Developing a Trading Framework from Scratch
- Thread starter fatrat
- Start date
+$54 this morning. I found a severe bug in my code and had to rely on manual calculations again. As a backup, I wrote a program for my TI-89 to forecast prices. Because things were in disarray this morning, I double checked the numbers against the 89, whose math I can trust. 
Whoever the guy was who said to manually trade in the Frostengine thread was 100% correct. There's all kinds of strange issues that pop up in the use of 3rd party software that one can't anticipate.
Total profit on my system over 5 days of execution +$265.
I started dabbling with C# and Windows Forms last night. Windows forms are beautiful. I'm definitely going to switch my reporting interface to C# for some of the modules.
Whoever the guy was who said to manually trade in the Frostengine thread was 100% correct. There's all kinds of strange issues that pop up in the use of 3rd party software that one can't anticipate.
Total profit on my system over 5 days of execution +$265.
I started dabbling with C# and Windows Forms last night. Windows forms are beautiful. I'm definitely going to switch my reporting interface to C# for some of the modules.
+$122 this morning on equities
+$47 on EUR/USD
Same general trading system, though I added a new variation on it for Forex.
I'm having to refactor my database. Anyone know out there a convenient place to get and parse end of day information? I need EOD prices for every single ticker and I want to have a job that stores these things into my database automatically, without me having to enter them into the database manually.
It would be better if I could specify the ECN/Exchange on which the prices came, but I'll settle for an accurate closing price.
+$47 on EUR/USD
Same general trading system, though I added a new variation on it for Forex.
I'm having to refactor my database. Anyone know out there a convenient place to get and parse end of day information? I need EOD prices for every single ticker and I want to have a job that stores these things into my database automatically, without me having to enter them into the database manually.
It would be better if I could specify the ECN/Exchange on which the prices came, but I'll settle for an accurate closing price.
I only have 3 months of data, but here's the equity curve on backtests of 1m intervals. Since the system looks at multiple stocks, it cycles between different products so the %'age of time in the market for any one product is not very long.
The system exercises no position sizing and just executes 100 share transactions. I was hoping the equity curve would smooth out with the liquidity-analysis code tacked onto the Nov and Dec data, but it seems to make little difference. I was hoping I could fine tune removal of liquidity with the Level-2 information, but it seems like that doesn't make as much difference on the equity curve as I would have hoped.
In other words, I really wasted a lot of time writing code that didn't need to be written.
Part of the problem may also be that I have a sort of hybrid-backtesting system that relies on external DLLs and also the garbage that is E-Signal's EFS. I also think the adaptive tick filters and filtering systems are wonky too. The sheer pain involved in getting to the state where I am now makes me think I won't have much of a FULLY automated trading system that looks like something I can be proud of for at least a good 6 months.
This is the system applied to NTAP (NASDAQ), is highly speculative, and seriously unwieldy. I don't know if I have enough capital to finance this strategy, and this equity curve is for the initial non-aggressive version of the strategy I want to roll out.
<img src="http://www.losingtrader.com/images/ntap_hf.jpg">
The system exercises no position sizing and just executes 100 share transactions. I was hoping the equity curve would smooth out with the liquidity-analysis code tacked onto the Nov and Dec data, but it seems to make little difference. I was hoping I could fine tune removal of liquidity with the Level-2 information, but it seems like that doesn't make as much difference on the equity curve as I would have hoped.
In other words, I really wasted a lot of time writing code that didn't need to be written.
Part of the problem may also be that I have a sort of hybrid-backtesting system that relies on external DLLs and also the garbage that is E-Signal's EFS. I also think the adaptive tick filters and filtering systems are wonky too. The sheer pain involved in getting to the state where I am now makes me think I won't have much of a FULLY automated trading system that looks like something I can be proud of for at least a good 6 months.
This is the system applied to NTAP (NASDAQ), is highly speculative, and seriously unwieldy. I don't know if I have enough capital to finance this strategy, and this equity curve is for the initial non-aggressive version of the strategy I want to roll out.
<img src="http://www.losingtrader.com/images/ntap_hf.jpg">
I am all kinds of retarded, so perhaps that basement-dwelling Walter Jennings character was right to question my mathematical ability.
I forgot to enable adaptive stops on the last backtest.
This is a backtest after I ran the discrete tick point data through a filter to remove high frequency noise and made calculations dynamically for the stop-loss parameter. The profit factor improved to 1.94. The system only works on Network Appliances (NASDAQ: NTAP).
<img src="http://www.losingtrader.com/ntap_hf2.jpg">
Much better. I feel less dumb now.
I forgot to enable adaptive stops on the last backtest.
This is a backtest after I ran the discrete tick point data through a filter to remove high frequency noise and made calculations dynamically for the stop-loss parameter. The profit factor improved to 1.94. The system only works on Network Appliances (NASDAQ: NTAP).
<img src="http://www.losingtrader.com/ntap_hf2.jpg">
Much better. I feel less dumb now.
+$45 this morning.
Note for future development:
There was an erroneous tick at the close for the sotck I was interested in. That tick caused a theoretical price discrepancy which caused the morning prices used for the trading system to screw up. So, I must add a tick cleanser so the numbers in the automated system realign with what my eye knows the numbers are supposed to be.
Note for future development:
There was an erroneous tick at the close for the sotck I was interested in. That tick caused a theoretical price discrepancy which caused the morning prices used for the trading system to screw up. So, I must add a tick cleanser so the numbers in the automated system realign with what my eye knows the numbers are supposed to be.
Long time no update. I started taking classes in preparation for graduate school, so my time got forked. I got a job to pay for graduate school and my job allows me the freedom to still be a trader; however, classes interrupt my trading day now. So it's even more important that I progress through the automation angle of this system. I intend to continue updates as well.
Just briefly checking in. I have not abandoned this system.
Just briefly checking in. I have not abandoned this system.