Trading with Market Statistics VI. Scaling In and Risk Tolerance

[darthtrader]

	Quote:	Dogpile
	then would it hold that every tick closer FROM +1.0/-1.0 StdDev TO the VWAP is superior location to the point of exactly +1.0/-1.0 Std Devs... seems intuitive that this would be the case and if so, it begs the question why trade necessarily at the +1.0/-1.0 std dev... if any entry superior to that level is also expected to be profitable then there is something to be said for improving location relative to that level through some technique/filter that adds a little value to that, ie X ATR's better than the +1/-1 Std Dev price where X can probably be calculated through back-testing to give you the most number of trades that are still profitable enough to offset the extra variance they create.

Dogpile, I think I burned out a few neurons reading this over and over trying to wrap my head around it. :cry:
I think you do have a point there though. The first part has to be correct.
Then if price is between the vwap and +1.0/-1.0 std dev, price never goes back to vwap so you wait for price to get back to +1.0/-1.0 std dev and then put on the trade in that same direction you obviously have not got in at the best point all else the same.
In your example what would you actually backtest X for though? Would that be the number of atrs away from std dev/vwap that have occured the most over Y time? That would then be an entry point to also watch for all else the same?

[Dogpile]

its hard for me to relate to this 'scaling in' thread given the example in the video. the location of the initial short is just a location I wouldn't ever consider shorting NQ -- and I trade NQ every day. I am not saying this method isn't profitable, just that personally -- I place a premium on having good trade location and use a hard but reasonable stop against that -- then often re-enter if I believe my initial read was correct and my stop just wasn't wide enough. I can see the logic in your approach Jerry but I just don't think that is optimal trade location -- hence my question above.

[dbntina]

Darthtrader...I was LMAO at your comment.

I remember reading Dogpile's post and thinking "I used to think I was semi-intelligent but know I realize I can't even follow/understand this line of thinking".

I guess it's probably because my trading iq on trader's laboratory is only like 5% so I will have to hang around until I catch on.

Anyway...that was funny.

Luis thanks for the info, I will check out ADE and see if I can't get that to work, then I will post the code...actually I will probably send it to Nick and a few others to check it out before posting to make sure it is accurate.

dbntina

[Dogpile]

darth,

I got sidetracked and meant to clean up/clarify that question -- it was pretty murky. lol.

My question was born out of seeing the example and not being at all comfortable shorting in the spot that this method (-1SD) shorted -- the initial short in the video. NQ will often make its high or low for the day before 11am EST then trend the opposite direction all day -- in fact it did that today. again, I am not saying jerry's method isn't profitable --- its just hard for me to relate to this particular example.

I was thinking just that;
VWAP is better 'location' than 1SD
and 1SD is better than 2SD
-- therefore 0.80 SD's is better than 1.0 SD.

the only reason NOT to wait for VWAP is because you are looking for incremental opportunities as it won't always get back to VWAP. Thus, you are passing up 'positive expectation' by NOT trading at 1 SD. but there should be a point between 1SD and VWAP that is more optimal than exactly 1.0 SD. Was just curious if this was correct to think like this or is there some other factor that I am just not 'getting'.

[jperl]

	Quote:	Dogpile
	jerry, was curious if this is true or false: given your assumption that SD will be generaly understated for any skewed distribution and therefore trading at the +1/-1 SD level will generally be profitable since it will be likely to go 2 SD's (if skewed) enough times to offset your stop outs... then would it hold that every tick closer FROM +1.0/-1.0 StdDev TO the VWAP is superior location to the point of exactly +1.0/-1.0 Std Devs... seems intuitive that this would be the case and if so, it begs the question why trade necessarily at the +1.0/-1.0 std dev

You are quite correct here Dogpile. If a move from the VWAP gives the smallest SD (and thus lowest volatility), then trading at a price just above the VWAP (in the case of positive skew) should have higher volatility and should be a superior location for a trade in the direction of the skew. I thought to myself that this should be correct, but then realized that the probabilites for continuation in the skew direction is declining the further you move from the VWAP, which offsets the higher expected volatility. Exactly where the "balance point is" I don't know. I decided not to pursue this because there other interferences which I call HUP which are virtually impossible to include in a more general computation. HUP will be discussed in a future thread.

	Quote:	Dogpile
	I got the sense that you were just trying to increase the number of trades and picked +1/-1 std dev as a number that seemed to have some logic to it. but I didn't 'get it' if there was another reason why this price was important.

You are correct here. The SD is a reference point rather than a point fixed in cement where you have to take a trade. When we discuss HUP in more detail in a later thread, you will see that more often then not, the entry for the trade is offset from the SD by the HUP. This can get quite complicated, so I don't want to get ahead of myself in a discussion of it here.

[jperl]

	Quote:	darthtrader
	Jerry, I'm a little confused by this part. Maybe it would have to do with going reverse on the trade but say in that example you had waited until price had gone back to the wvap to even start the trade. Wouldn't you have not scaled in at the vwap and put on your full risk tolerance right away since that will be the highest probability trade you can get?

Good question and good observation darth. Here's the problem.In a short, if price is already below the 1st SD, (or above the 1st SD in a long) the probability of it returning to the VWAP diminishes with every tick that it moves down. Thus if you expect to take a trade at all, you have to find another entry point, since price may never return to the VWAP. You could of course develop a trading style in which you only take VWAP trades, but you may have to wait a long time for an entry. The 1st SD is then the next best entry target.

	Quote:	darthtrader
	I also don't see where above the vwap you would even want to scale in since if price had gone up to the pvp would'nt you have used that as your system stop?

Yes, if you enter at the VWAP, scaling in becomes dangerous. At the PVP, other things can happen which we are going to cover in the next thread. Scaling-in there or at the next SD becomes a touchy situation, definitely not for newbies. Advance traders might do so, but can easily get caught with there pants down (as I did today, but I didn't lose my pants, only my shirt).

[Dogpile]

let me just make a point that I didn't initially get but explain it another way than the way Jerry did initially since it helps me to understand it if I try to explain it a different way anyway. (please correct anything that is wrong here).

the premise of this strategy is that for any non-normal (skewed) distribution, standard deviation will understate the 'true' volatility. thus, 2 std devs is not as far away as you might first think by the std dev calculation (ie, 2 std devs means something different if the the distribution were instead 'normal' -- normally distributed). we have observed many cases where price will not return to VWAP -- despite VWAP serving as the 'mean' for the distribution in the std dev calculation.

thus, for a skewed distribution -- you can kind of think as 1 SD as the 'new mean' (though that term 'mean' probably isn't right -- just think of it as a new potential 'pivot' since price might very well not return to the VWAP price) -- and price should be expected to rotate between the VWAP and 2 SDs (or more) -- the price pivots that are above and below 1 std dev. and if price doesn't do what is expected --- then many times you will be exiting since the PVP might have changed -- or price will trade down far enough away from VWAP just due to pure volatility to allow you to get out.

so skewed distributions are kind of funky -- you are using VWAP in the calculation as you would the 'mean' -- but it really shouldn't be thought of as the mean.

[Dogpile]

btw, until we figure out the PVP code -- I am using a screen like this -- call it poor man's PVP -- 2 windows side by side with font shrunk down to '5' (times new roman)

[darthtrader]

First Dogpile, I'm a total newb. Please don't take anything I say on here as rhetorical. That first part was ment to be funny because it was pretty true as a newb. Does that second part of my response make any sense? I think what you said make sense but would like to know how you would back test such an idea.

	Quote:	jperl
	Good question and good observation darth. Here's the problem.In a short, if price is already below the 1st SD, (or above the 1st SD in a long) the probability of it returning to the VWAP diminishes with every tick that it moves down. Thus if you expect to take a trade at all, you have to find another entry point, since price may never return to the VWAP. You could of course develop a trading style in which you only take VWAP trades, but you may have to wait a long time for an entry. The 1st SD is then the next best entry target.