Hi guys, hi from Andrea Unger.
Today I would like to discuss why…a question which often happens to arise…
Why do the trading systems deliver worse performance in live trading rather than in backtest?
So actually this is a common question because people work, develop a trading system and once it goes live after a while it delivers poorer performance or it even becomes a loser.
The point is that on one side obviously, we tend to develop something that looks nice, because you would not try something that loses money and even though we might be skilled we might be smart enough, we will somehow insert some level of overfitting, which means we will insert rules that even though are logical, force a little bit the system and therefore adapt too much on the past.
It’s like if we go to a tailor and we get a suit and we use measures from our body which are not the very last but are an average of a diary we kept over the years and therefore once we have our new suit we put it on and we discover that we put on weight and so we no longer fit exactly how we wished into it, which is…
It does not be we have to take the very last condition of the market, of course, it’s just an example of how average measures of the past do not always meet the expectations of the future.
So let’s say, we can limit obviously overfitting we can limit it based on our experience on our common sense and so on, but then there is another point.
When we develop a system we normally find an edge in the market, some inefficiency that is there.
The point is that we are not the only one out there doing this and the higher the number of traders discovering that inefficiency the weaker the inefficiency starts to become, which means that the higher number of traders trading a certain edge in the market the lower the edge comes out to be.
If we discover, we know, that we can enter here and exit here for some reason and we do it, if we are hundred thousand doing the same, each other would try to maybe enter a bit higher and exit a little bit earlier and that kills the inefficiency itself.
At the end, it is no longer an advantage for anybody.
It’s like when you go on holiday, on vacation and you come back on Sunday and everybody comes back on Sunday so obviously it creates an awful amount of traffic so there is the clever, the smart return on Monday and that was something I did myself and people did it and they found no traffic but slowly more and more people started coming back on Monday and at the end of this story also on Monday there were queues all over the roads.
So the solution would be either you come back on a normal…on one of the days or you can bake on Tuesday, which might be possible, but it might create problems on your job for example, because you need one extra day off and at the end of the all story you have no longer a real advantage on coming back on Tuesday, because ok, you don’t find a queue on the road but you need an extra day off which is not what you desire or even not what your boss wants you to do.
So the same thing is on the markets, it might happen that we find an edge we trade it but if the edge becomes so small at the end the gain we could make out of that edge is no longer compensating the costs that we face to trade that edge.
So if I gain 10 and I spend 20 to trade those 10 I’m an idiot or obviously, I should avoid doing that, this is the point.
So, overfitting, even kept to a minimum level somehow goes into the systems.
Each rule we had is sort of overfitting and this is something that we cannot avoid.
We can avoid the heavy overfitting, well, we put rules just to clean up the system as much as possible, but we will never avoid completely a small level of overfitting.
On the other side, the edges in the market tend to disappear or to become smaller and smaller depending on the number of players that discover these edges.
This is the main reason why things that worked in the past may stop working or work to a much lower extent in the future.
This is it, I hope it helped, this is the reason why you have to be… keep on being curious and study and always try to find new ways to trade markets, new instrument.
You’ve always to be active with your mind in the markets see you next time ciao from Andrea Unger.