Thanks guys, please bear with me here. I am new to this but fairly smart so if you guys explain it me clearly I will get it. Here is my detailed exposition. And it is really important for me to understand this. @tobbe, pick the ratio historic back-adjust. Let's pick an example. Say we have 3 contracts. F1 is the front month( expires at the end of June, 2013, say on June 28th 2013), F2 is the back month, expires at the end of July, 2013, and F3 is the one after, expires at the end of August, 2013. Say time begins on June 1st, 2013. Now from June 1st to June 28th, I go about living my life as usual, I grab data for F1, F2, F3 and use them in my backtest as the front, back, and the "back after back" contracts. Come June 28th, I need to do something to make a continuous futures series. My understanding (which may be flawed) To create a continuation chart for F1 - On June 28th (i.e the day F1 expires), Say the expiry price for F1 is x. We also have a close price for F2 which is y and F3 which is z on the same day. To create the continuous contract for F1, multiply all previous prices (i.e from June 1st to June 28th 2013) by y/x, which is the ratio of F2 expiry/F1 expiry on June 28th, 2013. The idea is simple, i.e on the expiry of the front month, the price of F1 should really be that of F2, because, F2 is really our new front month contract. The jump in question is the difference, y - x. Doing the math, we have F1 price on June 28th (after the backadjust) = y/x*x = y, and hence now we have eliminated the jump , because y(new F1 price on June 28th)- y(F2 price on June 28th) = 0 Please stop me if I am wrong here only and read no further Now assuming that I am correct, we can extend this idea to create a continuation series for F2. Here we will use the prices of F3 to do the calculation. And then we can extend this to F3, but then there is no F4, because it doesn't exist. Does that make more sense now? Please let me know, It is really important for me to understand this.
Huh?? The price of F2 has different delivery , and all the variables associated with that delivery are different (ie weather, storage, transport, political, bugs, famine, gmo wheat, currency, etc etc etc). All I see you trying to do is remove time, event, and environment premium or discount. Or maybe disregard a backwardation or contango situation. Either way, it's insane imo! The price of F1 should be the price of F1!! Not being a math guy I can't help much, but it seems to me that with the difference of the cash price and the futures price on the first day of the futures contract, you can determine (and eliminate) at least that initial premium or discount whenever you want, without the BS of the next one and the next one and the next one, and so on and so on and so on. Hopefully nothing happens to your trading account when something occurs during the month to multiply or reverse the initial premium or discount! Good luck
Ganesha - I may not understand what you wrote, but it *sounds* like you're saying hypothetical futures contract F4 does not exist and your project to create a continuous futures contract comes to an abrupt end after stitching together a contract from F1 through F3. That doesn't make sense to me. I'm not aware of any futures with less than four contract months per year (some have 12) and I'm pretty sure they all trade at least 2 and up to 8 years out (listed contracts from 2015 to 2021). In other words, you're never going to run out of F4, or F5 or whatever. Time passes -- old contracts expire and new months are listed. Note: I don't know what futures contract(s) you are interested in, but even if it's an equity index and you don't see quotes for several years into the future, as the nearby months expire, new ones will be added (listed). You won't "run out."
FWIW to the previous post, VIX mini futures only run 3 near term months. As to the OP's question, and how I interpreted it: You just don't calculate the constant maturity for the time period between F3-F4 because the data just doesn't exist...deal with it? i.e. say you were trying to calculate the constant maturity for the Feb-March 2014 time period in VIX futures (regular size). March 2014 vix future isn't listed yet. Outside of maybe calculating a synthetic product from SPX options, you just can't, because there is no exchange data yet?
You still didn't answer why you've picked the "ratio historic back-adjust" method. What benefits does it have for you over the more commonly chosen "backward price adjustment" method, and more importantly, what are the drawbacks? For example, there will be less volatility in your adjusted contract than in the real front month, does this matter? Secondly, any reason you're doing this manually yourself? As Jack pointed out most platforms have a "continuous" contract version of all futures, as does many of the tick vendors. And lastly, however you do this, you'll have to figure out at what date to roll. It's not necessarily the expiration date, could be volume based etc. Another comparison: http://www.ipedr.com/vol29/48-CEBMM2012-R00003.pdf
@ to be, I apologize I don't completely understand the nuances of different adjustment methods. One thing I can say that I prefer the ratio method over the difference method owing to the risk of negative prices. Please let me ask you this though, does my reasoning seem ok to you? I don't have access to ready made vendor data for my application and will have to do it myself.
Apologies for sounding negative, but I think you're missing the point. You can't trade the prices you're dreaming up, and, any analysis will therefore be misleading. Plus, as someone said earlier, the contango/backwardation will be entirely messed up with what you're trying (and failing) to do. How'd you go, finding an ETF to .. solve the whole problem? Chase the money, and don't get tangled up in retail pseudo-quant mumbo jumbo.
I disagree, standardizing futures contracts is a very commonly use exercise for backtesting. I know of ample examples of people who trad futures, just trying to understand the procedure. These are real contract which don't have a liquid ETF