43) Correlation Heatmaps for Trading Strategies | Reducing Portfolio Risk

[@medievalbatsy]

Liking before watching is the true premise here! Thanks again Martyn for keeping up with those great videos.

[@seanjohnson7145]

Thanks Martyn!!! Great mini series - looking forward to portfolio balancing!

[@jocollo979]

Great job with this mini-serie video regarding to portfolio correlation and how to measure it. Thanks!! 👏🏼😃

[@Estrav.Krastvich]

It was ultra quality of studying experience for me. Thank you so much for this highly specialized course! Special thanks for your pure (I suppose British) English. It was absolute pleasure for me as a non-native speaker to understand every peace of information you share).

[@alastairferris6184]

Hi, It would be great if you did a mini series on how you set up your excel sheets and graphs. It would be very handy for us non excel users.

[@Martian4x]

Thanks for putting these together.

[@leonjbr]

Clear and usefull.

[@georgimanov8257]

Dear Martyn, your mini-series was very helpful for me in understanding the topic of diversification. However, is it possible to make a video where you present your view on when a trading system stop working and consequently we should stop using it?

[@godfredamoah6297]

Hi Martyn, when estimating the correlation of results from different strategies, do you only consider the negative results? Thus, do you set negative results to 1 and positive results to zero. I suspect that this would generate a more practical view of the correlation among the strategies within the portfolio in periods of distress, which is exactly what diversification is intended to reduce. Thanks!

[@Darwinexchange]

It's very timely that you posted this comment because I have been thinking about this myself over the past week. The way that I was considering this is as follows... If we imagine that there are two strategies that are perfect (this will never happen of course but it helps to inform the thought process). These two perfect strategies will always increase their own equity curves. So, in terms of correlation they would have an extremely high correlation value. But they are both fine to trade together because they don't increase risk, despite the correlation value between them being high. So I agree with you that it might be beneficial to only consider values where one or both of the strategies exhibit a loss. I wouldn't however use the binary method (0's and 1's) that you propose. I'd probably use the actual % change values. Hope this helps. Martyn

[@Estrav.Krastvich]

10 of 10.

[@valueray]

Which values in correlation are acceptable in your opinion?

[@Estrav.Krastvich]

Less than 0.5 I think.

[@bishalsahoo3934]

That's some great content Martyn! Thank you. I had one doubt. While computing the correlation coefficient, should we use Pearson correlation or Spearman (ranking based)? Pearson would capture linear relationships whereas Spearman can cover monotonic relationships as well. I am confused about which is the right choice here?

[@Darwinexchange]

What a great question. My feeling is that Pearson (which I use in the video) is actually correct. My reasoning here is that I don't think the relationship between two assets would ever be monotonic. However if you are a statistics expert (which I do NOT profess to be), then I would love to hear your reasoning if you think otherwise. I'm always happy for challenge and like nothing more than improving my own thinking and rationale by learning from others. Thanks, Martyn