Scalping the HSI & the K200 (SIM)

[boofis]

Arvo session goals:
*Take notes on how it behaves with euro openings and these new ETH.
*Reassess levels and expectations on busy vs quiet times.

 

[boofis]

.....Missed the start of that cause NT and IB stopped communicating, and was still able to take 40 ticks.

Still don't get how the asian markets move relative to each other. HSI drops a 170 points, Kospi stays flat just ticking away lol.

" Its the move after the heavy hits thats counts." Waiting....

 

[CanOz]

Meanwhile a dropped 668 HKD on the mini trying to catch that falling knife....400 was brokerage...enuf of that, need more screen time and more setups before i try that again

 

[boofis]

TH, or any other thin market traders, care to chime in on working positions. UP to this point I haven't been adding to a trade that goes against me just cause it's what I learnt through Radge but he's primarily focused on long term equities rather than bullet like futs.

 

[Trembling Hand]

TH, or any other thin market traders, care to chime in on working positions. UP to this point I haven't been adding to a trade that goes against me just cause it's what I learnt through Radge but he's primarily focused on long term equities rather than bullet like futs.

Its not thats its thin - its that its fast. When its doing nothing its easy to add position. but when its moving its very hard. I just make a decision to enter and hit out, mostly as much as I can get while its still a good R:R. If it runs away without me I just steam and move on. If it goes against me I normally haven't got much more to add.

So F knows....

 

[boofis]

Good to know mate, appreciate the input. A sloppy click of the mouse ( as I did on my 2nd trade) and lost a good 10 ticks purely cause I wanted to make sure I got a fill doh, so I think I'll set the hot keys up.

 

[Trembling Hand]

Boy she is wild and woolly today. My P& L swings are like Wayne Swans budget projections.....


Positive... nah... negative... hold-on.. positive... nah negative!!...

 

[boofis]

Haha, I've put the gun back in its holster to save on ammo

Are there any asian markets one can actually spread against, it doesn't seem like anything has enough of a correlation to anything else one would just end up in the red on two positions lol.

 

[sinner]

Haha, I've put the gun back in its holster to save on ammo

Are there any asian markets one can actually spread against, it doesn't seem like anything has enough of a correlation to anything else one would just end up in the red on two positions lol.

Spread the top 4 stocks in the index against the index.

 

[CanOz]

Spread the top 4 stocks in the index against the index.

Yeah, there's an idea....i think you can trade HK equities with IB and the tops four stocks must be a huge portion of the index...

 

[Trembling Hand]

Spread the top 4 stocks in the index against the index.

I was collecting the data for a few months on the 44 stocks and the HSI. Stats strangely vary greatly from day to day but most def the stocks lag the index moves, especially at the extremes.

In the end I didn't bother doing anything........ seemed too much like .......work!!

 

[sinner]

I was collecting the data for a few months on the 44 stocks and the HSI. Stats strangely vary greatly from day to day but most def the stocks lag the index moves, especially at the extremes.

In the end I didn't bother doing anything........ seemed too much like .......work!!

Isn't that exactly what you'd want? If they move too closely together then the spread is no good for trading. The lag (especially at extremes!) is whats providing you with opportunities to spread that have an R:R worth taking.

I suggested the top 4 top stocks mostly for simplicity, didn't want to get into a long technical thing about it, but you can do this really easily with a bit of programming:

1. Each day build 11, 4 stock portfolios from the index universe (equal weight to each stock).
2. Rank the portfolios based on their intraday cointegration scores for the last N days. Need to figure out the sweetspot but my guess is N=3~5 is probably good.
3. Trade the best ranking portfolio as the other leg of your spread for the day. You may wish to apply a hedge ratio based on volatility or whatever to either leg.

Once you've got a simple algorithm like this you can build on it to generate portfolios which aren't equal weight, etc to fine tune cointegration rankings.

 

[Trembling Hand]

Isn't that exactly what you'd want? If they move too closely together then the spread is no good for trading. The lag (especially at extremes!) is whats providing you with opportunities to spread that have an R:R worth taking.

What I mean is the spread between the futs and the cash index constituents varied greatly. So it made it hard to see how one would take entry signals when 1 day the spread would be 15 points avg diff the next day 35.

I suggested the top 4 top stocks mostly for simplicity, didn't want to get into a long technical thing about it, but you can do this really easily with a bit of programming:

1. Each day build 11, 4 stock portfolios from the index universe (equal weight to each stock).
2. Rank the portfolios based on their intraday cointegration scores for the last N days. Need to figure out the sweetspot but my guess is N=3~5 is probably good.
3. Trade the best ranking portfolio as the other leg of your spread for the day. You may wish to apply a hedge ratio based on volatility or whatever to either leg.

Once you've got a simple algorithm like this you can build on it to generate portfolios which aren't equal weight, etc to fine tune cointegration rankings.

Like I said... too much work..:

 

[boofis]

Isn't that exactly what you'd want? If they move too closely together then the spread is no good for trading. The lag (especially at extremes!) is whats providing you with opportunities to spread that have an R:R worth taking.

I suggested the top 4 top stocks mostly for simplicity, didn't want to get into a long technical thing about it, but you can do this really easily with a bit of programming:

1. Each day build 11, 4 stock portfolios from the index universe (equal weight to each stock).
2. Rank the portfolios based on their intraday cointegration scores for the last N days. Need to figure out the sweetspot but my guess is N=3~5 is probably good.
3. Trade the best ranking portfolio as the other leg of your spread for the day. You may wish to apply a hedge ratio based on volatility or whatever to either leg.

Once you've got a simple algorithm like this you can build on it to generate portfolios which aren't equal weight, etc to fine tune cointegration rankings.

Prob just easiest for me to spread against the MHI considering I'm not dropping 100 lots like the big fella above though the mhi concept wouldn't scale beyond 2 contracts prob so really not even worth starting....square 1 lol.

 

[sinner]

What I mean is the spread between the futs and the cash index constituents varied greatly. So it made it hard to see how one would take entry signals when 1 day the spread would be 15 points avg diff the next day 35.

That is always one of the major risks when it comes to any mean reversion bet, that your timeseries becomes nonstationary or the bounds of stationarity change too quickly to be usefully predictable.

You can use Hurst exponents, or much more practically, an approximation thereof (like the one proposed below), to determine when it's appropriate to bet on mean reversion.

Dubovikov and others. Dimension of the Minimal Cover and Local Analysis of Fractal Time Series, 2004
http://pubs.giss.nasa.gov/docs/2004/2004_Dubovikov_etal.pdf

(I have found the variation index as proposed in the paper to be quite useful way of calculating approximate Hurst exponents without significant computation required compared to the exponents themselves)

 

[boofis]

So we had one 8 tick winner on the k200 open followed by 1 tick winner and then a Stop loss hit of 4 ticks and then a scratch.
Had one good trade on the MHI followed by a 15 tick stop out trying to fade but I was too early and ambitious, followed by a 1 tick winner (scratch lol).

Trying to employ patience and longevity of account while slowly adding to the trade ideas and implicit knowledge of how these two behave.

 

[boofis]

DAX and Seng, what's leading/lagging which one to where and what outcome :dunno: The arvo session is still sim territory for me.

 

[Trembling Hand]

DAX and Seng, what's leading/lagging which one to where and what outcome.

Thats not the way to think about.

More to the point. Thats not the way to frame a trade. Too reactionary.

 

[boofis]

Point taken.

Need more time on arvo to figure something out then, morning currently makes sense. Arvo currently gets me... To the bat cave of NT to practise!

 

[skyQuake]

Dubovikov and others. Dimension of the Minimal Cover and Local Analysis of Fractal Time Series, 2004
http://pubs.giss.nasa.gov/docs/2004/2004_Dubovikov_etal.pdf

(I have found the variation index as proposed in the paper to be quite useful way of calculating approximate Hurst exponents without significant computation required compared to the exponents themselves)

Cheers! Looks like a decent approximation.

P.S. I have also discovered showing a chart to any 5 year old to be a fairly effective method as well.