SPI 200 Futures!

[blink231]

Many thanks for your charts.

Not sure why the thread got closed (censored).

There is some algorithm in there that quotes at a much tighter spread than what the "old" index arb bots were doing. If I remember right index arb had a 4-5 point spread so the bots were quoting outside of that ... but this is definitely no longer the case. Also check the overnight SPI order book. It used to be empty and super illiquid. But now somebody is constantly quoting and maintaing quite a tight spread.

It must be some fairly strong (in size) algo that spreads us against whatever ... S&P futures would be my first guess .... but I don't really know.

 

[CanOz]

How about using a long term chart of the XJO and an ATR?

 

[Trembling Hand]

Nothing gobsmacking, just curious whether anyone has a good understanding of why the range dropped so markedly at the end of 2011?

Because IMO prop has given up trading it. There are NO more locals.

 

[blink231]

Following graph shows SPI 10day MA intraday range since 2010. We've been that low before. However interestingly volatility seems to quickly shoot back up once we have reached these low levels. This is obviously related to world events ... so maybe I am just reading something from this chart that does not really exist. I guess we are about to find out.

 

[CanOz]

Its just volatility....

I pulled up a half dozen indices around the world and they're all the same....a bull market is usually accompanied by low volatility...

The SPI also has low volume and is at times almost illiquid looking.

CanOz

 

[sinner]

How about using a long term chart of the XJO and an ATR?

Just for the record CanOz, when you are looking at a longer term chart of ATR, or StdDev or similar measures you should really normalise by a mean price or current close because 10pts ATR when the index is priced 100pts is different from a 10pts ATR when the index is prices 1000pts.

The most common normalised historical vol measure is probably 'bollinger bandwidth'. This only takes into account closing prices, you can do the same with ATR.


The pane 3 is BollingerBandwidth(20,1) -- that is, 1 stddev normalised by the 20 period mean.
The pane 2 pane is ATR(14)/Close -- that is 14 day average of True Range normalised by the current close
The pane 1 that is my own personal code so I won't disclose it, but you can see it's not that different anyway. Less smoothing basically.

However, since XJO is not a really representative due to staggered opening times etc I've included BHP to show a better example of intraday vol.

In this case pane 1 is bandwidth, pane 2 is prop sinner, pane 3 is normalised ATR.

 

[Trembling Hand]

There is a problem with range as a way to measure volatility for an intraday trader.

Rarely do you actually trade the daily range. If you have days when it just opens and grinds in 1 direction very slowly all day it is very hard to trade. If you fluff up a few trades in the morning going against the direction by the time you flip there is nothing left to get you back to even. Sooner or later you should start seeking out something that is trade-able.

 

[sinner]

There is a problem with range as a way to measure volatility for an intraday fader.

Rarely do you actually trade the daily range. If you have days when it just opens and grinds in 1 direction very slowly all day it is very hard to trade.

There, fixed it for you.

I have a couple of completely dynamic setups which hold until the close, they generally trigger and perform nicely (both psychologically and R:R) on the trend days you describe.

Just looked at the stats for one of those setups, it's triggered only 68 times since Jan 1 2008 (admittedly not on the SPI) profit factor 1.4ish.

 

[Trembling Hand]

My point was in reference to why there is no more prop traders and locals on the SPI. Just small retail fighting it out with the arb bots. Prop traders are active, 20 trades a day at least. No way in the world you'll find that on the SPI.

 

[sinner]

My point was in reference to why there is no more prop traders and locals on the SPI. Just small retail fighting it out with the arb bots. Prop traders are active, 20 trades a day at least. No way in the world you'll find that on the SPI.

Yeah fair enough I guess, but dunno why the prop traders have to trade all day every day, instead of trading market regime appropriate setups?

I mean, if one of those setups I mentioned above was triggered I would be in and watching/waiting for the close, rather than looking for an exit to enter the next trade, market is telling me something on those days.

 

[Trembling Hand]

Yeah fair enough I guess, but dunno why the prop traders have to trade all day every day, instead of trading market regime appropriate setups?

I mean, if one of those setups I mentioned above was triggered I would be in and watching/waiting for the close, rather than looking for an exit to enter the next trade, market is telling me something on those days.

Funny isn't it. Retail wants a few good setups, maybe even 1 per day. Prop wants 100.

 

[sinner]

Funny isn't it. Retail wants a few good setups, maybe even 1 per day. Prop wants 100.

Well I'm happy to trade higher frequency if I think the market regime means I can get away with it. And classify myself as retail.

If the regime is indicating the market is likely to run away from you, then standing aside or joining in (even if it means you enter at the open and exit at the close) seems to make more sense to me than keep standing in the way.

 

[blink231]

There is a problem with range as a way to measure volatility for an intraday trader.

Rarely do you actually trade the daily range. If you have days when it just opens and grinds in 1 direction very slowly all day it is very hard to trade. If you fluff up a few trades in the morning going against the direction by the time you flip there is nothing left to get you back to even. Sooner or later you should start seeking out something that is trade-able.

Couldn't agree more. Though I could still successfully trade the SPI until a few weeks back. The "grinding" has always been a problem. But what has been bad just got a whole lot worse recently. Now the contract indeed seems to be finally completely dead.

Time to move on to better pastures.

 

[CanOz]

Just for the record CanOz, when you are looking at a longer term chart of ATR, or StdDev or similar measures you should really normalise by a mean price or current close because 10pts ATR when the index is priced 100pts is different from a 10pts ATR when the index is prices 1000pts.

The most common normalised historical vol measure is probably 'bollinger bandwidth'. This only takes into account closing prices, you can do the same with ATR.

View attachment 46762
The pane 3 is BollingerBandwidth(20,1) -- that is, 1 stddev normalised by the 20 period mean.
The pane 2 pane is ATR(14)/Close -- that is 14 day average of True Range normalised by the current close
The pane 1 that is my own personal code so I won't disclose it, but you can see it's not that different anyway. Less smoothing basically.

However, since XJO is not a really representative due to staggered opening times etc I've included BHP to show a better example of intraday vol.
View attachment 46763
In this case pane 1 is bandwidth, pane 2 is prop sinner, pane 3 is normalised ATR.

Thanks for the clarification Sinner. You are much more technical than I, and more so its seems since i was last on here frequently. To what do you owe your knowledge?

Cheers,


CanOz

 

[sinner]

Thanks for the clarification Sinner. You are much more technical than I, and more so its seems since i was last on here frequently. To what do you owe your knowledge?

Cheers,


CanOz

I would say the technicality comes from mostly reading a **** tonne of financial research as well as reading quant blogs and speaking to quants who are willing to talk to noobs.

The CompSci background allowed me to pick up a lot if it really quickly since there is a huge overlap. From the compsci perspective it's just a time-series which needs processing.

First to admit the main reason I never blew up is because the market didn't punish early transgressions too hard (I put this down to pure luck) and the huge trending environs which took place post GFC when I was getting the 'swing' of things. I bought a lot of resource stocks in Nov 2008 and sold them when I moved to Melb. I bought a lot of silver and sold it into the spike. I traded the **** out GBPUSD and FTSE during the BP thing.

 

[CanOz]

Interesting Sinner, are a fully systematic trader yet?

What markets do you now trade? DO you trade the SPI?

CanOz

 

[sinner]

Interesting Sinner, are a fully systematic trader yet?

I've developed and using models for momentum, options vol, mean reversion, machine learning, stat arb, swing trades, vol breakout. All systematic, running on equities/equity indices, each has an 'on/off' depending on market regime. Usually only one/two are running at any given time.

Working on an options collar model, research is promising and based on returns I anticipate to drop the mean reversion and swing. Maybe even wrap it around the momentum strategy and turn off that models regime switch. Should reduce transaction costs by a chunk.

But no not fully systematic since I trade discretionary as well, AUDUSD or EURUSD depending on session. Never managed to systematise what seems to pay intraday (despite sooooo many attempts). I also speculate in OTM options, pretty unsuccessfully. In both cases I don't risk a lot.

I have a bunch of really robust daily forex models but the trade size required is way bigger than me, so they're sitting on the shelf until one of those OTM options returns 500,000,000%

What markets do you now trade? DO you trade the SPI?

Overall systematic: ASX XSO stocks, three US sector ETFs, SPY, QQQ and two related futures markets. I also spend a lot of time each day looking at markets I don't trade.

In regards SPI, nope, if I'm intraday trading Tokyo hours it will be AUDUSD these days.

 

[CanOz]

You have been busy Sinner!

CanOz

 

[CanOz]

When trading sycom do i need to use a different contract? For example, i have the APM2, IB's SPI contract up on my chart now, and it includes sycom, so i should be able to trade after hours on IB then right? eSignal has two contracts, one for day and one for night, but IB has just the one yeah?

CanOz

 

[Trembling Hand]

Yep IB has just the one contract. As they should its the same contract.