The subject of quitting a paying job to get into trading has been discussed at length on other threads by people who have done it. The cost of a trading education in the school of hard knocks is expensive, but not insurmountable. Some people make it in a prop-shop, most don't. Again, discussed on other threads by people who have been there.
Pair Trading Strategy Journal
- Thread starter jonnysharp
- Start date
AFAIK PTF doesn't support intaday loopbacks. Which lookback period do you use for your STDEV calcs (I believe you're not using any correlation filters at all and you STDEV calcs are based on daily EST closing prices)?
Thanks for this Steven, I'll check out the other forums.
Quote from BigSalad:
The lookback is on daily data, but with DtnIQ real time you can update at desired intervals during the day. Different lookbacks for different crosses...
If loopback is on daily data (I believe that you still use hourly prices for your loopback calcs), then I would assume that most of your trades are getting closed within 1 or 2 days period, right?
What's the max loss in % you would keep before closing the trade and what is your average win % from a single trade?
Hi Bigsaled, sounds interesting, never even looked at forex with ptf... How lose are the settings for forex if you dint mind my asking... Are you using the new default 100 100 100 settings that come on ptf. I tried them on equities but prefer the older 20 14 14 since they have treated me well... I get a win ratio of 70 % with them but like you said, positions sizing is key imo....
i did however start taking trades with the new settings on another small account.
I'm curious, has anyone tested multiple time frame pairs. Taking trades when lets say the daily and the 60 or 30 are both +-2 or whatever std one uses...
Nkny
I tried paper trading on different loopbacks. When you increase stdev loopbacks you get less trades and it takes more time for them to converge, however, it slightly improves avg. returns from each such a trade.
However, from the capital usage perspective I found those improvements insignificant and less profitable after 18-20 days.
I tried also Stdev loopbacks to be less than 20, however, the most optimal is confirmed to be still around 18-20 days, while 14-16 days give you more trades each day and decrease converging time for the good trades.
I use 90 day, 180 day and 360 day correlations for my trading decisions, while finding 30 day correlation to be insignificant and totally uncorrelated with avg. returns.
However, from the capital usage perspective I found those improvements insignificant and less profitable after 18-20 days.
I tried also Stdev loopbacks to be less than 20, however, the most optimal is confirmed to be still around 18-20 days, while 14-16 days give you more trades each day and decrease converging time for the good trades.
I use 90 day, 180 day and 360 day correlations for my trading decisions, while finding 30 day correlation to be insignificant and totally uncorrelated with avg. returns.
Just a thought. A,B,C&D instruments together have >80% hist correlation coeff.
If A/B ratio line does not seem to be bouncing from 2std BBands& reverting to 20ma by 13days~half life but C/D has revert to it's own 20ma mean line and beyond, can we consider the convergence of (A/B)/(C/D) instead?
If A/B ratio line does not seem to be bouncing from 2std BBands& reverting to 20ma by 13days~half life but C/D has revert to it's own 20ma mean line and beyond, can we consider the convergence of (A/B)/(C/D) instead?
The uptick in volatility this week has been great.
In fact March has been great. 59 closed trades with only 12 losers so far this month.
It is a given that trending markets are not good for stat arb.
My experience has been that strong up trending markets like we saw in Q4 are more toxic for pairs trading than down trending markets.
Q4 has been my only losing qtr pairs trading, with only 3 out of the 6 previous months profitable. What is very reassuring is I am still profitable on balance. The best test of any trading method in my opinion is how much it loses during drawdown periods.
In fact March has been great. 59 closed trades with only 12 losers so far this month.
It is a given that trending markets are not good for stat arb.
My experience has been that strong up trending markets like we saw in Q4 are more toxic for pairs trading than down trending markets.
Q4 has been my only losing qtr pairs trading, with only 3 out of the 6 previous months profitable. What is very reassuring is I am still profitable on balance. The best test of any trading method in my opinion is how much it loses during drawdown periods.
Indeed. Trending market esp. slow up trends seems to be less favorable than fast moving down market for pairs to converge more frequently...which brings to the point: How do you use beta(systematic relative risk) VS STD (volatility) of individual instrument when pairs trading?