Alex, I agree that the correlation between ES and ER2 is pretty high (I trade the ER2 full-time and the ES is my main reference). I also have traded them against each other a few times. But how on earth do you get a correlation of >0.95, by changing the contract size? Isn't the correlation entirely dependant on the relationship between the indices, rather than money management? Also, regarding your high figure, what kind of data (timeframe/tick/frequency) are you basing your correlation quantifications on, what are your session settings and what length is your reference period? TIA & Cheers, Scientist
Since I am scalping the DAX and YM, correlations are extemely important for my type of trading. I'm not sure if you can do this in eSignal, but in CQG you can plot the correlation as the market is moving. It is interesting to see how the correlation whips around all day (depending on the bar interval, of course). I will attach a chart of several main ones in a minute. If you tell me what time frame you are interested in, I can save time and do those also. Pick 4 product pairs and you will have the charts. Any suggestions?
FuturesTrader, I've never used CQG, this is highly interesting stuff! If one had available calculation of correlation live and on-the-fly as you describe, then surely one could use quantified maximum correlative deviation in the past, base probability models on those figures, and exploit extraordinary deviations from the current norm as they occur in the live markets? I think this could be quantitatively superior to trading pairs based on "standard" divergence models, as most pairs traders do. Myself, I scalp the DAX as well and the ER2 (but not YM), also intraday swings in the ES and currency futures. Would love to see the charts on those 4 product pairs I have on the top of my list of correlation coefficients! Cheers, - S
Try 1m, 3m, 5m, 15m. Number of bars computed? Not sure, the minimum possible preferably, for real-time evaluation? Yes, AB is the symbol for the ER2.
Ok. I have attached 6 pairs. I assumed AB in your pairs is the ER2. A couple of things to consider when looking at these: All times are CST in the USA (Chicago). CQG automatically only correlates the period of the market where both products are open. Hence, the ES is open all day but the dax is open for 11 hours. The correlation is only for those 11 hours. DD= FDAX, EP = ES, the rest are obvious. I based the correlation on a 10-bar lookback with a 5 minute time interval. I can do as small as 1 bar/1 min which is useless, but it is there nonetheless. [/list=1] Just as not to confuse others, correlations are different than spreads. I have a live spread ratio calculator if anyone wants to check their pairs against mine. I find my spread ratio pairs to be very close and am able to plot those in CQG as well to see how the spread has moved over time between 2 contracts. If anyone is interested in that, I can post an image of the latest spreads. If you want any further correlations or whatever, just post a request and I will put it up. It can be done in a blink.
caught in the act, damn. i just didn't have the patience to get into details about what i really meant - which was not a straight bet that DAX-ES would decouple, but that it would de-"correlate" relative to YM-ES.