Adventures in Automation

If he was holding both ES and NQ at the same time, leverage would be about 2.3 for a 100k account. Closer to no leverage if he was only holding one at a time.

Also depends on whether he considers 100k to be his notional balance or something different.
 
fan27 said:
Results for January are as follows.

8 trades. All winners.
Up $10,097.29 for a 10.34% gain.

See attached.
Note. There will always be 1 share of QQQ and SPY on statement as it was a gift to my son that I hold in my account.
More...
Congratulations. Is the 10.34% an annualized gain? If not, what would be annualized?
 
BrooksRimes said:
Congratulations. Is the 10.34% an annualized gain? If not, what would be annualized?
More...
Thanks. YTD gain which is just the month of January. My first trade was the day after Christmas so I do not have enough live trading to determine annualized...but backtest results average over 100% annually.
 
Currently making enhancements to the PortfolioOptimization runner. Previously, I had the ability to to filter strategies and then choose the the order combination with the highest profit factor. For example, let's say we had:

strat1: orderCombo1, orderCombo2, orderCombo3
strat2: orderCombo1, orderCombo2
strat3: orderCombo1

The runner would pick the orderCombo with the highest ProfitFactor for each strategy. What I just built yesterday and this morning is the ability to rank strategies on different result metrics and weight those metrics. For example, I might have:

HighestProfitFactor: 30%
HightestTotalResult: 40%
LowestMaxDrawdown: 30%

So now, each strategy will be ranked based on those weighted metrics and the strategy with the highest ranking will be selected. Of course I am able to test different variations of metrics and weightings very quickly. I am also going to apply the same technique for ranking Portfolios.
 
fan27 said:
Currently making enhancements to the PortfolioOptimization runner. Previously, I had the ability to to filter strategies and then choose the the order combination with the highest profit factor. For example, let's say we had:

strat1: orderCombo1, orderCombo2, orderCombo3
strat2: orderCombo1, orderCombo2
strat3: orderCombo1

The runner would pick the orderCombo with the highest ProfitFactor for each strategy. What I just built yesterday and this morning is the ability to rank strategies on different result metrics and weight those metrics. For example, I might have:

HighestProfitFactor: 30%
HightestTotalResult: 40%
LowestMaxDrawdown: 30%

So now, each strategy will be ranked based on those weighted metrics and the strategy with the highest ranking will be selected. Of course I am able to test different variations of metrics and weightings very quickly. I am also going to apply the same technique for ranking Portfolios.
More...
Adding weighted metrics for strategy and portfolio selection turns out to be a major enhancement. At the portfolio level, previously I would just rank by ProfitFactor but I am finding if I add equal weighting to lowest max drawdown and highest total result, almost all other metrics improve.

I am going to spend the rest of the month porting my research over to a high powered Linux instance on AWS and hunt for more trade-able strategies. My two core laptop is just not cutting it anymore.
 
FasterQuant running without issue on an Ubuntu instance. Tomorrow I am going to create a compute optimized instance with 76 cores (compared to two on my laptop).

Time to release the Kraken!!

 
fan27 said:
FasterQuant running without issue on an Ubuntu instance. Tomorrow I am going to create a compute optimized instance with 76 cores (compared to two on my laptop).

Time to release the Kraken!!

More...
VERY GOOD, DO IT! :)
 
Great project ! (I'm following also your other threads). I like the "grey box" approach (don't get this the wrong way :) ). Keep it up.
 
Lot's of good things happening!

Just met with a friend to test a proof of concept with NinjaTrader. We were able to have strategies "talk" to each other while live trading which means the portfolio capabilities of FasterQuant can be implemented in NinjaTrader live trading. So the flow will go something like this:

1. FasterQuant identifies strategies and generates strategy code to plug into NinjaTrader.
2. Portfolio of strategies is forward tested via live trading in NinjaTrader and via forward testing in FasterQuant.
3. Testing results from NinjaTrader and FasterQuant should match up.

Looks like I also might be able to get some consulting jobs through AlgoTerminal from their customer base, which may or not involve FasterQuant. There are not many programmers out their that a very familiar with the platform.

And lastly, there is a potential opportunity brewing which would be a full time quant position that would allow me to continue trading my own account and continue with FasterQuant. Also, it is a remote position :) I will post more details if it materializes.
 
fan27 said:
Lot's of good things happening!
And lastly, there is a potential opportunity brewing which would be a full time quant position that would allow me to continue trading my own account and continue with FasterQuant. Also, it is a remote position :) I will post more details if it materializes.
More...
Like I told you months ago, you won't die poor ;)
 
You must log in or register to reply here.
Back
Top