Code: https://docs.google.com/spreadsheets/d/1CGV-e51EmIzje44CO4nvZJV6QYkbgo8tzYcLL7_rk3w/edit?usp=sharing Not having high hopes that anyone will bite - but I'm trying to analyze my pairs trades performance over the last 2 years. Attached is a spreadsheet which shows my daily pnl, my margin at work and a running total. The first graph is a running total in $$$ - the 2nd graph is a running total in % of margin - you can see in column B (margin) that my buying power reduction fluctuates quite a bit - from 40k margin on the lower end to 260k margin on the higher end. The 2nd tab shows a running daily average return percentage over the last 200 days normalized for margin - so view into potential draw down over 11 month. I'm curious to how I should look at this - in my head this all looks great and I should put all my money to work with it - however another way to look at it is there's big sideways period without any real return. Please keep ideas like "just try it" - or "this looks terrible" to yourself - not sure it's helping anyone. If you know how to analyze this and point out potential pitfalls - I'd love to hear from you - if you have a ton of experience in this - i'd be happy to pay for a couple of hours of consulting. If you've made it this far in the post - thx and I hope you can show me some blind spots here
Nope - not you - definitely my user error - should work now via the link - pls let me know if it doesn't - thx for opening btw !!
It depends on your goals. In my case, I don't do system trading, I have a portfolio and most of my trading is off margin of that portfolio, so my IRR has to cover the margin rate (which is now increasing) and the market benchmark (S&P and Nasdaq). In other words, if I'm not generating alpha, why trade? However, the caveat is I've been doing it for long enough where brief periods of poor performance wouldn't make me stop, just make me adjust. I think the last 2 years has been unusual and shouldn't be the basis of the next 2 years, or 20 years, so be careful of drawing conclusions. I wouldn't put all my eggs into this one trade, but that's just me. GL. Edit: I just realized this was likely about currency and not long/short pairs trading that I sometimes do with stock and options. I don't know anything about currency, so disregard.
I'm not seeing the significance of the system mechanics when analyzing the P&L - how do you look at this differently? To me it's an input - output analysis - i put X $$$ to work - I got Y $$$ back. Would it help to overlay S&P or VIX in the graph - I'm not even seeing much correlation between those and the system return. If this wasn't pairs but let's say a momentum based futures system with this P&L - what would you conclude?
RAR risk-adjusted return would be my one and only concern, while of course always trading and/or investing within logical safe margin levels.
RaR over 2 years is 1.97 or 197%. However over a 6 month period the numbers look very different - from negative .4 to 2.96.
I am on mobile and away from home, so I am going to comment on what I did with someone in the past about pairs trading. We looked at correlations between those pairs and set a correlation parameter in order to trigger trades. That parameter was the threshold to start analysing data, we did not care about pairs that were correlated, we only considered the ones that weren't. The principle was to look for breaks in that correlation that showed anomalies in those markets with the simple idea that if pairs weren't correlated before and suddenly they started behaving correlated they should be uncorrelated again at some point, unless a major event happened. I believe arbitrage folks could add more into it. The reason why we looked at uncorrelated pairs is because pretty much everyone else look at the ones that are correlated first and when there's a break in that correlation, the pressure coming from market players is that strong that the pairs become correlated again almost instantly. It showed a profitable system at a low risk. The problem was that those opportunities were rare and the system took weeks to trigger a trade. Past performances won't lead you to a better system as those breaks in correlation are random and can be for any reason. The main point is to improve the trigger that would catch anomalies in those markets. We normally tend to look at ratios, drawdowns and stops in order to search for an improvement but that would only improve the past as that is the input data, and it is not guaranteed to behave the same way in the future.