trade more than one strategy on a particular instrument).
I mix it up. Different instruments, different timeframes, different strategy types, etc. To keep it simple, I have a handful of ES strategies - I trade each in its own account. That keeps things simple (for me). There are other ways to do this - trading multiple strategies in the same account for the same instrument.
I read few chapter of the books and will complete this weekend. Seems quite excellent book. Congratulations!.
searched online and went thru couple of presentation of yours on system designing and topsteptrader webinar....again impressed with your work.
My question 1.
i am working on few strategies couple of them are based on simple open close of the day, ie i do not have any parameter that can be optimized. How to do optimization or walk forward in such cases?
My question 2.
few strategies are giving good results but again no optimization parameter as stop loss-target all are fixed
for instance check this for 9 months data (5m in and 7 min) success rate is less 40%
Net profit : above 135% net Including slippage commision
max dd is less than 45%,
sharp ratio payoff ration/profit factor all above 1.5
k ratio below : 0.0500
The Following stats was for 2years data with 10m tf
(success rate is great(over 50%) hence less dd)
Net profit : above 225% net Including slippage commision
max dd is less than 25%,
sharp ratio payoff ration/profit factor all above 1.5
k ratio below : 0.0500
can i go with strategy? is it good to go for intraday? or swing ? how can i improve such strategy? without optimization or i need to add some optimization parameter?
My question 3.
how to design strategy for options(only)?
can i go for competition with less capital.(say strategy needs 1m) but can work with 10000/ and competition entry is the same.
minimum capital gives a change of winning as 20000 or 30000 can end up as winning but that cannot be done with 1 min or 100 thousand).
thanks for book/seminars and this thread
happy trading.
If you have nothing to optimize, then you cannot do optimization. You can do walkforward, but your parameters will never change. You could develop the strategy of 2000-2005 data, for example, the the "walkforward" would be 2006-present - all out of sample results.
How does the strategy perform on truly unseen (out of sample) data? That is the real key.
Options are hard to design strategies for, unless you have access to all the options price data. I typically don;t backtest options strategies, because of the difficulty in getting good data.
I use amibroker and it does not perform walk-forward if we do not have optimization parameter? how to do it manually?
i am working on some systems all with static parameter, but confused how to select the strategy.
I want to trade 2 index, all the strategies work well with one index but fail miserably with other.
some strategy have good net profit decent trades good profit factor good sharpe but drawdown is huge and payoff ratio is bad
other strategy has all the components good with less drawdowns but no of trades are huge just like intraday and winning trades ration less than 40%
one strategy has all the parameters and expentancy excellent ...winning less than 30% net profit is over 300% 2 years data but dd is huge. If we book profit after certain levels winning increases and dd also decreases but it eats up profit and annual return....
all the above are performing well on incubation period(paper trades) in lines of expected equity curve, but how to select which to trade as trading all of them is not possible for me with limited funds and it also gives contra trades so cant trade together. Please suggest.
If you have no parameters, and did no optimization, then you'd just evaluate the strategy performance on unseen data. Technically it is walkforward / out of sample, although there is nothing to do, other than evaluate the single strategy performance during the time in question.
What are your goals for annual return, and for maximum drawdown? Do either of these strategies meet your goals? I use my goals to determine if a strategy is acceptable. It either is, or it is not. Maybe both are acceptable to you, maybe neither. If both are, then you have to decide which one best meets your goals, or trade both if possible (I don;t understand why you can;t trade both together if they are different indexes)...
