How to tell if a strategy has lost its "edge"? World Champ example.

[TraderJimmy]

I read and hear about strategies losing edge. I understand that "edge" in this context is the unique element(s) of the strategy that allow it to have a positive expectancy...or make it profitable.

Pro's often talk about a strategy that they used 10 years ago which doesn't work anymore, because too many people discovered it, or the markets changed.

World Champ example
I was listening to a podcast with a former trading World Champ winner - and he ran through his breakout strategy which combined highest highs and RSI. So I coded it up, adding all the good things that I've learned on this forum, like position sizing, index filters, slippage etc. And it actually back tested really well.

So does this mean that:

1. it still has an edge,
2. my changes to the strategy (eg index filter) actually created a different strategy
3. my backtesting my not be robust enough

How do you know if a strategy has lost it's edge?

 

[peter2]

Sorry for the late reply, but your question is not a soup question. (ref movie Finding Forrester).

I must refer you to Howard Bandy as I believe he addresses this issue in one or more of his books.

Firstly. Yes maybe to all of your three options. Your back tested system may have an edge, it most likely is a different system or your system uses a different trading universe, or your system tested a different tine frame, etc. And yes, your back testing or your analysis of the results maybe flawed.

Trading system edges go up and down in response to the changing market conditions. They can even go negative when market conditions change drastically or very slowly and you didn't notice. Monitoring the system's edge is a must.

How you do that and how you find the right parameters to watch, well that will take a book and a lot of system experience.

Edge, btw, good description of an edge. Most systems don't have anything "unique" about them. Yes there are some that exploit something unique in a market. Pure quant systems will find unique edges. Generally this uniqueness doesn't last long or occur very infrequently. (eg going long ASX on Melb Cup day).

The edge most systems exploit is a mathematical edge and this is governed by the management of the trades. Trade management that allows the average win to be larger than the average loss exploits this mathematical edge. There's nothing unique about buying break-outs but it's possible to create a positive edge with good management in the right market environment.

 

[captain black]

I must refer you to Howard Bandy as I believe he addresses this issue in one or more of his books.

All of Bandy's books discuss this issue to some degree. His "Modelling Trading System Performance" is one I reference repeatedly, particularly the "Is it broken?" chapter. Not only does it discuss how to measure whether a trading system has lost it's edge but also important issues such as sample size etc. that determine whether or not an edge actually exists.

 

[captain black]

Dr Bandy's website also has some great introductory content in the resources section:

Resources > Blue Owl Press

This page hosts information about software, data, books, videos, articles, courses, and websites that you might find useful in trading system development. The material available is extensive, of varying quality and usefulness, and rapidly changing. I will try to keep the list reasonable...

blueowlpress.com

Several Youtube videos with Howard as well.

 

[TraderJimmy]

Thanks for the replies @peter2 and @captain black. This makes a lot of sense.

I find Bandy's work super valuable, but quite a tough slog to get through - it takes a few reads.

I also saw that he used to post here a lot, which is pretty cool, some great content there to trawl through on many topics. Though his last post was " I give up" back in 2017

 

[peter2]

I find Bandy's work super valuable, but quite a tough slog to get through - it takes a few reads.

Phew, I thought I was the only one to find his use of language difficult to understand. I have no doubt that his work would be much more popular if his books were easier to understand.

 

[TraderJimmy]

Reminds me of one of Nick's talks, where Bandy is in the crowd and steps up to answer a question about this very topic.


Nick's eyes oll back, and he says "so for everyone else out there..."

I wish I understood it straight away, but I'm still a padawan.

 

[frugal.rock]

I wish I understood it straight away, but I'm still a padawan.

If you do a small amount of study and work through a few actual exercises on "statistics", you should find it relatively simple.
It just sounds all technical like, but it's just the unfamiliar terms used.... IMO.
Basic statistics is really basic mathematics.... but difficult language.