How to develop robust trading systems | Nick Radge, The Chartist
ฝัง
- เผยแพร่เมื่อ 27 พ.ค. 2016
- EP 064: The casino edge, mean reversion strategies, and how to develop robust trading systems w/ Nick Radge, The Chartist
More interviews: chatwithtraders.com
Free guides: chatwithtraders.com/ebooks
Twitter: / chatwithtraders
Facebook: / chatwithtraders
Instagram: / chatwithtraders_
Soundcloud: / chat-with-traders
Stitcher: www.stitcher.com/podcast/chat-...
![004: Nick Radge discusses the best trading systems and what makes successful traders. [AUDIO ONLY]](/img/n.gif)
This is one of best interviews there is very little subjective and "this doesnt work" bias in Nicks explanations. Pretty much all straight up facts, and the fact that he pointed clearly how important frequency in trading is (which is heavily under-looked) is great.
Im building a system using a roulette model, frequency is defintely something im considering. Much much more to it than roulette, it was more a inspired by situation
thanks - great one! Anything on curve fitting or robustness is gold
Listened to several episodes of Chat with traders recently and this is one of my favorites.
Instead of looking for a perfect system, one instead should be trying to break their system with extensive testing. That idea was pure gold!
Nick is one of the best sources of knowledge out there on systems trading. I recommend his book Unholy Grails. It's simple but is very informative on what robust strategies look like.
Great guy - Nick Radge - Thanks for sharing your knowledge !
Great
Technical
1. Systems Thinking / Strategy / Plan - Cognitive Biases, etc.
2. Back Testing / Record Keeping - Review
This guy is a zero bullshit trader. Everything he speaks about is based on data and statistics... brilliant interviews.
Wow, Nick Radge is very articulated! That is something I noticed in top tier traders!
You can construct a dynamic edge by analyzing the behaviour of mean reversions. Nick Radge explains this concept well.
Great interview, very informative. Can I ask what trading platform Nick uses to chart stocks with on a daily basis?
+Paul Christo, awesome to hear you liked the interview. Nick uses AmiBroker (amibroker.com/) for charting, I'm fairly sure...
Awesome Aaron, keep up the excellent work!
+speeddemonsliderule, thanks!!
Thought I’d review some older episodes. Enjoyed this greatly.
I see that not just Nick but other traders as well stress the importance of having a strategy. I am afraid these are newbie questions, but I must ask... Is the vertical credit spread a strategy? I understand it works best with high volatility and that there are many variables. But could I expect to be profitable with it? Thanks for the time and effort!
+speeddemonsliderule - I wish I could answer this for you, but I'm not very clued-up on options...
Thanks Aaron, if there was a way you could keep these question in mind for when you interview specifically an options trader I would certainly appreciate it. Anyway, as I said before, keep up the excellent work I enjoy listening to all these interviews you do a lot!
I don't think it's a strategy in and of itself - it's just a method to apply options... TO a strategy. I would say.
I suggest, when you design a strategy, you need more details, such as:
in what markets?
how do you put the strikes of the options in the spread?
signals for entering the trade (or filtering on when NOT to enter a trade)?
when or at what prices to exit?
etc...
you better do back tests before you strategy launch..
On the subject of robustness, if I had a tech indicator, say moving average, and changing the period from 36 to 42 makes it dramatically change from winner to loser or vice versa -- that's a bad, bad sign that what you're doing is not robust. It should be able to tolerate parameters that are "in the ballpark".
why?
This is why you optimise your parameters against multiple products to find something that works consistently In a given market.
Regression to the mean is a common method with edge, but what parameters are optimal?
Only back testing against multiple products will tell you.
@@geoffl Lets say the indicator was highly profitable at 38 but lost a lot of money at 40. You're only kidding yourself if you think it could not flip the other way next year.
@@scottydog9997 Agreed. That's a great way to test robustness, especially if you were not able to go far back in time. Just be realistic about the similarities of the security/market. For example, an indicator that works well in the US/Canada equity indexes might do terrible in the more volatile Latin American equities. Don't throw it away -- just be aware of it's limitations.
@@overcastfriday81 Yeah that's exactly right. Its quite funny, even back testing can't confirm whether your method will work or not on lower time framea due to order flow.
This backtesting system I built, only took price into account, and when trying to implement my imbalance algo into the market, I would never get filled at those prices due to the amount of people at that price point.
Yet in my back test with a win rate of 50%, it showed I would earn 500% a month on one stock.
Reality hit home when trying to forward test with small size.
does anyone work with eoddata?
the best food for brain
i like your podcast, but this one put me off. not that i buy systems, but he near the end he mentioned that he sells a system for reversion to the mean; but at the beginning said there really is no 'mean' used. traditional mean reversion trades based on a return to a price based on valuation (p/e; p/s; p/b, etc.); technical trading/analysis trades in expectation that price will revert back to an average price (m/a or range). this was a strange one (for me).
He said “Mean reversion tends to be a little more consistent”
@@tonic973 if it works for you, great. Mean reversion is one of the most ambiguous approaches to swing or longer term trading. Which mean? Why that mean and why now? We've all seen price continue to move away from a (preconceived) mean only to keep going, not to 'revert' for months or years. Mean reversion makes more sense with fundamental and/or quantitative models (i.e. grain, heating oil, wheat etc. In the context of supply) but if it works for you, roll with it.
@@JP-wq6sb i totally get what you mean , excuse the pun ha. But what Nick meant by there isn’t really a mean was that there isn’t a specific mean or rather mean isn’t objective. I think meant mean is subjective. Each strategy will have its own ‘mean’. It could be a simple pullback to a 50 or 200 ema. Or back to fair value etc.
@@tonic973 Summon, again, if it works for you that's great. I personally think that mean reversion is one of the least useful strategies UNLESS there are at least 2 or 3 supporting causes for that reversion that one can identify and quantify in real time. reversion to an EMA is quite possibly one of the most inconsistent strategies. You are correct: it is subjective, to the point of near-worthless (IMO). The notion seems logical, and it's easy to curve fit in a backtest, but I challenge you to try to break the method and I suspect you'll find it quite easy to throw mean reversion into the garbage bin. when it comes to technical analysis, I believe that more effort should be applied to find the flaws and cracks in any presumed edge. That's not a negative thing, it's part of the math. All the best.
Do one with Fous would be nice
50
I like the guests but this podcast is missing quality follow up questions
Agreed, really enjoy the podcast greatly. I think Aaron could do a little better with follow questions rather than "Hey that's great, moving on"...
there's no such thing as a robust system because of arbitrage.
Eventually, more people will find your "robust" strategy and compete away any edge you had.
You must continually find new edge.
This is really debatable.
Successful traders tend to have one edge they focus on as their foundation.
Regression to the mean, and momentum trading strategies tend to be used quite a lot, but it is also market and context dependant.
Of course where your statement has weighting is based on how a markets personality changes. How its participants change, thus new edge should be explored to adapt to this new context.
E.g flipping your bias based on current trend and new information.
I get the feeling this dude doesn't know what an API is.
radge is a cookie cutter
I really disagree with this definition of robustness.
Robustness is the quality of a system (optimized or not) consistently demonstrating a passing level of whatever objective function performance is required on UNSEEN DATA on a GIVEN MARKET. That's it.
It has nothing to do with "applying to different markets" or any other such naive thinking. Markets have different characteristics and for very good reasons. It's silly to expect a screwdriver to be a hammer. You will really narrow your range of approaches if you think that Strategy X should prove itself on the R2K but also pass a basket of low-vol commodity ETFs. Markets are in different regimes, vol conditions, and are driven by different factors. You really should not seek "robustness in different markets". That is really bullshit nonsense and is a terrible idea.
He actually understand what you are saying. There is a sentence in which he says there could be robust systems designed for specific markets.
I was confused when he said you need robustness across markets and later said that a system may be better in some sectors than other sectors, specifically raw materials verses individuals.