Anyone here have experience with how to deal with corporate actions and how one should adjust the price series accordingly so that, eg, your VaR makes sense? Obviously if it’s a share split one simply adjusts for that. The problem for me is what if it’s a special dividend or something whereby the price does a big enough jump to generate a VaR scare? Is there a rule out there, or a specific way people deal with this in terms of massaging a price series so that this doesn’t happen? Any help or direction would be much appreciated.
What’s different about adjusting price for a split than for any other corporate action?
Well the maths for a split, and the logic for me, are seemingly straightforward. The share splits 10 for 1, you multiply the price by 10. When it comes to dividends, it just seems very nebulous - so the share price is x, they pay a dividend of y that is larger than usual producing a price drop of, say, 15%…what do you do? Where is the line? That price drop will come through in VaR unless you have adjusted the data accordingly. I just want to know if there is a hard and fast rule about dealing with these things, or do you randomly adjust some share price series and not others. I’m happy to be asking a completely stupid question (I am a muscian after all by training, sorry in that case!), just wondered if anyone had any input here.
mschumacher Wrote: ------------------------------------------------------- > Well the maths for a split, and the logic for me, > are seemingly straightforward. The share splits 10 > for 1, you multiply the price by 10. > Almost certainly not - you back-adjust so the current price in your data matches the market price. > > When it comes to dividends, it just seems very > nebulous - so the share price is x, they pay a > dividend of y that is larger than usual producing > a price drop of, say, 15%…what do you do? Multiply all prices prior to ex-dividend price by 0.85 > Where > is the line? There is no line except the one that says you adjust for all dividends and corporate actions. > That price drop will come through in > VaR unless you have adjusted the data accordingly. Yep > I just want to know if there is a hard and fast > rule about dealing with these things, or do you > randomly adjust some share price series and not > others. > Of course not > > I’m happy to be asking a completely stupid > question (I am a muscian after all by training, > sorry in that case!), just wondered if anyone had > any input here. There might be some technical analysis applications for which the reinvested multiplicative back-adjusting isn’t right but these are not about VaR but about system trading.
JoeyDVivre Wrote: ------------------------------------------------------- > mschumacher Wrote: > -------------------------------------------------- > ----- > > Well the maths for a split, and the logic for > me, > > are seemingly straightforward. The share splits > 10 > > for 1, you multiply the price by 10. > > > Almost certainly not - you back-adjust so the > current price in your data matches the market > price. You telling me that if a share splits 10:1 you don’t just take each days closing price up till that point and multiply it by 10? Maybe that’s where I’m going wrong…it will make all corporate actions more consistent if that is the case, just doesn’t seem to make intuitive sense? Do you then adjust for each and every dividend, no matter how large or small? Thanks for your time btw, much appreciated.
mschumacher Wrote: ------------------------------------------------------- > JoeyDVivre Wrote: > -------------------------------------------------- > ----- > > mschumacher Wrote: > > > -------------------------------------------------- > > > ----- > > > Well the maths for a split, and the logic for > > me, > > > are seemingly straightforward. The share > splits > > 10 > > > for 1, you multiply the price by 10. > > > > > Almost certainly not - you back-adjust so the > > current price in your data matches the market > > price. > > You telling me that if a share splits 10:1 you > don’t just take each days closing price up till > that point and multiply it by 10? It’s a really bad plan because your price for MSFT, e.g., would look like 12000 (or I don’t know what) and you wouldn’t be able to tell errors from real prices. Everyone back-adjusts. > Maybe that’s > where I’m going wrong…it will make all corporate > actions more consistent if that is the case, just > doesn’t seem to make intuitive sense? > > Do you then adjust for each and every dividend, no > matter how large or small? Sure (although getting good data like this is important and can be a challenge). > > Thanks for your time btw, much appreciated.
screw Var
Thank-you for the insightful post nailit.
you got it, JoeyD 
nailit Wrote: ------------------------------------------------------- > screw Var post of the year? short and to the point?