Skip to content

RESEARCH

tempo_000010_orologio_panorama[1]

NIGHT RANGE AND HOW IT IMPACTS ON POSITION SIZING

I’m writing this research post because I’m facing a really difficult problem: night SPY range follow VIX values. With high VIX values we see high average gain or loss during the night and viceversa with low VIX values. I made the backtest from 1993 to present. The table starts with VIX value at the close. I split the numbers in 5 VIX points intervals. The AW and AL (average winning and average losing) in % grow from minimun 0,08% (AW first number) to a maximum value of 3,35% when VIX is greater than 80 at the close. I’m asking if it’s better to adjust our number of future contracts or future margins? Or is it better to simply bet the same number of contracts each night with the same margin too? I really don’t know… If you’ve got any idea just write…

< > NT %W PF AW AL MAX MIN
10 0 4 25 0,19 0,08 -0,14 0,08 -0,34
< > NT %W PF AW AL MAX MIN
15 10 949 54,58 1,19 0,21 -0,21 1,16 -1,23
< > NT %W PF AW AL MAX MIN
20 15 831 54,87 1,17 0,34 -0,35 1,9 -2,03
< > NT %W PF AW AL MAX MIN
25 20 421 54,87 1,14 0,48 -0,51 2,91 -2,68
< > NT %W PF AW AL MAX MIN
30 25 198 56,57 1,11 0,62 -0,73 2,68 -3,67
< > NT %W PF AW AL MAX MIN
35 30 101 53,47 1,07 0,89 -0,96 5,52 -2,94
< > NT %W PF AW AL MAX MIN
40 35 51 56,86 1,03 0,89 -1,14 2,59 -2,9
< > NT %W PF AW AL MAX MIN
45 40 52 55,77 1,12 0,95 -1,06 4,09 -3,14
< > NT %W PF AW AL MAX MIN
50 45 30 50 0,76 1,06 -1,39 2,65 -2,89
< > NT %W PF AW AL MAX MIN
55 50 15 60 1,33 1,55 -1,74 3,31 -2,77
< > NT %W PF AW AL MAX MIN
60 55 14 42,86 1,03 2,28 -1,66 4,71 -2,86
< > NT %W PF AW AL MAX MIN
65 60 9 22,22 0,24 1,77 -2,07 2,67 -4,34
< > NT %W PF AW AL MAX MIN
70 65 12 50 0,98 2,07 -2,1 6,07 -8,32
< > NT %W PF AW AL MAX MIN
75 70 3 66,67 3,16 2,66 -1,68 3,02 -1,68
< > NT %W PF AW AL MAX MIN
80 75 1 0 0 #DIV/0! -1,23 0 -1,23
< > NT %W PF AW AL MAX MIN
85 80 2 100 #DIV/0! 3,35 #DIV/0! 4,04 0

 

Subscribe Nightly Patterns

SUBSCRIPTION BUTTON

3 Comments
  1. Alex Argyros permalink

    I’m afraid that I will be of no help, but have you thought of asking Rob Hanna? He has been tremendously helpful to me with less intriguing questions.

    Liked by 1 person

    Reply
  2. Terence Doherty permalink

    Hi Marco,

    This is very intriguing, and I suspect it can be exploited. Here are my suggestions (just shooting from the hip here):

    1) The first thing I would do would be to split the trades into winning trades and losing trades, then analyze the relationship between the magnitude of the VIX and returns separately. I suspect there is a relationship between trade direction and VIX: as the spot VIX increases, short winners exceed long winners, both in terms of percent successful trades, and in terms of the ratio of profits on winners/losses on losers.

    2) I suspect another factor is at work here. I think as the VIX decreases, you are getting increasing input from noise, and trading noise is leading to whipsaws and just losing trades because the sensitivity of your methods for detecting trend direction and strength is unacceptably low.

    3) Another way to think about this is analogous to a signal:noise ratio. When the VIX drops below some cutpoint, your signal:noise level decreases so that you are not reliably distinguishing signal from noise anymore.

    4) To me, the VIX is not really a direct measure of volatility. Rather, it measures predicted volatility by evaluating the volatility of pricing of two hypothetical options contracts on the SPX that expire in 15 and 43 days on a rolling basis. That prediction can and frequently is wrong, in terms of predicting what the spot VIX will be at some point in the future. That’s not at all to suggest that the VIX is worthless (it definitely is valuable), but it does suggest that the VIX might not be the best benchmark for our purposes. The futures trade based on price, not expected volatility, so it strikes me that a price-based metric would be better for comparative/analytic purposes. So as a more direct measure of price variability, I suggest you re-do the analysis, except use the actual variability in price (average true range). So, calculate the absolute value of the daily price range as a percentage of the average price.

    5) I think calculating the ratio of mean and median profits on winners/losses on losers will be more instructive.

    Terry

    Like

    Reply
  3. aimdal permalink

    Thank you Terry for these good ideas!

    1 You are right for the VIX signing the market direction. I this post I just looked at the VIX absolute value not the daily returns. I will do it. I can anticipate when VIX rise, market fall. Larry Connors was one of the pioneers with this trading style.

    2-3 I never thought about avoiding the lowest range nights. But can be useful exploiting it as a noise filter in order to avoid paying useless commissions and slippage.

    4 And I agree with you that Average True Range is the best tool for our purposes. I will backtest everything again with it.

    5 I prefer PF (Profit Factor) because alone can tell if a system is good or not. Anyway I will try to add it.

    Again thanks for advising!

    Marco

    Like

    Reply

Leave a Reply

Fill in your details below or click an icon to log in:

Gravatar
WordPress.com Logo

You are commenting using your WordPress.com account. ( Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. ( Log Out /  Change )

Cancel

Connecting to %s

%d bloggers like this: