Quote from tradingjournals:
Let us see if you can walk the talk. If you have two instruments A and B with volA and VolB as volalilities. How do you design the spread? Write the major steps and the equations down, the tests you would do to decide on the spread, etc? You can add notation as you need.
>> Most of the spreads I trade are fully hedged and basis point for basis point. e.g. STIRS vs. Bonds, different bond duration, etc. You don't need to sit there and custom-make spreads. But there are a select few that we can do anyway. We are not quant-based but I did learn this.
>> Anyway here's a basic guideline as to how I would create a spread from scratch for equities (assuming you believe they somewhat follow each other). It's not mathematical, but its based on observation and feel and watching it over time.
[Step 1.] First you need two products you believe fundamentally should relate to each other. I watch the Australian spi vs. the S&P500. The correlation function (using CQG) varies on the time interval for 15-min to 60-min but it hovers between 0.4 up to 0.8. So they're positively correlated to each other.
[Step 2.] Now we will calculate a "perfect theoretical spread" assuming both spreads moved in a perfect ratio.
Spreads do not move perfectly in unison. This must be done so you know you're on the right track.
S&P500
> S&P500 current price = 1808.25
> 1% move = ~18.0825 x 4 = ~72 ticks
> This represents 72 x $12.50 U.S./tick divided by 0.90, AUD rate) = $1,000
So if you entered the spread, and you
made 1% on the E-minis leg you would make $1,000 AUD
SPI
>Australian SPI Price = 5330
> 1% move = ~53 ticks = $25/tick x 53 = $1,325 AUD
So if you entered the spread, and you
lost 1% on the Spi leg you would lose $1,325 perfectly hedged
Total net change:
You made $1,000 on the E-minis. But you lost $1,325 on the Spi on the same 1% move if they moved in line. You need to be over-hedged in the E-minis to make up the Spi loss by approximately 30%.
So if you buy 10 spi, you need to be short 13 eminis provided the exchange rate stays the same, assuming the spread moves perfectly in unison.
[Step 3.] If only spreading was perfect. Unfortunately, it requires time and devotion. If you actually watch most inter-country spreads, you will see that two legs are guided by individual fundamental moves that don't impact the other leg.
Now, we want to ignore what the 'perfect ratio' says, but instead find out
exactly, based on charts/data, how much each leg has been moving during that specific time-period session.
Also, in CQG you can apply a Correlation function to spi & emini and find out its always between 0.30 to 0.80 on a 15min to 60min candle during the designated zone of Europe to U.S.A.
Most important part is here, I now applied a
6-hour ATR. My charting program doesn't allow me to collect specific amounts of data for certain times. But I used a 6-hour ATR because that seems to be the average trade time anyway. You can toggle around with it but most of your spreading execution will come based off experience.
This is filtered out for the exact same time. Notice how the ATR of the Spi (left) and the Eminis (right) follow each other. That means as the minis moves more, the spi also moves more as the weeks go on, and vice versa. So we know they're moving relatively in line with each other based on the volatility of the minis. This is what I want to see. I want to see the E-mini volatility transcending over to the other leg.
The above chart was used just to check the reliability of the spread. When the E-minis is dead (low volatility) will the spi probably be low as well? And what about the other way? The answer is yes. I used a value of 14 for the 6-hr ATR.
When I applied a small 3-slot ATR to the 6-hr ATR the chart can give us more recent information:
From the chart we can take the most recent (blue) data set. The spi moved 7-ticks x $25 = $175 , and the E-minis moved 3.35 x 4 ticks x $12.50 divided by 0.90 = $186
If you were to overlay both the data-sets you would find that the difference in ATR between the two legs varies by about ~30%. In fact, it ends up being totally random whether or not you knew to overhedge minis by 30%, keep it flat (0% extra), or under-hedge it by -30% against the other Leg.
[Step 5.] We know the perfect ratio but like I said, its always best to investigate the spreads and observe their activity during different timezones.
Nothing beats plotting a chart of a simple uniform spread and working around it. For example, I know that the E-minis leads the Spi therefore in all my trades I will consistently
underhedge the E-minis by 30%.
Instead of 10 spi to 13 eminis, I will do 10 to 10. This is because, based on feel/experience, the E-minis moves around a lot more and does a more volume. And I am assuming that the
spi is wrong and the e-minis are right. So if the eminis are up, and someone has sold the spi, I want to be long spi but not fully-hedged short minis... because if they run down they should crash enough to let me out anyway. But if they refuse to come and squeeze up, I will need a lot of 'breathing room' on that leg because the spi needs to catch up.
When the minis goes somewhere, it often bounces back and f*cks around as well.
Also from experience, being fully-hedged and copping an e-mini reversal can be very very deadly.
Real Ratio: E-minis require extra 30% contracts than spi. E.g. 10 spi to 13 eminis
Based on current market conditions/ATR: A safer spread ratio is actually under-weighting e-minis by 30% so leaving it 10 to 10.
You can apply this to any other spread out there as long as you believe they fundamentally follow each other over a certain time period.
And most importantly, nothing beats watching the spread like a hawk every day and observing how both legs react to each other. Plot the chart, and look for value. With enough hard work you will see opportunities of cheap/expensive relative value plays.
Finally, it's common to see people who spread the DAX vs. EUROSTOXX to under-hedge the DAX slightly. Even though they know the perfect hedge, they know the DAX is more whippy and can run. So if it runs onside it should flick you out, but if it runs offside it's going to do you a lot of damage and the Eurostoxx won't catch up.
